diff --git a/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json b/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..de9e98acac39a0935f7baee5baa9a6b98eb8cd8d
--- /dev/null
+++ b/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json
@@ -0,0 +1,200 @@
+{"filepath": "data/multi_event_test/syn_1.wav", "onoffCaption": "cat meowing at 0.5-2.0, 3.0-4.5 and whistling at 5.0-6.5 and explosion at 7.0-8.0, 8.5-9.5", "frequencyCaption": "cat meowing two times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_6.wav", "onoffCaption": "whistling at 2.0-6.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_8.wav", "onoffCaption": "cow mooing at 1.954-4.954, 6.219-9.219", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_11.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5 and dog barking at 5.0-7.0", "frequencyCaption": "burping belching two times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_16.wav", "onoffCaption": "duck quacking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_18.wav", "onoffCaption": "door knocking at 0.138-2.518, 3.708-6.088 and door slamming at 2.798-4.798", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_21.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_26.wav", "onoffCaption": "whistling at 0.2-4.2", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_28.wav", "onoffCaption": "cow mooing at 0.0-1.0 and spraying at 1.0-2.0", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_32.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_35.wav", "onoffCaption": "car horn honking at 0.5-2.5, 3.0-5.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_43.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5 and burping belching at 5.0-7.0, 7.5-9.5 and explosion at 4.8-7.8", "frequencyCaption": "dog barking two times and burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_44.wav", "onoffCaption": "sneeze at 0.5-1.5", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_50.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0 and sneeze at 6.0-7.0 and train horn at 8.0-10.0", "frequencyCaption": "car horn honking two times and sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_57.wav", "onoffCaption": "dog barking at 0.0-2.0, 3.0-5.0 and cow mooing at 6.0-9.0", "frequencyCaption": "dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_59.wav", "onoffCaption": "door slamming at 0.0-1.0 and explosion at 1.5-4.5, 5.0-8.0", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_60.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_67.wav", "onoffCaption": "whistling at 0.204-5.379", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_69.wav", "onoffCaption": "door knocking at 0-1, 2-3, 4-5", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_73.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3 and sneeze at 3-4, 4-5", "frequencyCaption": "door knocking three times and sneeze two times"}
+{"filepath": "data/multi_event_test/syn_74.wav", "onoffCaption": "spraying at 0.5-1.0, 1.5-2.0 and gunshot at 3.0-4.0, 5.0-6.0, 7.0-8.0", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_test/syn_82.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_91.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_96.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_98.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_101.wav", "onoffCaption": "dog barking at 0.464-2.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_106.wav", "onoffCaption": "burping belching at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_test/syn_108.wav", "onoffCaption": "sneeze at 0.5-1.5", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_112.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_115.wav", "onoffCaption": "duck quacking at 0.3-2.3 and tapping clicking clanking at 2.5-5.5, 6.0-9.0", "frequencyCaption": "duck quacking one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_122.wav", "onoffCaption": "door knocking at 0-1, 3-4, 6-7", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_125.wav", "onoffCaption": "cow mooing at 1.5-4.5, 5.5-8.5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_131.wav", "onoffCaption": "whistling at 0-1, 2-3 and cat meowing at 1-2", "frequencyCaption": "whistling two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_136.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.5-5.5 and whistling at 6.0-8.0, 8.5-9.5 and woman laughing at 2.0-4.0", "frequencyCaption": "sheep goat bleating two times and whistling two times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_138.wav", "onoffCaption": "gunshot at 0.0-1.0 and tapping clicking clanking at 1.5-5.0", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_140.wav", "onoffCaption": "door knocking at 0.00-2.00, 3.00-5.00, 6.00-8.00", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_147.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_149.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0 and spraying at 5.5-6.0, 7.0-7.5", "frequencyCaption": "car horn honking two times and spraying two times"}
+{"filepath": "data/multi_event_test/syn_153.wav", "onoffCaption": "cat meowing at 0-1.0", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_154.wav", "onoffCaption": "cat meowing at 0.5-1.5 and door knocking at 2-3.5", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_163.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_164.wav", "onoffCaption": "whistling at 0.204-5.379, 7.724-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_test/syn_170.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_177.wav", "onoffCaption": "thump thud at 0-1 and cow mooing at 1-2", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_179.wav", "onoffCaption": "cow mooing at 1.954-4.602, 5.719-8.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_181.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_186.wav", "onoffCaption": "gunshot at 0.0-1.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_188.wav", "onoffCaption": "gunshot at 0-1, 1-2 and duck quacking at 2-3", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_192.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_195.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_3.wav", "onoffCaption": "tapping clicking clanking at 0.5-3.0, 4.0-7.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_4.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_13.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_14.wav", "onoffCaption": "sneeze at 0.38-1.38", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_23.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.5-3.5", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_24.wav", "onoffCaption": "woman laughing at 2.782-5.368", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_30.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.871-7.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_37.wav", "onoffCaption": "thump thud at 0.0-1.5, 5.0-6.5 and door knocking at 1.5-3.5, 6.5-8.5 and burping belching at 3.5-4.5", "frequencyCaption": "thump thud two times and door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_39.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_41.wav", "onoffCaption": "thump thud at 0.0-1.0", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_test/syn_48.wav", "onoffCaption": "cat meowing at 0-1, 1-2, 2-3", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_52.wav", "onoffCaption": "gunshot at 0.0-1.0 and duck quacking at 1.5-2.5 and tapping clicking clanking at 3.0-4.0", "frequencyCaption": "gunshot one times and duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_55.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_62.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_65.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_71.wav", "onoffCaption": "door slamming at 0.0-1.0, 2.0-3.0 and whistling at 4.0-8.0", "frequencyCaption": "door slamming two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_76.wav", "onoffCaption": "dog barking at 0.464-2.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_78.wav", "onoffCaption": "explosion at 0.0-2.0, 2.5-4.5 and duck quacking at 5.0-7.0, 7.5-9.5", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_80.wav", "onoffCaption": "door slamming at 0.0-1.0 and sheep goat bleating at 2.0-4.0", "frequencyCaption": "door slamming one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_85.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_87.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_89.wav", "onoffCaption": "car horn honking at 0.0-2.0 and cat meowing at 2.5-4.0", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_93.wav", "onoffCaption": "dog barking at 0-2, 2-4", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_94.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0 and thump thud at 1.0-3.0, 4.0-6.0 and sheep goat bleating at 2.0-4.0, 7.0-9.0", "frequencyCaption": "gunshot three times and thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_103.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_104.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_110.wav", "onoffCaption": "train horn at 0-1 and duck quacking at 1-2 and cow mooing at 2-3", "frequencyCaption": "train horn one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_117.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.5-6.5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_119.wav", "onoffCaption": "train horn at 0.0-2.0 and door knocking at 2.5-4.5, 5.0-7.0", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_120.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.871-5.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_127.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_129.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_133.wav", "onoffCaption": "duck quacking at 2.203-4.203, 5.361-7.361", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_134.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_142.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.0-3.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_145.wav", "onoffCaption": "door knocking at 0.002-2.092, 2.842-5.601 and whistling at 1.9-10.0", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_151.wav", "onoffCaption": "dog barking at 0.121-2.121, 3.824-5.824, 7.767-9.767", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_156.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_158.wav", "onoffCaption": "tapping clicking clanking at 1.5-4.5, 5.5-8.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_161.wav", "onoffCaption": "spraying at 0-1, 2-3", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_166.wav", "onoffCaption": "woman laughing at 1.672-3.955", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_168.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_172.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_175.wav", "onoffCaption": "cow mooing at 0-3 and spraying at 3-6", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_183.wav", "onoffCaption": "explosion at 0.0-2.0, 2.1-4.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_184.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_190.wav", "onoffCaption": "whistling at 0.0-1.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_197.wav", "onoffCaption": "tapping clicking clanking at 0.032-2.032, 2.532-4.532", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_199.wav", "onoffCaption": "duck quacking at 0.0-2.0 and cat meowing at 2.5-4.5", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_200.wav", "onoffCaption": "explosion at 1.0-3.0, 4.0-6.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_2.wav", "onoffCaption": "door knocking at 0.0-1.0", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_5.wav", "onoffCaption": "burping belching at 0.359-2.774", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_test/syn_12.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0 and sneeze at 2.5-3.5", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_test/syn_15.wav", "onoffCaption": "tapping clicking clanking at 2.992-6.432", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_22.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_25.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.391-7.391", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_31.wav", "onoffCaption": "woman laughing at 0-1, 2-3", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_36.wav", "onoffCaption": "door slamming at 0.5-1.5, 2.0-3.0, 3.5-4.5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_38.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_40.wav", "onoffCaption": "door knocking at 0.138-2.518, 3.708-6.088 and cow mooing at 6.91-9.447", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_46.wav", "onoffCaption": "door slamming at 1.145-2.085, 3.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_47.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0 and cow mooing at 6.0-8.0, 8.5-10.0", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_49.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0 and tapping clicking clanking at 0.0-4.0, 5.5-9.5", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_51.wav", "onoffCaption": "train horn at 0.873-4.633, 5.147-8.907", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_53.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_54.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_63.wav", "onoffCaption": "train horn at 0-1 and cat meowing at 2-3 and dog barking at 4-5", "frequencyCaption": "train horn one times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_70.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_77.wav", "onoffCaption": "cow mooing at 0.0-3.0", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_79.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_81.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_86.wav", "onoffCaption": "whistling at 0-1 and woman laughing at 1-3, 3-5", "frequencyCaption": "whistling one times and woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_88.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.0-6.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_92.wav", "onoffCaption": "door slamming at 0.0-1.0, 2.0-3.0, 4.0-5.0 and tapping clicking clanking at 6.0-7.0", "frequencyCaption": "door slamming three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_95.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_102.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_105.wav", "onoffCaption": "train horn at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_test/syn_111.wav", "onoffCaption": "whistling at 0.204-2.79, 4.0-6.586 and door slamming at 7.0-8.0", "frequencyCaption": "whistling two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_116.wav", "onoffCaption": "burping belching at 1.0-3.0, 4.0-6.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_118.wav", "onoffCaption": "sneeze at 0.0-1.0", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_121.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.65-2.65, 3.65-5.65", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_126.wav", "onoffCaption": "sneeze at 0.373-2.332 and car horn honking at 1.03-5.542, 6.081-10.0", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_128.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0 and door knocking at 3.5-5.5, 6.0-8.0", "frequencyCaption": "sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_132.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0 and spraying at 2.5-3.0, 4.0-4.5, 5.5-6.0 and duck quacking at 6.5-7.5, 8.0-9.0, 9.5-10.0", "frequencyCaption": "sheep goat bleating one times and spraying three times and duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_135.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_139.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_143.wav", "onoffCaption": "spraying at 0.0-1.0 and explosion at 1.5-4.5", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_144.wav", "onoffCaption": "duck quacking at 0-1, 2-3, 4-5", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_150.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_157.wav", "onoffCaption": "train horn at 0-3.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_159.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_160.wav", "onoffCaption": "spraying at 0.0-1.0 and whistling at 1.0-3.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_167.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5 and gunshot at 5.0-7.0", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_169.wav", "onoffCaption": "sneeze at 0.373-2.332, 3.255-5.716", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_173.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_174.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_176.wav", "onoffCaption": "woman laughing at 1.625-3.98, 4.735-6.981", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_182.wav", "onoffCaption": "cow mooing at 0.0-3.0 and gunshot at 4.0-5.0", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_185.wav", "onoffCaption": "spraying at 0.22-0.847 and door knocking at 2.797-5.334", "frequencyCaption": "spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_189.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_191.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_193.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_196.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_test/syn_198.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_7.wav", "onoffCaption": "spraying at 0.0-1.0 and burping belching at 1.5-2.5", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_9.wav", "onoffCaption": "cow mooing at 0.0-3.0", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_10.wav", "onoffCaption": "door knocking at 2-4, 5-7", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_17.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_19.wav", "onoffCaption": "gunshot at 0.0-1.0 and spraying at 1.5-2.5", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_27.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_33.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5 and car horn honking at 5.0-7.0", "frequencyCaption": "dog barking two times and car horn honking one times"}
+{"filepath": "data/multi_event_test/syn_34.wav", "onoffCaption": "sheep goat bleating at 1.575-3.575", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_42.wav", "onoffCaption": "tapping clicking clanking at 0.0-2.0, 2.5-4.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_45.wav", "onoffCaption": "cat meowing at 0.5-1.5 and train horn at 2.0-6.0", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_56.wav", "onoffCaption": "tapping clicking clanking at 0.961-4.401, 6.37-9.81", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_58.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_test/syn_61.wav", "onoffCaption": "explosion at 0.5-3.5, 4.0-7.0 and train horn at 7.5-10.0 and woman laughing at 1.0-4.0", "frequencyCaption": "explosion two times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_66.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_68.wav", "onoffCaption": "car horn honking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_72.wav", "onoffCaption": "spraying at 0.0-0.6, 1.0-1.6 and thump thud at 2.0-3.6 and dog barking at 4.0-6.0", "frequencyCaption": "spraying two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_75.wav", "onoffCaption": "explosion at 0.5-2.5, 2.501-4.501", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_83.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_84.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.642-7.642", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_90.wav", "onoffCaption": "gunshot at 0.2-1.2", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_97.wav", "onoffCaption": "cat meowing at 0.5-1.5", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_99.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.0-4.0", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_100.wav", "onoffCaption": "cat meowing at 0.0-2.0 and sheep goat bleating at 3.0-5.0, 6.0-8.0, 9.0-10.0", "frequencyCaption": "cat meowing one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_107.wav", "onoffCaption": "spraying at 0.0-1.5, 2.0-3.5 and dog barking at 4.0-6.0, 7.0-9.0 and tapping clicking clanking at 1.6-3.1, 3.6-5.1", "frequencyCaption": "spraying two times and dog barking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_109.wav", "onoffCaption": "cow mooing at 0.0-3.0 and gunshot at 3.5-4.5", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_113.wav", "onoffCaption": "whistling at 0.742-5.917 and tapping clicking clanking at 2.992-6.432", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_114.wav", "onoffCaption": "car horn honking at 0-2 and door knocking at 2-4", "frequencyCaption": "car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_124.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_130.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_137.wav", "onoffCaption": "door knocking at 0-1 and cow mooing at 2-3 and gunshot at 4-5", "frequencyCaption": "door knocking one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_141.wav", "onoffCaption": "sneeze at 0.33-1.403, 2.759-3.832", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_146.wav", "onoffCaption": "sneeze at 0.0-1.0, 2.0-3.0 and cat meowing at 4.0-5.0", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_148.wav", "onoffCaption": "duck quacking at 0-1, 2-3 and cow mooing at 4-5", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_152.wav", "onoffCaption": "tapping clicking clanking at 0.0-1.0, 1.5-2.5 and train horn at 3.0-7.0", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_155.wav", "onoffCaption": "tapping clicking clanking at 0.0-1.0 and gunshot at 2.0-3.0 and cat meowing at 4.0-5.0", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_162.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_165.wav", "onoffCaption": "thump thud at 0.0-1.5, 2.0-3.5 and whistling at 4.0-7.0", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_171.wav", "onoffCaption": "spraying at 0.0-0.5, 1.5-2.0, 3.0-3.5 and thump thud at 4.0-5.0 and sheep goat bleating at 5.5-6.5, 7.0-8.0", "frequencyCaption": "spraying three times and thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_178.wav", "onoffCaption": "door slamming at 0.355-2.581 and woman laughing at 0.964-3.319", "frequencyCaption": "door slamming one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_180.wav", "onoffCaption": "spraying at 0.0-1.0 and cow mooing at 2.0-5.0", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_187.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_194.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
diff --git a/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json b/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..9310b20b6657974a4e9a39f0ddf9c0ab7c252c6f
--- /dev/null
+++ b/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json
@@ -0,0 +1,400 @@
+{"filepath": "data/single_event_multi_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 1.674-5.019", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_6.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_8.wav", "onoffCaption": "door slamming at 0-1", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_11.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_16.wav", "onoffCaption": "thump thud at 0-1", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_18.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_21.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_26.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_28.wav", "onoffCaption": "sneeze at 0-1, 2-3", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_32.wav", "onoffCaption": "cow mooing at 0-3.309", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_35.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_43.wav", "onoffCaption": "thump thud at 0-1, 2-3", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_44.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_50.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_57.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_59.wav", "onoffCaption": "woman laughing at 2.0-4.5, 5.0-7.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_60.wav", "onoffCaption": "cat meowing at 1-2, 3-4", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_69.wav", "onoffCaption": "burping belching at 2.0-3.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_73.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_74.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_82.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-8.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_91.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_96.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_98.wav", "onoffCaption": "woman laughing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_101.wav", "onoffCaption": "burping belching at 0.871-1.871, 2.871-3.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_106.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_108.wav", "onoffCaption": "cat meowing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_112.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_115.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.0-4.0, 5.0-8.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_125.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_131.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_136.wav", "onoffCaption": "tapping clicking clanking at 1-3, 6-8", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_138.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_140.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_147.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.5-5.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_149.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_153.wav", "onoffCaption": "cow mooing at 0-1, 2-3", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_154.wav", "onoffCaption": "train horn at 0-1, 2-3", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_163.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_164.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_170.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_177.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_179.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_181.wav", "onoffCaption": "door knocking at 0-1.0", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_186.wav", "onoffCaption": "sneeze at 0.3-1.3, 2.3-3.3", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_188.wav", "onoffCaption": "explosion at 0-2", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_192.wav", "onoffCaption": "cat meowing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_195.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_4.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.032-1.032, 2.032-3.032, 4.032-5.032", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_14.wav", "onoffCaption": "tapping clicking clanking at 0-1, 2-3", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_23.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_24.wav", "onoffCaption": "thump thud at 1-2, 3-4", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_30.wav", "onoffCaption": "explosion at 0-1, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_37.wav", "onoffCaption": "tapping clicking clanking at 0.5-2.5, 3-5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_39.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_41.wav", "onoffCaption": "car horn honking at 1-2, 3-4", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_48.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_55.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_62.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_65.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_71.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_76.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_78.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_80.wav", "onoffCaption": "car horn honking at 0-1", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_85.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_87.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_89.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_93.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_94.wav", "onoffCaption": "burping belching at 0.871-2.871, 5.218-7.218", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_103.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_104.wav", "onoffCaption": "thump thud at 0-2, 3-5", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_110.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_117.wav", "onoffCaption": "tapping clicking clanking at 0-2", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_119.wav", "onoffCaption": "duck quacking at 0.235-2.235, 3.085-5.085", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_120.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_127.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_129.wav", "onoffCaption": "burping belching at 0.5-1.5, 2.5-3.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_133.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_134.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_142.wav", "onoffCaption": "cow mooing at 1-2, 3-4", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_145.wav", "onoffCaption": "cow mooing at 0.0-2.0, 3.0-5.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_151.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_156.wav", "onoffCaption": "thump thud at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_158.wav", "onoffCaption": "burping belching at 1.5-3.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_161.wav", "onoffCaption": "car horn honking at 1.5-3.5, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_166.wav", "onoffCaption": "burping belching at 0-1, 1-2, 2-3", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_168.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_172.wav", "onoffCaption": "woman laughing at 0.0-2.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_175.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_183.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_184.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_190.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_197.wav", "onoffCaption": "car horn honking at 1.817-4.404, 5.85-8.437", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_199.wav", "onoffCaption": "car horn honking at 0.664-3.129", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_200.wav", "onoffCaption": "train horn at 0-2", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 0.5-2.5", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_12.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_15.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_22.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_25.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_31.wav", "onoffCaption": "sneeze at 0.0-1.0", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_36.wav", "onoffCaption": "sheep goat bleating at 0-1, 1-2", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_38.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_40.wav", "onoffCaption": "woman laughing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_46.wav", "onoffCaption": "tapping clicking clanking at 1.5-5.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_47.wav", "onoffCaption": "cow mooing at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_49.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_51.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_53.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_54.wav", "onoffCaption": "cow mooing at 1.954-6.383, 7.52-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_63.wav", "onoffCaption": "explosion at 0.0-3.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_64.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_70.wav", "onoffCaption": "dog barking at 1-2", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_77.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_79.wav", "onoffCaption": "train horn at 0-2.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_81.wav", "onoffCaption": "tapping clicking clanking at 0-1", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_86.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_88.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_92.wav", "onoffCaption": "door slamming at 0-1, 2-4", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_95.wav", "onoffCaption": "woman laughing at 1.5-3.5, 4.0-6.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_102.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_111.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_116.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_118.wav", "onoffCaption": "sheep goat bleating at 0.5-1.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_121.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_126.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_128.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_132.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_135.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_139.wav", "onoffCaption": "spraying at 0-1, 1-2", "frequencyCaption": "spraying two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_143.wav", "onoffCaption": "door knocking at 0.645-2.772, 3.875-6.782, 7.405-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_144.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_150.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_157.wav", "onoffCaption": "explosion at 0.5-1.5, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_159.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.5-3.5", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_160.wav", "onoffCaption": "woman laughing at 0.0-2.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_167.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_169.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_173.wav", "onoffCaption": "explosion at 0-3", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_174.wav", "onoffCaption": "duck quacking at 0.2-2.2, 3.2-5.2", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_176.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_182.wav", "onoffCaption": "car horn honking at 0.653-3.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_185.wav", "onoffCaption": "dog barking at 0-2, 3-5", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_189.wav", "onoffCaption": "burping belching at 0-1, 2-3", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_191.wav", "onoffCaption": "tapping clicking clanking at 0.0-4.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_193.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_196.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_198.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_7.wav", "onoffCaption": "door slamming at 0-2, 2-4", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_9.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.0-3.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_10.wav", "onoffCaption": "door slamming at 0.0-1.0", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_17.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_19.wav", "onoffCaption": "thump thud at 1.9-4.5, 5.5-8.1", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_20.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_27.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_29.wav", "onoffCaption": "woman laughing at 0-1", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_33.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_34.wav", "onoffCaption": "dog barking at 0-1, 1.5-2.5, 3-4", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_42.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_45.wav", "onoffCaption": "woman laughing at 0-1", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_56.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_58.wav", "onoffCaption": "spraying at 0.5-1.5", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_61.wav", "onoffCaption": "sheep goat bleating at 0.8-2.8, 3.8-5.8", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_66.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_68.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_72.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_75.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_83.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_84.wav", "onoffCaption": "burping belching at 0-3", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_90.wav", "onoffCaption": "whistling at 1-2", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_97.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_99.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_100.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_107.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_113.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_114.wav", "onoffCaption": "explosion at 0.0-3.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_124.wav", "onoffCaption": "woman laughing at 0-1, 1-2", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_130.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_137.wav", "onoffCaption": "train horn at 0-1, 2-3, 4-5", "frequencyCaption": "train horn three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 0.5-3.5", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_146.wav", "onoffCaption": "sneeze at 0.8-1.8", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_148.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_152.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_155.wav", "onoffCaption": "spraying at 0.033-1.519", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_162.wav", "onoffCaption": "explosion at 0-1, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_165.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_171.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_178.wav", "onoffCaption": "tapping clicking clanking at 1-3, 4-6", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_187.wav", "onoffCaption": "explosion at 1.5-4.5", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_194.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_11.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_16.wav", "onoffCaption": "burping belching at 0.5-3.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_18.wav", "onoffCaption": "burping belching at 1.0-2.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_21.wav", "onoffCaption": "burping belching at 0.5-1.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_26.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_28.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_32.wav", "onoffCaption": "gunshot at 0-1, 2-3, 4-5", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_35.wav", "onoffCaption": "woman laughing at 0-2, 3-5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_43.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_44.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_50.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_57.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_59.wav", "onoffCaption": "gunshot at 0-1, 2-3, 4-5", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_60.wav", "onoffCaption": "cow mooing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_69.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.1-4.1", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_73.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_74.wav", "onoffCaption": "cow mooing at 0-2, 3-5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_82.wav", "onoffCaption": "woman laughing at 2.0-6.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_85.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_91.wav", "onoffCaption": "tapping clicking clanking at 1.0-4.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_96.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_98.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_101.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_106.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_108.wav", "onoffCaption": "gunshot at 0.2-2.2, 3.2-5.2", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_112.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.871-5.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_115.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.0-5.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_125.wav", "onoffCaption": "explosion at 1.0-3.0, 3.5-5.5", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_131.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_136.wav", "onoffCaption": "car horn honking at 0-1", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_138.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_140.wav", "onoffCaption": "train horn at 0-1, 1-2", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_147.wav", "onoffCaption": "explosion at 0-1, 2-3, 4-5", "frequencyCaption": "explosion three times"}
+{"filepath": "data/single_event_single_identity_test/syn_149.wav", "onoffCaption": "spraying at 0.1-1.1, 1.2-2.2, 3.3-4.3", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_153.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_154.wav", "onoffCaption": "explosion at 0.0-1.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_163.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_164.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_170.wav", "onoffCaption": "burping belching at 0.5-2.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_175.wav", "onoffCaption": "explosion at 2.941-5.813", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_177.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_179.wav", "onoffCaption": "explosion at 0.0-4.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_181.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_186.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 2.5-4.5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_188.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_190.wav", "onoffCaption": "woman laughing at 0-1, 2-3", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_192.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_195.wav", "onoffCaption": "cow mooing at 1.5-3.5, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_14.wav", "onoffCaption": "woman laughing at 0-2", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_23.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_24.wav", "onoffCaption": "dog barking at 0.311-2.711", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_30.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_37.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_39.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_41.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 2.75-4.75", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_46.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_48.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_55.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_62.wav", "onoffCaption": "burping belching at 1.5-3.5, 4.0-6.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_65.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.0-6.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_71.wav", "onoffCaption": "tapping clicking clanking at 0.5-3.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_76.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_78.wav", "onoffCaption": "train horn at 0-1, 2-3", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_80.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_87.wav", "onoffCaption": "car horn honking at 1.817-4.12, 6.106-8.453", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_89.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_93.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_94.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_103.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_104.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_110.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_117.wav", "onoffCaption": "cat meowing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_119.wav", "onoffCaption": "car horn honking at 0.0-2.0", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_120.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_127.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3-5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_129.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_133.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_134.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_142.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_145.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_151.wav", "onoffCaption": "burping belching at 1-3", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_156.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_158.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_161.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_166.wav", "onoffCaption": "tapping clicking clanking at 0.032-3.472, 4.758-7.489", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_168.wav", "onoffCaption": "explosion at 2.941-5.813", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_172.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_183.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_184.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_197.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_199.wav", "onoffCaption": "dog barking at 0-2, 2-4", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_200.wav", "onoffCaption": "thump thud at 2.224-5.891, 7.389-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_12.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_15.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_22.wav", "onoffCaption": "whistling at 2.603-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_25.wav", "onoffCaption": "explosion at 0.0-2.0, 2.5-4.5", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_31.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_36.wav", "onoffCaption": "dog barking at 0.5-1.5", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_38.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_40.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_single_identity_test/syn_47.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_49.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_51.wav", "onoffCaption": "cat meowing at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_53.wav", "onoffCaption": "cat meowing at 0-2", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_54.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_63.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_70.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_77.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_79.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_81.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_86.wav", "onoffCaption": "door knocking at 1-2", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_88.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_92.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_95.wav", "onoffCaption": "thump thud at 0.0-1.0", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_102.wav", "onoffCaption": "thump thud at 0-1, 2-3", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_111.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_116.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_118.wav", "onoffCaption": "cat meowing at 0-3", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_121.wav", "onoffCaption": "door knocking at 1.155-5.305", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_126.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_128.wav", "onoffCaption": "tapping clicking clanking at 0-1", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_132.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_135.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_139.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_143.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_144.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_150.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_157.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_159.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_160.wav", "onoffCaption": "tapping clicking clanking at 1.5-3.5, 5-7", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_167.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_169.wav", "onoffCaption": "train horn at 0-3.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_173.wav", "onoffCaption": "thump thud at 0-1", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_174.wav", "onoffCaption": "cat meowing at 0-1.2, 2-3.2, 4-5.2", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_182.wav", "onoffCaption": "door knocking at 1-3, 4-6", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_185.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_189.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_191.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_196.wav", "onoffCaption": "door knocking at 0-1, 2-3, 4-5", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_198.wav", "onoffCaption": "explosion at 0.0-1.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_10.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_17.wav", "onoffCaption": "burping belching at 0.871-5.871, 7.218-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_19.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_27.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 1-3, 4-6", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_33.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_34.wav", "onoffCaption": "burping belching at 0-1, 2-3", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_42.wav", "onoffCaption": "dog barking at 2.579-4.579", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_45.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_56.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_58.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_61.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_66.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_68.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_72.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_75.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_83.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_84.wav", "onoffCaption": "woman laughing at 0-2, 2-4", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_90.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_97.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_99.wav", "onoffCaption": "door knocking at 1-3, 4-6", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_100.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.891-5.891", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_107.wav", "onoffCaption": "woman laughing at 0-2, 5-7", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_113.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_114.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_123.wav", "onoffCaption": "woman laughing at 2.777-6.165", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_124.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_130.wav", "onoffCaption": "duck quacking at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_137.wav", "onoffCaption": "dog barking at 0.5-2.5, 3.0-5.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 2.782-5.368", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_146.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_148.wav", "onoffCaption": "thump thud at 2-3", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_152.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_155.wav", "onoffCaption": "woman laughing at 0.5-2.5, 3.0-5.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_162.wav", "onoffCaption": "door knocking at 0-1, 1-2", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_165.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_171.wav", "onoffCaption": "woman laughing at 2.672-5.672", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_176.wav", "onoffCaption": "burping belching at 0.5-3.5, 4.5-7.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_178.wav", "onoffCaption": "sheep goat bleating at 1-3, 4-7", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_187.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_193.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_194.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 0.5-1.5, 2.5-3.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_single_identity_test/syn_4.wav", "onoffCaption": "car horn honking at 0.664-3.129, 4.357-7.014", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.0-2.0", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_6.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_7.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_8.wav", "onoffCaption": "burping belching at 2.861-8.462", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_9.wav", "onoffCaption": "burping belching at 0.569-4.438", "frequencyCaption": "burping belching one times"}
diff --git a/data/meta_data/test-onoff-control_multi-event.json b/data/meta_data/test-onoff-control_multi-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..7c57e688ed2805ae7ac07be7b61dc10e52d3b82e
--- /dev/null
+++ b/data/meta_data/test-onoff-control_multi-event.json
@@ -0,0 +1,200 @@
+{"filepath": "data/multi_event_test/syn_1.wav", "onoffCaption": "cat meowing at 0.393-1.783, 3.975-5.365 and whistling at 0.861-5.455 and explosion at 2.089-4.841, 5.738-8.538", "frequencyCaption": "cat meowing two times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_6.wav", "onoffCaption": "whistling at 2.093-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_8.wav", "onoffCaption": "cow mooing at 1.177-3.977, 5.15-7.774", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_11.wav", "onoffCaption": "burping belching at 1.039-3.198, 4.085-6.244 and dog barking at 3.119-5.119", "frequencyCaption": "burping belching two times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_16.wav", "onoffCaption": "duck quacking at 0.799-2.799, 3.634-5.634, 6.976-8.976", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_18.wav", "onoffCaption": "door knocking at 1.225-3.352, 5.173-7.3 and door slamming at 5.439-7.678", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_21.wav", "onoffCaption": "dog barking at 2.947-4.947, 6.186-8.186", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_26.wav", "onoffCaption": "whistling at 2.848-7.442", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_28.wav", "onoffCaption": "cow mooing at 2.639-5.263 and spraying at 8.565-9.667", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_32.wav", "onoffCaption": "duck quacking at 0.039-2.039, 3.171-5.171, 5.938-7.938", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_35.wav", "onoffCaption": "car horn honking at 2.31-5.271, 5.91-8.871", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_43.wav", "onoffCaption": "dog barking at 2.157-4.157, 5.953-7.953 and burping belching at 2.431-5.388, 6.452-8.611 and explosion at 4.8-7.552", "frequencyCaption": "dog barking two times and burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_44.wav", "onoffCaption": "sneeze at 2.638-6.791", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_50.wav", "onoffCaption": "car horn honking at 0.874-3.835, 4.429-7.39 and sneeze at 1.814-5.167 and train horn at 2.818-7.898", "frequencyCaption": "car horn honking two times and sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_57.wav", "onoffCaption": "dog barking at 3.007-5.007, 6.103-8.103 and cow mooing at 3.017-5.641", "frequencyCaption": "dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_59.wav", "onoffCaption": "door slamming at 0.035-2.274 and explosion at 3.857-6.609, 7.377-10.0", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_60.wav", "onoffCaption": "train horn at 0.062-3.062", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_67.wav", "onoffCaption": "whistling at 1.616-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_69.wav", "onoffCaption": "door knocking at 0.237-2.801, 4.117-6.681, 7.378-9.942", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_73.wav", "onoffCaption": "door knocking at 0.045-2.172, 2.718-5.282, 6.027-8.591 and sneeze at 2.92-6.273, 6.847-9.032", "frequencyCaption": "door knocking three times and sneeze two times"}
+{"filepath": "data/multi_event_test/syn_74.wav", "onoffCaption": "spraying at 0.38-1.176, 3.06-3.856 and gunshot at 1.729-3.729, 4.367-6.367, 7.031-9.031", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_test/syn_82.wav", "onoffCaption": "dog barking at 0.497-2.497, 4.187-6.187", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_91.wav", "onoffCaption": "gunshot at 0.501-2.501, 3.148-5.148", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_96.wav", "onoffCaption": "door slamming at 0.154-2.393, 3.23-4.641, 5.232-7.471", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_98.wav", "onoffCaption": "thump thud at 1.835-4.135, 6.505-9.18", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_101.wav", "onoffCaption": "dog barking at 0.681-2.681", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_106.wav", "onoffCaption": "burping belching at 0.093-3.05, 3.962-6.121, 7.309-9.468", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_test/syn_108.wav", "onoffCaption": "sneeze at 3.287-7.44", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_112.wav", "onoffCaption": "woman laughing at 1.823-4.587, 6.243-9.007", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_115.wav", "onoffCaption": "duck quacking at 0.044-1.862 and tapping clicking clanking at 0.436-3.876, 5.547-7.6", "frequencyCaption": "duck quacking one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_122.wav", "onoffCaption": "door knocking at 1.266-3.83, 4.854-7.418, 7.929-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_125.wav", "onoffCaption": "cow mooing at 2.954-5.754, 6.384-9.008", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_131.wav", "onoffCaption": "whistling at 0.666-5.26, 5.984-8.335 and cat meowing at 0.904-2.294", "frequencyCaption": "whistling two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_136.wav", "onoffCaption": "sheep goat bleating at 0.226-2.226, 3.707-5.707 and whistling at 1.058-5.652, 6.943-10.0 and woman laughing at 2.749-7.207", "frequencyCaption": "sheep goat bleating two times and whistling two times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_138.wav", "onoffCaption": "gunshot at 0.785-2.785 and tapping clicking clanking at 5.685-9.125", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_140.wav", "onoffCaption": "door knocking at 0.341-2.468, 3.382-5.946, 7.206-9.77", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_147.wav", "onoffCaption": "door slamming at 0.305-1.716, 2.95-4.361, 5.691-7.102", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_149.wav", "onoffCaption": "car horn honking at 0.666-3.35, 5.748-8.432 and spraying at 7.494-8.29, 8.904-9.7", "frequencyCaption": "car horn honking two times and spraying two times"}
+{"filepath": "data/multi_event_test/syn_153.wav", "onoffCaption": "cat meowing at 3.029-4.355", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_154.wav", "onoffCaption": "cat meowing at 2.044-3.37 and door knocking at 2.866-5.43", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_163.wav", "onoffCaption": "sheep goat bleating at 0.139-2.139, 3.188-5.188, 6.077-8.077", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_164.wav", "onoffCaption": "whistling at 0.15-4.744, 6.868-8.971", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_test/syn_170.wav", "onoffCaption": "dog barking at 0.286-2.286, 3.801-5.801", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_177.wav", "onoffCaption": "thump thud at 0.593-2.893 and cow mooing at 4.617-7.241", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_179.wav", "onoffCaption": "cow mooing at 2.754-5.378, 6.145-8.769", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_181.wav", "onoffCaption": "cow mooing at 3.381-6.181, 7.936-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_186.wav", "onoffCaption": "gunshot at 0.131-2.131", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_188.wav", "onoffCaption": "gunshot at 0.785-2.785, 3.847-5.847 and duck quacking at 2.99-4.99", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_192.wav", "onoffCaption": "spraying at 1.763-2.865, 5.335-6.437", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_195.wav", "onoffCaption": "thump thud at 2.422-5.097, 5.945-8.245", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_3.wav", "onoffCaption": "tapping clicking clanking at 2.711-6.151, 7.783-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_4.wav", "onoffCaption": "door slamming at 3.076-4.487, 6.877-8.288", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_13.wav", "onoffCaption": "duck quacking at 0.012-2.012, 3.202-5.202, 7.582-9.582", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_14.wav", "onoffCaption": "sneeze at 1.853-6.006", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_23.wav", "onoffCaption": "sneeze at 0.109-4.262, 6.151-8.608", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_24.wav", "onoffCaption": "woman laughing at 3.051-7.509", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_30.wav", "onoffCaption": "burping belching at 3.234-6.191, 7.597-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_37.wav", "onoffCaption": "thump thud at 1.883-4.558, 6.153-8.453 and door knocking at 2.227-4.791, 5.771-8.335 and burping belching at 6.746-8.905", "frequencyCaption": "thump thud two times and door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_39.wav", "onoffCaption": "train horn at 2.197-5.197, 5.755-8.755", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_41.wav", "onoffCaption": "thump thud at 1.465-3.765", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_test/syn_48.wav", "onoffCaption": "cat meowing at 0.07-1.396, 3.738-5.064, 6.912-8.238", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_52.wav", "onoffCaption": "gunshot at 0.761-2.761 and duck quacking at 0.994-2.994 and tapping clicking clanking at 5.144-8.584", "frequencyCaption": "gunshot one times and duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_55.wav", "onoffCaption": "sneeze at 2.529-6.682, 7.206-9.677", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_62.wav", "onoffCaption": "woman laughing at 0.152-2.916, 5.112-7.934", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_65.wav", "onoffCaption": "gunshot at 3.755-5.755, 6.54-8.54", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_71.wav", "onoffCaption": "door slamming at 0.023-2.262, 4.712-6.123 and whistling at 1.979-6.573", "frequencyCaption": "door slamming two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_76.wav", "onoffCaption": "dog barking at 0.741-2.741", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_78.wav", "onoffCaption": "explosion at 0.11-2.862, 4.292-7.044 and duck quacking at 2.338-4.156, 5.898-7.716", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_80.wav", "onoffCaption": "door slamming at 0.695-2.106 and sheep goat bleating at 0.985-2.985", "frequencyCaption": "door slamming one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_85.wav", "onoffCaption": "door knocking at 4.074-6.201", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_87.wav", "onoffCaption": "explosion at 0.371-3.123, 5.335-8.087", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_89.wav", "onoffCaption": "car horn honking at 2.099-5.06 and cat meowing at 5.989-7.315", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_93.wav", "onoffCaption": "dog barking at 0.988-2.988, 5.289-7.289", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_94.wav", "onoffCaption": "gunshot at 1.463-3.463, 4.41-6.41, 7.226-9.226 and thump thud at 1.729-4.404, 6.318-8.993 and sheep goat bleating at 1.895-3.895, 5.909-7.909", "frequencyCaption": "gunshot three times and thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_103.wav", "onoffCaption": "whistling at 2.759-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_104.wav", "onoffCaption": "duck quacking at 4.149-5.967", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_110.wav", "onoffCaption": "train horn at 0.111-5.191 and duck quacking at 0.894-2.894 and cow mooing at 5.062-7.862", "frequencyCaption": "train horn one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_117.wav", "onoffCaption": "sheep goat bleating at 3.487-5.487, 7.705-9.705", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_119.wav", "onoffCaption": "train horn at 2.056-5.056 and door knocking at 2.912-5.039, 5.997-8.124", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_120.wav", "onoffCaption": "burping belching at 2.114-5.071, 5.723-8.68", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_127.wav", "onoffCaption": "whistling at 1.653-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_129.wav", "onoffCaption": "door knocking at 0.592-2.719, 3.326-5.453, 6.255-8.382", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_133.wav", "onoffCaption": "duck quacking at 1.444-3.262, 4.595-6.413", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_134.wav", "onoffCaption": "car horn honking at 0.439-3.123, 5.193-7.877", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_142.wav", "onoffCaption": "sneeze at 0.338-4.491, 5.776-7.91", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_145.wav", "onoffCaption": "door knocking at 0.308-2.872, 4.395-6.959 and whistling at 0.583-9.383", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_151.wav", "onoffCaption": "dog barking at 0.368-2.368, 3.112-5.112, 5.983-7.983", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_156.wav", "onoffCaption": "car horn honking at 0.03-2.714, 3.401-6.085, 6.775-9.459", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_158.wav", "onoffCaption": "tapping clicking clanking at 3.057-6.497, 7.876-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_161.wav", "onoffCaption": "spraying at 0.049-1.151, 2.004-2.8", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_166.wav", "onoffCaption": "woman laughing at 1.442-5.9", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_168.wav", "onoffCaption": "sheep goat bleating at 0.016-2.016", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_172.wav", "onoffCaption": "door knocking at 0.153-2.28, 3.142-5.706, 6.305-8.869", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_175.wav", "onoffCaption": "cow mooing at 0.61-3.41 and spraying at 3.012-4.114", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_183.wav", "onoffCaption": "explosion at 0.192-5.114, 5.844-8.596", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_184.wav", "onoffCaption": "sheep goat bleating at 0.322-2.322", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_190.wav", "onoffCaption": "whistling at 2.571-7.165", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_197.wav", "onoffCaption": "tapping clicking clanking at 1.043-4.483, 5.786-9.226", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_199.wav", "onoffCaption": "duck quacking at 3.246-5.246 and cat meowing at 7.245-8.635", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_200.wav", "onoffCaption": "explosion at 3.045-5.797, 7.133-9.196", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_2.wav", "onoffCaption": "door knocking at 2.42-4.984", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_5.wav", "onoffCaption": "burping belching at 3.676-5.835", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_test/syn_12.wav", "onoffCaption": "sheep goat bleating at 1.611-3.611 and sneeze at 5.808-9.161", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_test/syn_15.wav", "onoffCaption": "tapping clicking clanking at 0.807-4.247", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_22.wav", "onoffCaption": "whistling at 3.354-7.948", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_25.wav", "onoffCaption": "burping belching at 2.316-5.273, 6.42-9.377", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_31.wav", "onoffCaption": "woman laughing at 0.674-5.132, 6.464-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_36.wav", "onoffCaption": "door slamming at 0.106-2.345, 2.885-5.124, 5.997-8.236", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_38.wav", "onoffCaption": "cat meowing at 0.245-1.571, 3.125-4.451, 5.016-6.342", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_40.wav", "onoffCaption": "door knocking at 2.051-4.178, 4.942-7.506 and cow mooing at 2.928-5.728", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_46.wav", "onoffCaption": "door slamming at 0.382-1.793, 2.674-4.913", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_47.wav", "onoffCaption": "spraying at 0.719-1.515, 2.813-3.915, 4.469-5.265 and cow mooing at 1.592-4.392, 4.998-7.798", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_49.wav", "onoffCaption": "sheep goat bleating at 0.44-2.44, 3.141-5.141 and tapping clicking clanking at 1.283-4.723, 6.144-8.215", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_51.wav", "onoffCaption": "train horn at 0.258-3.258, 4.737-7.277", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_53.wav", "onoffCaption": "dog barking at 0.072-2.072, 3.076-5.076, 6.003-8.003", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_54.wav", "onoffCaption": "train horn at 0.347-3.347, 4.652-7.652", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_63.wav", "onoffCaption": "train horn at 0.507-3.507 and cat meowing at 7.463-8.789 and dog barking at 7.612-9.612", "frequencyCaption": "train horn one times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 1.521-3.521", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_70.wav", "onoffCaption": "whistling at 2.267-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_77.wav", "onoffCaption": "cow mooing at 0.75-3.55", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_79.wav", "onoffCaption": "dog barking at 1.282-3.282, 4.117-6.117, 6.789-8.789", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_81.wav", "onoffCaption": "gunshot at 0.019-2.019, 2.851-4.851, 5.918-7.918", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_86.wav", "onoffCaption": "whistling at 1.438-6.032 and woman laughing at 2.351-5.115, 6.601-9.365", "frequencyCaption": "whistling one times and woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_88.wav", "onoffCaption": "sheep goat bleating at 3.021-5.021, 6.26-8.26", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_92.wav", "onoffCaption": "door slamming at 0.346-1.757, 2.569-3.98, 5.839-7.25 and tapping clicking clanking at 2.508-5.948", "frequencyCaption": "door slamming three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_95.wav", "onoffCaption": "door slamming at 2.522-3.933, 5.673-7.084, 8.486-9.897", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_102.wav", "onoffCaption": "door knocking at 2.145-4.272, 4.881-7.008", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_105.wav", "onoffCaption": "train horn at 0.682-3.682, 4.465-6.698, 7.809-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_test/syn_111.wav", "onoffCaption": "whistling at 0.032-4.626, 6.182-10.0 and door slamming at 0.753-2.164", "frequencyCaption": "whistling two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_116.wav", "onoffCaption": "burping belching at 3.577-5.736, 6.261-9.218", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_118.wav", "onoffCaption": "sneeze at 3.124-6.477", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_121.wav", "onoffCaption": "car horn honking at 0.782-3.743, 4.51-7.194, 7.76-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 2.222-4.222, 6.493-8.493", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_126.wav", "onoffCaption": "sneeze at 2.136-6.289 and car horn honking at 2.473-5.434, 7.027-9.711", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_128.wav", "onoffCaption": "sheep goat bleating at 0.291-2.291 and door knocking at 0.293-2.42, 3.227-5.791", "frequencyCaption": "sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_132.wav", "onoffCaption": "sheep goat bleating at 0.295-2.295 and spraying at 0.328-1.124, 2.065-3.167, 4.421-5.217 and duck quacking at 0.387-2.387, 2.967-4.785, 5.384-7.384", "frequencyCaption": "sheep goat bleating one times and spraying three times and duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_135.wav", "onoffCaption": "tapping clicking clanking at 0.458-3.898, 5.425-8.865", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_139.wav", "onoffCaption": "thump thud at 2.477-4.777, 6.095-8.77", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_143.wav", "onoffCaption": "spraying at 2.679-3.475 and explosion at 5.945-10.0", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_144.wav", "onoffCaption": "duck quacking at 1.162-2.98, 3.994-5.994, 8.158-9.976", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_150.wav", "onoffCaption": "gunshot at 1.946-3.946, 4.6-6.6, 7.322-9.322", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_157.wav", "onoffCaption": "train horn at 1.991-7.071", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_159.wav", "onoffCaption": "door slamming at 3.182-5.421, 7.675-9.086", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_160.wav", "onoffCaption": "spraying at 0.179-0.975 and whistling at 3.947-10.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_167.wav", "onoffCaption": "burping belching at 0.386-3.343, 4.105-6.264 and gunshot at 4.772-6.772", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_169.wav", "onoffCaption": "sneeze at 0.56-4.713, 5.69-7.783", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_173.wav", "onoffCaption": "sheep goat bleating at 0.834-2.834, 3.932-5.932, 6.656-8.656", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_174.wav", "onoffCaption": "dog barking at 0.021-2.021, 2.529-4.529, 5.505-7.505", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_176.wav", "onoffCaption": "woman laughing at 2.645-5.409, 7.198-9.435", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_182.wav", "onoffCaption": "cow mooing at 0.007-2.807 and gunshot at 1.124-3.124", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_185.wav", "onoffCaption": "spraying at 2.564-3.666 and door knocking at 6.756-9.32", "frequencyCaption": "spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_189.wav", "onoffCaption": "door slamming at 2.717-4.956, 5.586-6.997", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_191.wav", "onoffCaption": "burping belching at 2.833-4.992, 6.271-8.43", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_193.wav", "onoffCaption": "cow mooing at 0.942-3.742, 4.83-7.454", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_196.wav", "onoffCaption": "spraying at 3.461-4.563", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_test/syn_198.wav", "onoffCaption": "gunshot at 1.546-3.546, 4.501-6.501, 7.428-9.428", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_7.wav", "onoffCaption": "spraying at 0.113-0.909 and burping belching at 0.623-3.58", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_9.wav", "onoffCaption": "cow mooing at 1.06-3.86", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_10.wav", "onoffCaption": "door knocking at 0.3-2.864, 5.022-7.586", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_17.wav", "onoffCaption": "dog barking at 3.791-5.791, 7.757-9.757", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_19.wav", "onoffCaption": "gunshot at 0.007-2.007 and spraying at 4.251-5.047", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 0.017-3.457, 5.475-7.882", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_27.wav", "onoffCaption": "dog barking at 2.012-4.012, 4.76-6.76", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 2.18-5.62, 6.49-9.93", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_33.wav", "onoffCaption": "dog barking at 2.805-4.805, 5.866-7.866 and car horn honking at 5.136-8.097", "frequencyCaption": "dog barking two times and car horn honking one times"}
+{"filepath": "data/multi_event_test/syn_34.wav", "onoffCaption": "sheep goat bleating at 1.113-3.113", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_42.wav", "onoffCaption": "tapping clicking clanking at 2.443-5.883, 7.179-9.684", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_45.wav", "onoffCaption": "cat meowing at 0.324-1.65 and train horn at 4.186-9.266", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_56.wav", "onoffCaption": "tapping clicking clanking at 1.696-5.136, 6.886-9.533", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_58.wav", "onoffCaption": "door slamming at 2.48-3.891", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_test/syn_61.wav", "onoffCaption": "explosion at 2.489-5.241, 5.792-8.521 and train horn at 2.512-7.592 and woman laughing at 6.424-9.188", "frequencyCaption": "explosion two times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_66.wav", "onoffCaption": "sheep goat bleating at 1.634-3.634", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_68.wav", "onoffCaption": "car horn honking at 0.051-3.012, 4.062-6.746, 7.319-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_72.wav", "onoffCaption": "spraying at 0.013-0.809, 1.742-2.844 and thump thud at 1.117-3.792 and dog barking at 6.065-8.065", "frequencyCaption": "spraying two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_75.wav", "onoffCaption": "explosion at 0.266-5.188, 6.431-9.183", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_83.wav", "onoffCaption": "whistling at 2.863-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_84.wav", "onoffCaption": "burping belching at 2.009-4.966, 6.768-8.927", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_90.wav", "onoffCaption": "gunshot at 0.175-2.175", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_97.wav", "onoffCaption": "cat meowing at 3.666-5.056", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_99.wav", "onoffCaption": "duck quacking at 0.697-2.515, 3.677-5.677", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_100.wav", "onoffCaption": "cat meowing at 0.122-1.512 and sheep goat bleating at 0.564-2.564, 3.078-5.078, 5.762-7.762", "frequencyCaption": "cat meowing one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_107.wav", "onoffCaption": "spraying at 0.005-1.107, 3.385-4.487 and dog barking at 1.269-3.269, 4.85-6.85 and tapping clicking clanking at 1.455-4.895, 5.47-8.91", "frequencyCaption": "spraying two times and dog barking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_109.wav", "onoffCaption": "cow mooing at 1.573-4.373 and gunshot at 7.482-9.482", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_113.wav", "onoffCaption": "whistling at 0.12-4.714 and tapping clicking clanking at 0.731-4.171", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_114.wav", "onoffCaption": "car horn honking at 3.216-5.9 and door knocking at 3.814-6.378", "frequencyCaption": "car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_124.wav", "onoffCaption": "gunshot at 2.794-4.794, 5.712-7.712", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_130.wav", "onoffCaption": "dog barking at 0.835-2.835, 3.911-5.911, 6.459-8.459", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_137.wav", "onoffCaption": "door knocking at 0.152-2.716 and cow mooing at 1.559-4.183 and gunshot at 5.826-7.826", "frequencyCaption": "door knocking one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_141.wav", "onoffCaption": "sneeze at 0.816-4.969, 5.643-9.796", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_146.wav", "onoffCaption": "sneeze at 0.145-4.298, 5.107-8.031 and cat meowing at 1.128-2.454", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_148.wav", "onoffCaption": "duck quacking at 3.185-5.003, 5.701-7.701 and cow mooing at 3.469-6.093", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_152.wav", "onoffCaption": "tapping clicking clanking at 0.851-4.291, 4.863-7.054 and train horn at 5.524-8.524", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_155.wav", "onoffCaption": "tapping clicking clanking at 0.869-4.309 and gunshot at 1.402-3.402 and cat meowing at 6.9-8.226", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_162.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.074-5.074, 5.829-7.829", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_165.wav", "onoffCaption": "thump thud at 0.322-2.622, 4.239-6.914 and whistling at 0.361-9.161", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_171.wav", "onoffCaption": "spraying at 1.23-2.332, 3.511-4.613, 5.79-6.892 and thump thud at 1.604-3.904 and sheep goat bleating at 1.985-3.985, 4.796-6.796", "frequencyCaption": "spraying three times and thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_178.wav", "onoffCaption": "door slamming at 0.233-2.472 and woman laughing at 6.658-10.0", "frequencyCaption": "door slamming one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_180.wav", "onoffCaption": "spraying at 2.203-3.305 and cow mooing at 4.398-7.198", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_187.wav", "onoffCaption": "sneeze at 2.13-6.283, 6.866-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_194.wav", "onoffCaption": "duck quacking at 2.028-3.846, 5.612-7.43", "frequencyCaption": "duck quacking two times"}
diff --git a/data/meta_data/test-onoff-control_single-event.json b/data/meta_data/test-onoff-control_single-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..6071ff006d0d2a2618a7aaf573812f9a058ca4ac
--- /dev/null
+++ b/data/meta_data/test-onoff-control_single-event.json
@@ -0,0 +1,400 @@
+{"filepath": "data/single_event_multi_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 0.258-1.584", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_6.wav", "onoffCaption": "tapping clicking clanking at 1.246-4.686", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_8.wav", "onoffCaption": "door slamming at 2.564-4.803", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_11.wav", "onoffCaption": "dog barking at 0.084-2.084, 2.908-4.908", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_16.wav", "onoffCaption": "thump thud at 0.776-3.451", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_18.wav", "onoffCaption": "sheep goat bleating at 3.833-5.833", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_21.wav", "onoffCaption": "sheep goat bleating at 2.491-4.491", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_26.wav", "onoffCaption": "tapping clicking clanking at 0.89-4.33", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_28.wav", "onoffCaption": "sneeze at 0.109-4.262, 6.151-8.608", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_32.wav", "onoffCaption": "cow mooing at 1.486-4.11", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_35.wav", "onoffCaption": "door slamming at 0.085-2.324, 4.153-5.564", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_43.wav", "onoffCaption": "thump thud at 2.551-4.851, 5.601-8.276", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_44.wav", "onoffCaption": "burping belching at 0.979-3.138, 4.115-7.072, 7.609-9.768", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_50.wav", "onoffCaption": "car horn honking at 1.566-4.25, 6.473-9.434", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_57.wav", "onoffCaption": "train horn at 3.341-8.421", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_59.wav", "onoffCaption": "woman laughing at 2.439-5.203, 6.08-8.827", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_60.wav", "onoffCaption": "cat meowing at 0.074-1.464, 3.742-5.068", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 3.535-6.159", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_69.wav", "onoffCaption": "burping belching at 0.799-2.958", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_73.wav", "onoffCaption": "whistling at 2.868-7.462", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_74.wav", "onoffCaption": "cat meowing at 1.655-3.045", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_82.wav", "onoffCaption": "thump thud at 1.925-4.6, 5.398-7.698", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_91.wav", "onoffCaption": "duck quacking at 0.497-2.497", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_96.wav", "onoffCaption": "cat meowing at 0.044-1.37, 3.201-4.591, 5.458-6.848", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_98.wav", "onoffCaption": "woman laughing at 2.458-6.916, 7.905-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_101.wav", "onoffCaption": "burping belching at 1.697-4.654, 5.403-7.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_106.wav", "onoffCaption": "gunshot at 0.047-2.047", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_108.wav", "onoffCaption": "cat meowing at 1.96-3.35, 4.662-5.988", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_112.wav", "onoffCaption": "train horn at 3.416-8.496", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_115.wav", "onoffCaption": "sheep goat bleating at 3.021-5.021, 6.26-8.26", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.126-4.566, 6.974-9.783", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_125.wav", "onoffCaption": "car horn honking at 3.106-5.79, 6.31-9.271", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_131.wav", "onoffCaption": "cow mooing at 2.423-5.047, 6.252-9.052", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_136.wav", "onoffCaption": "tapping clicking clanking at 0.672-4.112, 5.733-7.916", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_138.wav", "onoffCaption": "burping belching at 0.093-3.05, 3.962-6.121, 7.309-9.468", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_140.wav", "onoffCaption": "duck quacking at 1.928-3.928, 5.108-6.926", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_147.wav", "onoffCaption": "burping belching at 2.269-4.428, 5.085-8.042", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_149.wav", "onoffCaption": "gunshot at 0.434-2.434", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_153.wav", "onoffCaption": "cow mooing at 3.209-5.833, 6.681-9.481", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_154.wav", "onoffCaption": "train horn at 3.48-6.48, 7.121-9.68", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_163.wav", "onoffCaption": "cow mooing at 1.335-3.959, 6.377-9.177", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_164.wav", "onoffCaption": "door slamming at 3.391-4.802, 5.918-8.157", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_170.wav", "onoffCaption": "whistling at 0.053-8.853", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_177.wav", "onoffCaption": "door knocking at 0.585-2.712, 4.192-6.756", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_179.wav", "onoffCaption": "gunshot at 2.477-4.477", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_181.wav", "onoffCaption": "door knocking at 2.753-5.317", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_186.wav", "onoffCaption": "sneeze at 2.336-6.489, 7.757-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_188.wav", "onoffCaption": "explosion at 1.933-6.855", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_192.wav", "onoffCaption": "cat meowing at 0.139-1.465, 2.845-4.235", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_195.wav", "onoffCaption": "duck quacking at 3.185-5.003, 5.701-7.701", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.203-3.16, 3.696-5.855", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_4.wav", "onoffCaption": "cat meowing at 1.562-2.888", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.838-4.278, 4.839-6.935, 7.732-9.827", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_14.wav", "onoffCaption": "tapping clicking clanking at 0.51-3.95, 5.245-8.17", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_23.wav", "onoffCaption": "cow mooing at 0.467-3.267, 4.388-7.012", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_24.wav", "onoffCaption": "thump thud at 3.239-5.539, 6.108-8.783", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_30.wav", "onoffCaption": "explosion at 2.75-5.502, 7.44-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_37.wav", "onoffCaption": "tapping clicking clanking at 2.357-5.797, 7.176-9.79", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_39.wav", "onoffCaption": "burping belching at 1.038-3.197, 4.613-7.57", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_41.wav", "onoffCaption": "car horn honking at 2.524-5.485, 6.594-9.278", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_48.wav", "onoffCaption": "train horn at 2.211-7.291", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_52.wav", "onoffCaption": "dog barking at 2.157-4.157, 5.953-7.953", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_55.wav", "onoffCaption": "spraying at 1.616-2.718, 3.653-4.449, 5.396-6.498", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_62.wav", "onoffCaption": "woman laughing at 0.881-5.339, 6.657-9.421", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_65.wav", "onoffCaption": "tapping clicking clanking at 1.976-5.416, 6.573-9.12", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_71.wav", "onoffCaption": "train horn at 2.442-7.522", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_76.wav", "onoffCaption": "door knocking at 0.618-3.182", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_78.wav", "onoffCaption": "door knocking at 0.065-2.192, 3.439-6.003", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_80.wav", "onoffCaption": "car horn honking at 3.533-6.494", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_85.wav", "onoffCaption": "gunshot at 1.931-3.931, 4.716-6.716, 7.891-9.891", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_87.wav", "onoffCaption": "thump thud at 1.759-4.059, 6.133-8.808", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_89.wav", "onoffCaption": "door knocking at 0.065-2.192, 3.164-5.728", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_93.wav", "onoffCaption": "whistling at 0.042-8.842", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_94.wav", "onoffCaption": "burping belching at 0.167-2.326, 3.873-6.83", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_103.wav", "onoffCaption": "dog barking at 1.282-3.282, 4.117-6.117, 6.789-8.789", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_104.wav", "onoffCaption": "thump thud at 1.988-4.663, 7.028-9.328", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_110.wav", "onoffCaption": "whistling at 1.555-6.149", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_117.wav", "onoffCaption": "tapping clicking clanking at 0.487-3.927", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_119.wav", "onoffCaption": "duck quacking at 2.537-4.355, 5.889-7.889", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_120.wav", "onoffCaption": "dog barking at 0.013-2.013, 3.064-5.064, 5.694-7.694", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_127.wav", "onoffCaption": "duck quacking at 0.78-2.78, 5.24-7.058", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_129.wav", "onoffCaption": "burping belching at 1.965-4.922, 6.696-8.855", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_133.wav", "onoffCaption": "train horn at 3.059-8.139", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_134.wav", "onoffCaption": "spraying at 0.184-0.98, 2.498-3.6, 4.402-5.198", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_142.wav", "onoffCaption": "cow mooing at 2.715-5.339, 6.568-9.368", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_145.wav", "onoffCaption": "cow mooing at 0.071-2.695, 4.586-7.386", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_151.wav", "onoffCaption": "duck quacking at 0.425-2.425, 4.73-6.548", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_156.wav", "onoffCaption": "thump thud at 0.071-2.746, 3.838-6.138, 7.435-9.735", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_158.wav", "onoffCaption": "burping belching at 0.027-2.186", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_161.wav", "onoffCaption": "car horn honking at 0.937-3.898, 5.036-7.72", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_166.wav", "onoffCaption": "burping belching at 0.323-3.28, 4.07-6.229, 7.049-9.208", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_168.wav", "onoffCaption": "door slamming at 0.115-1.526, 2.595-4.834, 5.389-7.628", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_172.wav", "onoffCaption": "woman laughing at 3.125-5.889", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_175.wav", "onoffCaption": "spraying at 0.007-0.803", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_183.wav", "onoffCaption": "woman laughing at 2.259-6.717, 7.786-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_184.wav", "onoffCaption": "door slamming at 3.397-4.808, 6.096-8.335", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_190.wav", "onoffCaption": "explosion at 0.228-5.15, 6.074-8.826", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_197.wav", "onoffCaption": "car horn honking at 3.732-6.416, 7.567-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_199.wav", "onoffCaption": "car horn honking at 1.911-4.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_200.wav", "onoffCaption": "train horn at 0.413-3.413", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 1.299-2.689", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_5.wav", "onoffCaption": "dog barking at 3.791-5.791, 6.571-8.571", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_12.wav", "onoffCaption": "tapping clicking clanking at 1.245-4.685", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_15.wav", "onoffCaption": "explosion at 3.815-6.567, 7.214-9.546", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_22.wav", "onoffCaption": "sheep goat bleating at 0.26-2.26, 3.592-5.592, 7.325-9.325", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_25.wav", "onoffCaption": "gunshot at 0.166-2.166, 3.749-5.749", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_31.wav", "onoffCaption": "sneeze at 3.917-8.07", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_36.wav", "onoffCaption": "sheep goat bleating at 2.86-4.86, 7.119-9.119", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_38.wav", "onoffCaption": "whistling at 2.996-7.59", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_40.wav", "onoffCaption": "woman laughing at 0.024-4.482, 5.882-8.646", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_46.wav", "onoffCaption": "tapping clicking clanking at 2.067-5.507", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_47.wav", "onoffCaption": "cow mooing at 0.008-2.808, 3.956-6.58, 7.995-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_49.wav", "onoffCaption": "dog barking at 3.464-5.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_51.wav", "onoffCaption": "whistling at 0.26-9.06", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_53.wav", "onoffCaption": "whistling at 0.748-5.342, 6.45-8.456", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_54.wav", "onoffCaption": "cow mooing at 0.48-3.28, 4.237-6.861", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_63.wav", "onoffCaption": "explosion at 1.214-6.136", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_64.wav", "onoffCaption": "whistling at 0.012-4.606, 5.649-8.052", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_70.wav", "onoffCaption": "dog barking at 4.239-6.239", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_77.wav", "onoffCaption": "train horn at 2.336-7.416", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_79.wav", "onoffCaption": "train horn at 2.15-7.23", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_81.wav", "onoffCaption": "tapping clicking clanking at 3.241-6.681", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_86.wav", "onoffCaption": "gunshot at 0.406-2.406, 4.136-6.136", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_88.wav", "onoffCaption": "car horn honking at 0.051-3.012, 4.062-6.746, 7.319-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_92.wav", "onoffCaption": "door slamming at 1.032-2.443, 4.422-6.661", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_95.wav", "onoffCaption": "woman laughing at 0.147-4.605, 5.939-8.703", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_102.wav", "onoffCaption": "duck quacking at 0.363-2.363, 2.979-4.797", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.253-1.664", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_111.wav", "onoffCaption": "dog barking at 0.562-2.562, 4.25-6.25", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_116.wav", "onoffCaption": "sheep goat bleating at 2.658-4.658", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_118.wav", "onoffCaption": "sheep goat bleating at 2.634-4.634", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_121.wav", "onoffCaption": "cat meowing at 2.182-3.508", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 2.042-4.042, 5.044-7.044", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_126.wav", "onoffCaption": "burping belching at 0.139-3.096, 4.403-6.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_128.wav", "onoffCaption": "train horn at 1.814-4.814", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_132.wav", "onoffCaption": "duck quacking at 1.582-3.582, 4.673-6.491", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_135.wav", "onoffCaption": "whistling at 1.414-6.008, 7.012-9.463", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_139.wav", "onoffCaption": "spraying at 1.819-2.615, 3.181-4.283", "frequencyCaption": "spraying two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_143.wav", "onoffCaption": "door knocking at 0.495-3.059, 4.039-6.166, 7.128-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_144.wav", "onoffCaption": "spraying at 0.584-1.686, 2.49-3.286, 3.892-4.688", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_150.wav", "onoffCaption": "duck quacking at 2.654-4.654", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_157.wav", "onoffCaption": "explosion at 2.478-5.23, 6.261-9.209", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_159.wav", "onoffCaption": "sneeze at 1.342-4.695, 6.662-9.384", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_160.wav", "onoffCaption": "woman laughing at 0.352-3.116", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_167.wav", "onoffCaption": "thump thud at 0.177-2.852, 4.459-6.759", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_169.wav", "onoffCaption": "gunshot at 0.088-2.088", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_173.wav", "onoffCaption": "explosion at 0.195-5.117", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_174.wav", "onoffCaption": "duck quacking at 0.089-2.089, 4.166-5.984", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_176.wav", "onoffCaption": "gunshot at 3.54-5.54, 7.238-9.238", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_182.wav", "onoffCaption": "car horn honking at 0.14-2.824", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_185.wav", "onoffCaption": "dog barking at 3.434-5.434, 6.333-8.333", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_189.wav", "onoffCaption": "burping belching at 0.432-3.389, 4.403-6.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_191.wav", "onoffCaption": "tapping clicking clanking at 2.168-5.608", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_193.wav", "onoffCaption": "dog barking at 3.219-5.219", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_196.wav", "onoffCaption": "duck quacking at 1.8-3.618", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_198.wav", "onoffCaption": "sheep goat bleating at 0.073-2.073", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_7.wav", "onoffCaption": "door slamming at 2.809-4.22, 6.263-8.502", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_9.wav", "onoffCaption": "sneeze at 0.07-4.223, 4.927-7.216", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_10.wav", "onoffCaption": "door slamming at 2.191-3.602", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_17.wav", "onoffCaption": "gunshot at 0.033-2.033", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_19.wav", "onoffCaption": "thump thud at 2.571-4.871, 6.726-9.401", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_20.wav", "onoffCaption": "dog barking at 2.557-4.557, 5.093-7.093, 7.963-9.963", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_27.wav", "onoffCaption": "whistling at 2.141-6.735, 7.84-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_29.wav", "onoffCaption": "woman laughing at 3.051-7.509", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_33.wav", "onoffCaption": "dog barking at 1.397-3.397, 5.014-7.014", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_34.wav", "onoffCaption": "dog barking at 0.691-2.691, 4.339-6.339, 7.597-9.597", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_42.wav", "onoffCaption": "door slamming at 0.111-1.522, 2.919-5.158", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_45.wav", "onoffCaption": "woman laughing at 0.913-5.371", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_56.wav", "onoffCaption": "cat meowing at 3.25-4.576", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_58.wav", "onoffCaption": "spraying at 2.012-2.808", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_61.wav", "onoffCaption": "sheep goat bleating at 0.44-2.44, 3.141-5.141", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_66.wav", "onoffCaption": "duck quacking at 0.199-2.199", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_68.wav", "onoffCaption": "door slamming at 0.555-1.966", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_72.wav", "onoffCaption": "duck quacking at 3.008-5.008", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_75.wav", "onoffCaption": "door slamming at 2.007-4.246, 5.403-6.814, 7.324-9.563", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_83.wav", "onoffCaption": "spraying at 0.42-1.522, 2.179-2.975, 4.216-5.012", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_84.wav", "onoffCaption": "burping belching at 1.998-4.955", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_90.wav", "onoffCaption": "whistling at 0.292-9.092", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_97.wav", "onoffCaption": "dog barking at 1.995-3.995", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_99.wav", "onoffCaption": "gunshot at 1.846-3.846, 5.067-7.067", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_100.wav", "onoffCaption": "gunshot at 2.965-4.965, 5.836-7.836", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_107.wav", "onoffCaption": "cat meowing at 0.382-1.772, 4.195-5.521, 7.481-8.871", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 1.827-3.217, 5.396-6.722, 8.387-9.777", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_113.wav", "onoffCaption": "door slamming at 1.281-3.52, 4.645-6.056", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_114.wav", "onoffCaption": "explosion at 2.267-7.189", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_124.wav", "onoffCaption": "woman laughing at 0.666-5.124, 7.521-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_130.wav", "onoffCaption": "gunshot at 2.672-4.672", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_137.wav", "onoffCaption": "train horn at 0.682-3.682, 4.465-6.698, 7.809-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 0.105-2.869", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_146.wav", "onoffCaption": "sneeze at 1.102-4.455", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_148.wav", "onoffCaption": "dog barking at 0.061-2.061, 3.265-5.265, 6.197-8.197", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_152.wav", "onoffCaption": "dog barking at 0.127-2.127", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_155.wav", "onoffCaption": "spraying at 3.549-4.651", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_162.wav", "onoffCaption": "explosion at 0.391-3.143, 3.673-5.706", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_165.wav", "onoffCaption": "whistling at 3.448-8.042", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_171.wav", "onoffCaption": "duck quacking at 2.752-4.752", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_178.wav", "onoffCaption": "tapping clicking clanking at 1.713-5.153, 6.827-9.222", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0.181-2.981", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_187.wav", "onoffCaption": "explosion at 2.424-5.176", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_194.wav", "onoffCaption": "gunshot at 2.339-4.339", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_11.wav", "onoffCaption": "door knocking at 3.808-5.935, 6.708-8.835", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_16.wav", "onoffCaption": "burping belching at 2.569-5.526", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_18.wav", "onoffCaption": "burping belching at 2.907-5.066", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_21.wav", "onoffCaption": "burping belching at 0.64-2.799", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_26.wav", "onoffCaption": "spraying at 0.127-0.923", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_28.wav", "onoffCaption": "train horn at 3.589-8.669", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_32.wav", "onoffCaption": "gunshot at 1.173-3.173, 3.96-5.96, 6.617-8.617", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_35.wav", "onoffCaption": "woman laughing at 0.948-5.406, 7.602-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_43.wav", "onoffCaption": "door slamming at 3.246-4.657, 6.312-7.723", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_44.wav", "onoffCaption": "dog barking at 1.211-3.211, 4.206-6.206, 6.728-8.728", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_50.wav", "onoffCaption": "door knocking at 0.488-3.052, 5.244-7.808", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_57.wav", "onoffCaption": "train horn at 0.177-5.257", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_59.wav", "onoffCaption": "gunshot at 0.24-2.24, 3.277-5.277, 7.394-9.394", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_60.wav", "onoffCaption": "cow mooing at 1.847-4.471, 6.336-8.96", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 2.819-5.443, 6.06-8.684", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_69.wav", "onoffCaption": "burping belching at 1.971-4.928, 6.428-9.385", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_73.wav", "onoffCaption": "dog barking at 0.094-2.094, 3.294-5.294, 6.771-8.771", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_74.wav", "onoffCaption": "cow mooing at 2.351-4.975, 5.558-8.182", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_82.wav", "onoffCaption": "woman laughing at 2.876-7.334", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_85.wav", "onoffCaption": "dog barking at 2.785-4.785", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_91.wav", "onoffCaption": "tapping clicking clanking at 1.295-4.735", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_96.wav", "onoffCaption": "door knocking at 0.452-2.579", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_98.wav", "onoffCaption": "door slamming at 2.339-4.578", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_101.wav", "onoffCaption": "spraying at 0.013-1.115, 1.805-2.907, 5.09-6.192", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_106.wav", "onoffCaption": "spraying at 2.518-3.314", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_108.wav", "onoffCaption": "gunshot at 3.946-5.946, 6.959-8.959", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_112.wav", "onoffCaption": "burping belching at 3.346-6.303, 7.74-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_115.wav", "onoffCaption": "explosion at 0.084-5.006", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 0.407-3.847", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_125.wav", "onoffCaption": "explosion at 0.371-3.123, 5.335-8.087", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_131.wav", "onoffCaption": "door slamming at 0.346-1.757, 2.569-3.98, 5.839-7.25", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_136.wav", "onoffCaption": "car horn honking at 0.066-2.75", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_138.wav", "onoffCaption": "explosion at 2.129-4.881", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_140.wav", "onoffCaption": "train horn at 1.872-6.952, 7.829-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_147.wav", "onoffCaption": "explosion at 0.38-3.132, 4.352-7.104, 7.977-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/single_event_single_identity_test/syn_149.wav", "onoffCaption": "spraying at 0.031-1.133, 1.86-2.962, 3.961-5.063", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_153.wav", "onoffCaption": "dog barking at 0.435-2.435, 4.016-6.016", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_154.wav", "onoffCaption": "explosion at 1.704-6.626", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_163.wav", "onoffCaption": "sneeze at 2.736-6.889", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_164.wav", "onoffCaption": "sneeze at 2.624-6.777", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_170.wav", "onoffCaption": "burping belching at 3.451-6.408", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_175.wav", "onoffCaption": "explosion at 1.902-6.824", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_177.wav", "onoffCaption": "door knocking at 3.219-5.346, 7.058-9.185", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_179.wav", "onoffCaption": "explosion at 2.521-5.273", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_181.wav", "onoffCaption": "train horn at 0.212-3.212", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_186.wav", "onoffCaption": "sheep goat bleating at 0.651-2.651, 3.512-5.512", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_188.wav", "onoffCaption": "whistling at 0.87-9.67", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_190.wav", "onoffCaption": "woman laughing at 0.484-3.248, 4.163-6.927", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_192.wav", "onoffCaption": "door knocking at 1.863-3.99, 5.187-7.314", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_195.wav", "onoffCaption": "cow mooing at 0.958-3.582, 5.272-7.896", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 3.109-6.549", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_14.wav", "onoffCaption": "woman laughing at 0.127-2.891", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_23.wav", "onoffCaption": "whistling at 0.074-8.874", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_24.wav", "onoffCaption": "dog barking at 0.91-2.91", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_30.wav", "onoffCaption": "whistling at 0.978-5.572", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_37.wav", "onoffCaption": "whistling at 2.107-6.701", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_39.wav", "onoffCaption": "whistling at 0.165-4.759, 5.362-9.956", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_41.wav", "onoffCaption": "sheep goat bleating at 0.023-2.023, 3.507-5.507", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_46.wav", "onoffCaption": "car horn honking at 1.978-4.939, 5.578-8.539", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_48.wav", "onoffCaption": "thump thud at 1.392-4.067, 5.357-8.032", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0.25-2.25, 3.486-5.486, 6.439-8.439", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_55.wav", "onoffCaption": "gunshot at 2.722-4.722, 6.936-8.936", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_62.wav", "onoffCaption": "burping belching at 0.459-3.416, 4.188-7.145", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_65.wav", "onoffCaption": "sheep goat bleating at 0.55-2.55, 4.457-6.457", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_71.wav", "onoffCaption": "tapping clicking clanking at 3.396-6.836", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_76.wav", "onoffCaption": "sheep goat bleating at 0.056-2.056, 3.47-5.47", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_78.wav", "onoffCaption": "train horn at 0.083-5.163, 6.748-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_80.wav", "onoffCaption": "whistling at 1.269-5.863, 6.498-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_87.wav", "onoffCaption": "car horn honking at 1.885-4.569, 5.797-8.481", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_89.wav", "onoffCaption": "train horn at 0.507-3.507", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_93.wav", "onoffCaption": "dog barking at 3.063-5.063, 6.381-8.381", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_94.wav", "onoffCaption": "duck quacking at 2.332-4.15", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_103.wav", "onoffCaption": "gunshot at 1.066-3.066", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_104.wav", "onoffCaption": "cat meowing at 0.488-1.878, 4.297-5.687, 6.263-7.653", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_110.wav", "onoffCaption": "whistling at 0.407-5.001", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_117.wav", "onoffCaption": "cat meowing at 1.091-2.481, 3.509-4.899", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_119.wav", "onoffCaption": "car horn honking at 0.202-2.886", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_120.wav", "onoffCaption": "door knocking at 2.729-5.293", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_127.wav", "onoffCaption": "sheep goat bleating at 2.262-4.262, 5.801-7.801", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_129.wav", "onoffCaption": "sheep goat bleating at 0.602-2.602, 4.548-6.548, 7.151-9.151", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_133.wav", "onoffCaption": "gunshot at 1.679-3.679, 5.98-7.98", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_134.wav", "onoffCaption": "sheep goat bleating at 0.091-2.091, 3.322-5.322", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_142.wav", "onoffCaption": "dog barking at 0.622-2.622, 5.087-7.087", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_145.wav", "onoffCaption": "train horn at 2.269-5.269", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_151.wav", "onoffCaption": "burping belching at 0.193-2.352", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_156.wav", "onoffCaption": "cow mooing at 1.573-4.373", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_158.wav", "onoffCaption": "door knocking at 1.174-3.301", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_161.wav", "onoffCaption": "spraying at 0.159-1.261, 2.033-3.135, 4.44-5.542", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_166.wav", "onoffCaption": "tapping clicking clanking at 1.641-5.081, 6.146-9.586", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_168.wav", "onoffCaption": "explosion at 3.277-8.199", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_172.wav", "onoffCaption": "gunshot at 1.58-3.58", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_183.wav", "onoffCaption": "duck quacking at 0.511-2.511", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_184.wav", "onoffCaption": "spraying at 0.044-0.84", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_197.wav", "onoffCaption": "sheep goat bleating at 2.317-4.317, 6.052-8.052", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_199.wav", "onoffCaption": "dog barking at 3.728-5.728, 6.93-8.93", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_200.wav", "onoffCaption": "thump thud at 1.717-4.017, 5.949-8.249", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_12.wav", "onoffCaption": "dog barking at 2.048-4.048", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_15.wav", "onoffCaption": "dog barking at 0.046-2.046, 4.09-6.09", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_22.wav", "onoffCaption": "whistling at 2.136-6.73", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_25.wav", "onoffCaption": "explosion at 1.944-4.696, 6.227-8.979", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_31.wav", "onoffCaption": "gunshot at 0.269-2.269, 3.559-5.559, 6.243-8.243", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_36.wav", "onoffCaption": "dog barking at 0.991-2.991", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_38.wav", "onoffCaption": "dog barking at 3.368-5.368, 6.043-8.043", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_40.wav", "onoffCaption": "sheep goat bleating at 0.185-2.185", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_single_identity_test/syn_47.wav", "onoffCaption": "door slamming at 0.106-2.345, 2.885-5.124, 5.997-8.236", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_49.wav", "onoffCaption": "duck quacking at 0.37-2.37", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_51.wav", "onoffCaption": "cat meowing at 0.245-1.571, 3.125-4.451, 5.016-6.342", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_53.wav", "onoffCaption": "cat meowing at 0.277-1.603", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_54.wav", "onoffCaption": "gunshot at 0.17-2.17, 4.644-6.644", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_63.wav", "onoffCaption": "door slamming at 1.788-4.027", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 1.736-3.736, 4.735-6.735, 7.944-9.944", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_70.wav", "onoffCaption": "sneeze at 0.231-4.384, 5.433-9.586", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_77.wav", "onoffCaption": "dog barking at 3.416-5.416, 5.973-7.973", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_79.wav", "onoffCaption": "tapping clicking clanking at 0.931-4.371", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_81.wav", "onoffCaption": "spraying at 2.201-2.997", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_86.wav", "onoffCaption": "door knocking at 0.221-2.785", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_88.wav", "onoffCaption": "cow mooing at 2.087-4.887, 6.12-8.92", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_92.wav", "onoffCaption": "train horn at 0.429-5.509, 6.408-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_95.wav", "onoffCaption": "thump thud at 2.906-5.581", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_102.wav", "onoffCaption": "thump thud at 2.581-4.881, 6.222-8.522", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.833-3.072, 4.449-6.688", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_111.wav", "onoffCaption": "door knocking at 1.124-3.688, 6.152-8.716", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_116.wav", "onoffCaption": "gunshot at 0.875-2.875, 4.735-6.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_118.wav", "onoffCaption": "cat meowing at 0.483-1.809", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_121.wav", "onoffCaption": "door knocking at 1.619-4.183", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_126.wav", "onoffCaption": "sheep goat bleating at 3.885-5.885, 7.836-9.836", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_128.wav", "onoffCaption": "tapping clicking clanking at 2.571-6.011", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_132.wav", "onoffCaption": "cat meowing at 2.927-4.317, 5.007-6.397, 6.922-8.312", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_135.wav", "onoffCaption": "door slamming at 3.195-5.434, 6.893-9.132", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_139.wav", "onoffCaption": "duck quacking at 2.765-4.583, 6.906-8.724", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_143.wav", "onoffCaption": "cat meowing at 2.231-3.621", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_144.wav", "onoffCaption": "cow mooing at 0.562-3.186, 4.31-6.934", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_150.wav", "onoffCaption": "dog barking at 0.436-2.436", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_157.wav", "onoffCaption": "sneeze at 3.222-7.375", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_159.wav", "onoffCaption": "sneeze at 2.417-6.57", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_160.wav", "onoffCaption": "tapping clicking clanking at 0.262-3.702, 5.703-9.143", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_167.wav", "onoffCaption": "cat meowing at 0.205-1.595, 2.703-4.093", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_169.wav", "onoffCaption": "train horn at 3.293-8.373", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_173.wav", "onoffCaption": "thump thud at 3.392-5.692", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_174.wav", "onoffCaption": "cat meowing at 2.478-3.804, 4.701-6.027, 7.098-8.424", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_182.wav", "onoffCaption": "door knocking at 2.598-4.725, 5.428-7.555", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_185.wav", "onoffCaption": "gunshot at 3.329-5.329, 6.811-8.811", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_189.wav", "onoffCaption": "door knocking at 2.566-4.693", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_191.wav", "onoffCaption": "cow mooing at 2.094-4.894", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_196.wav", "onoffCaption": "door knocking at 0.398-2.525, 3.558-5.685, 6.802-8.929", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_198.wav", "onoffCaption": "explosion at 3.575-6.327", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_10.wav", "onoffCaption": "duck quacking at 0.107-1.925", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_17.wav", "onoffCaption": "burping belching at 0.839-2.998, 4.442-6.601", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_19.wav", "onoffCaption": "cat meowing at 2.357-3.683, 5.023-6.349", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 2.446-5.886, 7.886-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_27.wav", "onoffCaption": "spraying at 0.301-1.403, 2.423-3.525, 4.539-5.641", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 0.69-4.13, 5.59-9.03", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_33.wav", "onoffCaption": "train horn at 2.016-7.096", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_34.wav", "onoffCaption": "burping belching at 3.636-5.795, 7.726-9.885", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_42.wav", "onoffCaption": "dog barking at 2.092-4.092", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_45.wav", "onoffCaption": "cat meowing at 2.902-4.228", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_56.wav", "onoffCaption": "train horn at 0.125-3.125", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_58.wav", "onoffCaption": "duck quacking at 0.179-2.179, 4.629-6.629", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_61.wav", "onoffCaption": "spraying at 2.685-3.787", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_66.wav", "onoffCaption": "cat meowing at 0.1-1.426, 2.691-4.017", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_68.wav", "onoffCaption": "duck quacking at 0.259-2.077", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_72.wav", "onoffCaption": "sneeze at 0.32-3.673, 4.809-8.162", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_75.wav", "onoffCaption": "door slamming at 3.048-4.459, 6.382-7.793", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_83.wav", "onoffCaption": "dog barking at 1.005-3.005, 5.367-7.367", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_84.wav", "onoffCaption": "woman laughing at 0.34-4.798, 6.685-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_90.wav", "onoffCaption": "dog barking at 0.965-2.965, 3.842-5.842, 7.713-9.713", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_97.wav", "onoffCaption": "gunshot at 1.924-3.924", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_99.wav", "onoffCaption": "door knocking at 3.167-5.294, 6.941-9.068", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_100.wav", "onoffCaption": "burping belching at 2.361-4.52, 5.23-7.389", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_107.wav", "onoffCaption": "woman laughing at 2.849-5.613, 6.83-9.594", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0.321-1.647, 2.314-3.64, 4.695-6.021", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_113.wav", "onoffCaption": "duck quacking at 1.194-3.012", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_114.wav", "onoffCaption": "duck quacking at 0.737-2.737, 3.972-5.972", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_123.wav", "onoffCaption": "woman laughing at 3.064-7.522", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_124.wav", "onoffCaption": "door slamming at 0.317-2.556, 3.904-6.143", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_130.wav", "onoffCaption": "duck quacking at 1.714-3.532, 4.074-5.892, 6.517-8.335", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_137.wav", "onoffCaption": "dog barking at 0.126-2.126, 2.714-4.714", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 3.098-7.556", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_146.wav", "onoffCaption": "dog barking at 0.087-2.087, 4.127-6.127", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_148.wav", "onoffCaption": "thump thud at 2.712-5.387", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_152.wav", "onoffCaption": "sheep goat bleating at 1.645-3.645, 5.29-7.29", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_155.wav", "onoffCaption": "woman laughing at 0.079-4.537, 5.539-9.997", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_162.wav", "onoffCaption": "door knocking at 0.465-2.592, 4.247-6.374", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_165.wav", "onoffCaption": "door slamming at 0.439-2.678", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_171.wav", "onoffCaption": "woman laughing at 0.467-3.231", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_176.wav", "onoffCaption": "burping belching at 0.432-2.591, 5.061-7.22", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_178.wav", "onoffCaption": "sheep goat bleating at 4.036-6.036, 6.704-8.704", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0.178-2.802", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_187.wav", "onoffCaption": "gunshot at 0.523-2.523, 3.427-5.427", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_193.wav", "onoffCaption": "tapping clicking clanking at 1.074-4.514, 6.811-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_194.wav", "onoffCaption": "train horn at 1.729-6.809", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 0.393-1.783, 3.975-5.365", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 2.278-3.668, 5.204-6.594", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.042-2.999, 4.324-7.281, 7.849-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_single_identity_test/syn_4.wav", "onoffCaption": "car horn honking at 0.38-3.341, 4.605-7.566", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.088-2.088", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_6.wav", "onoffCaption": "explosion at 2.796-7.718", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_7.wav", "onoffCaption": "dog barking at 2.565-4.565", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_8.wav", "onoffCaption": "burping belching at 0.45-3.407", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_9.wav", "onoffCaption": "burping belching at 0.775-3.732", "frequencyCaption": "burping belching one times"}
diff --git a/data/meta_data/train.json b/data/meta_data/train.json
new file mode 100644
index 0000000000000000000000000000000000000000..c14872a174e0c402586c8572b0104c8929c5df1e
--- /dev/null
+++ b/data/meta_data/train.json
@@ -0,0 +1,5000 @@
+{"filepath": "data/multi_event_train/syn_21.wav", "onoffCaption": "door slamming at 0.45-1.991, 3.019-5.8, 6.623-8.102", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_74.wav", "onoffCaption": "train horn at 2.817-5.697, 6.941-9.151", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_91.wav", "onoffCaption": "door knocking at 1.155-5.305", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_106.wav", "onoffCaption": "duck quacking at 0.309-2.309, 3.42-5.42, 6.714-8.714 and cow mooing at 2.038-5.007, 6.542-8.58", "frequencyCaption": "duck quacking three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_153.wav", "onoffCaption": "cow mooing at 1.592-4.602, 6.719-9.729 and explosion at 3.329-6.882", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_320.wav", "onoffCaption": "train horn at 3.325-5.48, 6.561-9.201", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_375.wav", "onoffCaption": "whistling at 0.48-4.964, 7.059-9.934", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_390.wav", "onoffCaption": "dog barking at 3.038-5.038, 5.585-7.585", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_505.wav", "onoffCaption": "thump thud at 1.59-6.04, 6.614-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_550.wav", "onoffCaption": "explosion at 0.707-5.707 and woman laughing at 3.44-5.677, 6.23-8.467", "frequencyCaption": "explosion one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_638.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_688.wav", "onoffCaption": "burping belching at 0.159-4.028, 6.032-8.977 and door knocking at 0.525-2.902, 4.436-6.813", "frequencyCaption": "burping belching two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_723.wav", "onoffCaption": "burping belching at 0.211-2.336, 2.942-5.466, 6.496-8.59 and dog barking at 0.78-2.78", "frequencyCaption": "burping belching three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_776.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_793.wav", "onoffCaption": "explosion at 2.941-5.813 and spraying at 4.494-5.575, 6.097-7.161", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_801.wav", "onoffCaption": "whistling at 2.268-8.349 and door slamming at 2.584-4.78, 5.618-7.814", "frequencyCaption": "whistling one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_854.wav", "onoffCaption": "woman laughing at 0.638-3.925, 5.136-7.364 and dog barking at 1.269-3.269", "frequencyCaption": "woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1035.wav", "onoffCaption": "tapping clicking clanking at 0.961-4.401, 6.37-9.81 and sheep goat bleating at 1.484-4.78 and door knocking at 1.608-4.664, 5.492-7.703", "frequencyCaption": "tapping clicking clanking two times and sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1060.wav", "onoffCaption": "spraying at 0.669-1.238, 2.488-3.057, 4.351-4.92 and whistling at 0.918-3.793, 5.909-8.784", "frequencyCaption": "spraying three times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1085.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1209.wav", "onoffCaption": "spraying at 1.431-2.181, 2.756-3.34, 4.475-5.722", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1213.wav", "onoffCaption": "gunshot at 3.324-5.324, 6.563-8.563", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1246.wav", "onoffCaption": "cow mooing at 1.954-6.383, 7.52-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1308.wav", "onoffCaption": "duck quacking at 0.295-2.295, 3.085-5.085, 5.734-7.734 and cat meowing at 3.868-4.879, 5.645-7.2 and cow mooing at 5.281-8.579", "frequencyCaption": "duck quacking three times and cat meowing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1312.wav", "onoffCaption": "dog barking at 3.089-5.089, 6.258-8.258", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1436.wav", "onoffCaption": "train horn at 2.446-5.806, 7.457-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1463.wav", "onoffCaption": "sneeze at 0.309-2.016 and car horn honking at 1.819-4.637 and cow mooing at 7.987-10.0", "frequencyCaption": "sneeze one times and car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1486.wav", "onoffCaption": "sheep goat bleating at 0.657-4.297, 4.952-8.592 and door slamming at 1.974-4.403, 5.259-7.688", "frequencyCaption": "sheep goat bleating two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1578.wav", "onoffCaption": "dog barking at 0.121-2.121, 3.824-5.824, 7.767-9.767", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1610.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1645.wav", "onoffCaption": "door knocking at 0.645-2.772, 3.875-6.782, 7.405-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1744.wav", "onoffCaption": "spraying at 0.033-1.519, 2.5-3.986, 4.812-6.298 and burping belching at 0.275-7.443", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1829.wav", "onoffCaption": "duck quacking at 0.235-2.235, 4.037-6.037 and burping belching at 1.845-5.024, 6.206-8.841", "frequencyCaption": "duck quacking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1899.wav", "onoffCaption": "woman laughing at 2.777-6.165, 7.557-10.0 and gunshot at 2.778-4.778", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1932.wav", "onoffCaption": "sheep goat bleating at 3.113-5.113, 5.767-7.767 and car horn honking at 3.453-5.453, 7.291-9.291 and spraying at 6.748-7.375", "frequencyCaption": "sheep goat bleating two times and car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1967.wav", "onoffCaption": "train horn at 0.91-4.39 and tapping clicking clanking at 1.016-4.456, 6.672-10.0", "frequencyCaption": "train horn one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1982.wav", "onoffCaption": "tapping clicking clanking at 2.58-6.02, 7.81-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3002.wav", "onoffCaption": "gunshot at 0.025-2.025, 3.214-5.214, 6.486-8.486", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3057.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3119.wav", "onoffCaption": "sheep goat bleating at 0.524-2.524, 3.378-5.378, 6.204-8.892", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3224.wav", "onoffCaption": "burping belching at 0.304-2.719 and woman laughing at 6.258-10.0", "frequencyCaption": "burping belching one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3271.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3294.wav", "onoffCaption": "cat meowing at 1.674-5.019 and spraying at 2.493-4.188", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3395.wav", "onoffCaption": "whistling at 0.204-5.379, 7.724-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3401.wav", "onoffCaption": "sneeze at 0.283-4.783", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3454.wav", "onoffCaption": "cow mooing at 0.34-3.309", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3627.wav", "onoffCaption": "sneeze at 2.845-4.091", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3672.wav", "onoffCaption": "duck quacking at 2.203-4.203, 5.361-7.361", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3697.wav", "onoffCaption": "cow mooing at 1.882-5.18", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3769.wav", "onoffCaption": "dog barking at 2.579-4.579", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3905.wav", "onoffCaption": "door knocking at 0.141-2.362, 2.877-5.098, 5.687-7.908", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3950.wav", "onoffCaption": "whistling at 2.603-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_149.wav", "onoffCaption": "door knocking at 0.413-2.54, 3.538-5.665, 6.803-8.93 and burping belching at 6.659-9.282", "frequencyCaption": "door knocking three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_221.wav", "onoffCaption": "cow mooing at 2.969-6.267, 7.423-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_250.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_274.wav", "onoffCaption": "tapping clicking clanking at 0.404-3.844, 5.186-8.626", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_291.wav", "onoffCaption": "dog barking at 1.244-3.244, 4.164-6.164 and door knocking at 3.598-6.168", "frequencyCaption": "dog barking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_404.wav", "onoffCaption": "gunshot at 0.415-2.545, 4.002-6.132", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_451.wav", "onoffCaption": "car horn honking at 0.154-3.067, 3.794-6.707, 7.909-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_606.wav", "onoffCaption": "cow mooing at 2.907-7.887", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_622.wav", "onoffCaption": "thump thud at 3.172-7.622", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_677.wav", "onoffCaption": "car horn honking at 0.664-3.129, 4.357-7.014", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_692.wav", "onoffCaption": "door slamming at 0.005-1.396 and burping belching at 0.843-4.349, 4.994-7.088 and cat meowing at 6.136-7.136, 8.336-9.336", "frequencyCaption": "door slamming one times and burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_739.wav", "onoffCaption": "car horn honking at 1.773-4.12, 6.106-8.453 and gunshot at 2.478-4.478", "frequencyCaption": "car horn honking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_789.wav", "onoffCaption": "train horn at 0.264-3.064, 3.64-6.44 and door slamming at 4.141-6.361, 7.801-10.0", "frequencyCaption": "train horn two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_900.wav", "onoffCaption": "door slamming at 0.098-1.098, 1.787-3.787, 4.856-7.735", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_955.wav", "onoffCaption": "thump thud at 2.224-5.891, 7.389-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_994.wav", "onoffCaption": "sneeze at 0.59-2.297", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1134.wav", "onoffCaption": "cow mooing at 3.267-6.236", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1145.wav", "onoffCaption": "train horn at 3.352-6.832", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1161.wav", "onoffCaption": "cow mooing at 2.373-5.671, 7.577-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1184.wav", "onoffCaption": "burping belching at 2.861-8.462", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1336.wav", "onoffCaption": "cow mooing at 4.099-7.109, 7.72-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1347.wav", "onoffCaption": "woman laughing at 1.672-3.955", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1408.wav", "onoffCaption": "spraying at 0.161-0.736, 2.006-4.59, 6.215-7.296 and burping belching at 1.503-5.503 and sheep goat bleating at 6.745-8.745", "frequencyCaption": "spraying three times and burping belching one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1479.wav", "onoffCaption": "car horn honking at 2.125-5.779, 7.871-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1513.wav", "onoffCaption": "door knocking at 3.332-5.635, 6.643-9.518", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1537.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1562.wav", "onoffCaption": "whistling at 3.153-8.328", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1587.wav", "onoffCaption": "cat meowing at 1.159-2.743, 4.625-6.753", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1711.wav", "onoffCaption": "gunshot at 0.349-2.349, 3.25-5.25 and whistling at 7.288-10.0", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1760.wav", "onoffCaption": "explosion at 2.27-5.138, 5.861-8.729", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1785.wav", "onoffCaption": "train horn at 1.794-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1833.wav", "onoffCaption": "thump thud at 0.115-2.454, 3.071-5.41", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1866.wav", "onoffCaption": "door knocking at 3.895-6.625", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1883.wav", "onoffCaption": "door knocking at 0.406-3.253 and woman laughing at 0.635-7.369 and explosion at 1.903-6.903", "frequencyCaption": "door knocking one times and woman laughing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1928.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1998.wav", "onoffCaption": "cat meowing at 0.728-2.283, 3.385-4.94, 5.621-7.176", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3018.wav", "onoffCaption": "burping belching at 2.502-4.625, 6.121-8.244", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3103.wav", "onoffCaption": "thump thud at 1.211-4.878, 7.154-9.492", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3156.wav", "onoffCaption": "cat meowing at 1.931-3.071, 4.724-5.864 and door knocking at 2.857-7.559", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3325.wav", "onoffCaption": "thump thud at 2.22-4.72 and dog barking at 3.589-5.589, 7.361-9.361", "frequencyCaption": "thump thud one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3354.wav", "onoffCaption": "gunshot at 0.26-2.26, 4.679-6.679", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3370.wav", "onoffCaption": "gunshot at 1.812-3.812", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3500.wav", "onoffCaption": "woman laughing at 0.391-3.091, 4.456-7.156", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3555.wav", "onoffCaption": "cow mooing at 1.331-4.3, 5.203-8.172", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3594.wav", "onoffCaption": "train horn at 0.651-4.051, 4.742-8.142", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3619.wav", "onoffCaption": "gunshot at 0.367-2.367", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3668.wav", "onoffCaption": "whistling at 2.959-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3702.wav", "onoffCaption": "gunshot at 1.811-3.811 and duck quacking at 7.641-9.641", "frequencyCaption": "gunshot one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3726.wav", "onoffCaption": "woman laughing at 1.491-3.728 and whistling at 2.496-6.98", "frequencyCaption": "woman laughing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3773.wav", "onoffCaption": "dog barking at 1.003-3.003 and door knocking at 5.52-9.288", "frequencyCaption": "dog barking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3796.wav", "onoffCaption": "cat meowing at 0.365-1.781, 2.545-3.82", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3804.wav", "onoffCaption": "explosion at 3.305-8.305 and spraying at 6.953-7.557", "frequencyCaption": "explosion one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3851.wav", "onoffCaption": "spraying at 0.101-0.952 and whistling at 1.043-6.218, 7.564-10.0", "frequencyCaption": "spraying one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_138.wav", "onoffCaption": "train horn at 3.379-5.819, 6.721-8.914", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_177.wav", "onoffCaption": "car horn honking at 0.74-3.24, 5.316-7.829 and thump thud at 1.929-4.429", "frequencyCaption": "car horn honking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_188.wav", "onoffCaption": "woman laughing at 0.562-2.799 and sheep goat bleating at 2.676-4.676, 6.672-8.672", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_192.wav", "onoffCaption": "thump thud at 3.782-6.829", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_205.wav", "onoffCaption": "cow mooing at 0.72-3.702, 5.102-8.084", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_304.wav", "onoffCaption": "tapping clicking clanking at 3.198-6.638", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_420.wav", "onoffCaption": "door slamming at 0.344-1.597, 3.048-4.026 and thump thud at 7.971-10.0", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_475.wav", "onoffCaption": "gunshot at 0.17-2.17, 3.09-5.09, 6.183-8.183", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_490.wav", "onoffCaption": "explosion at 0.558-5.558, 6.884-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_521.wav", "onoffCaption": "door knocking at 0.623-3.193 and door slamming at 6.292-6.792, 8.204-8.704", "frequencyCaption": "door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_649.wav", "onoffCaption": "door slamming at 3.421-4.594", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_653.wav", "onoffCaption": "explosion at 0.055-5.055, 7.24-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_748.wav", "onoffCaption": "whistling at 0.244-4.728 and explosion at 0.417-3.008 and door slamming at 1.002-2.393, 3.236-4.627", "frequencyCaption": "whistling one times and explosion one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_752.wav", "onoffCaption": "whistling at 1.609-9.994", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_825.wav", "onoffCaption": "thump thud at 3.622-6.122, 7.47-9.97", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_870.wav", "onoffCaption": "explosion at 0.987-4.856, 5.954-8.826", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_924.wav", "onoffCaption": "sneeze at 1.0-3.085, 4.232-6.317", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_971.wav", "onoffCaption": "car horn honking at 0.235-5.142", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1011.wav", "onoffCaption": "cow mooing at 1.805-4.774, 5.439-7.469", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1110.wav", "onoffCaption": "gunshot at 2.921-4.921, 7.411-9.411", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1262.wav", "onoffCaption": "train horn at 0.21-4.65, 6.639-8.799", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1278.wav", "onoffCaption": "gunshot at 3.554-5.554, 7.567-9.567", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1287.wav", "onoffCaption": "dog barking at 3.277-5.277, 6.629-8.629", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1363.wav", "onoffCaption": "spraying at 0.161-1.065, 3.194-4.369, 6.556-7.064", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1379.wav", "onoffCaption": "cat meowing at 3.426-4.692", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1386.wav", "onoffCaption": "door knocking at 1.508-3.668 and cow mooing at 5.206-8.216", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1447.wav", "onoffCaption": "sneeze at 0.818-3.135, 3.983-6.3 and door knocking at 1.957-4.527", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1546.wav", "onoffCaption": "spraying at 1.401-2.252, 3.819-4.67 and whistling at 1.789-6.273, 7.051-10.0", "frequencyCaption": "spraying two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1634.wav", "onoffCaption": "burping belching at 0.231-2.266 and spraying at 5.82-6.57, 8.991-9.858", "frequencyCaption": "burping belching one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1735.wav", "onoffCaption": "tapping clicking clanking at 2.463-5.903", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1817.wav", "onoffCaption": "sneeze at 1.669-3.665, 5.139-8.104", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1842.wav", "onoffCaption": "door slamming at 2.786-3.786", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1858.wav", "onoffCaption": "burping belching at 1.238-3.361", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1916.wav", "onoffCaption": "thump thud at 0.025-4.475", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1943.wav", "onoffCaption": "car horn honking at 0.017-3.192, 3.948-7.123, 7.76-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1959.wav", "onoffCaption": "door slamming at 3.143-5.369, 6.461-7.48", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3026.wav", "onoffCaption": "explosion at 0.682-5.682 and gunshot at 5.466-7.466", "frequencyCaption": "explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3069.wav", "onoffCaption": "car horn honking at 3.111-6.765 and tapping clicking clanking at 4.73-8.17", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3073.wav", "onoffCaption": "train horn at 1.238-5.438", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3096.wav", "onoffCaption": "burping belching at 0.356-3.535", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3127.wav", "onoffCaption": "gunshot at 2.644-5.15, 5.815-7.908", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3168.wav", "onoffCaption": "cow mooing at 3.298-6.28", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3172.wav", "onoffCaption": "sheep goat bleating at 0.162-2.162 and spraying at 4.646-5.497", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3197.wav", "onoffCaption": "cow mooing at 0.258-3.268, 4.919-7.929", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3200.wav", "onoffCaption": "thump thud at 3.356-5.695, 6.815-9.154", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3301.wav", "onoffCaption": "car horn honking at 0.679-3.192 and sheep goat bleating at 0.684-2.684, 5.024-7.024 and gunshot at 0.815-2.815, 3.554-5.554", "frequencyCaption": "car horn honking one times and sheep goat bleating two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3425.wav", "onoffCaption": "sneeze at 3.563-5.88", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3470.wav", "onoffCaption": "duck quacking at 2.5-4.5, 6.729-8.729", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3524.wav", "onoffCaption": "door knocking at 4.11-8.485", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3571.wav", "onoffCaption": "train horn at 2.494-5.134, 7.013-9.653", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3656.wav", "onoffCaption": "door knocking at 0.171-3.724 and door slamming at 7.536-8.685", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3757.wav", "onoffCaption": "woman laughing at 1.808-4.508 and door slamming at 7.615-9.094", "frequencyCaption": "woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3820.wav", "onoffCaption": "gunshot at 0.478-2.478, 4.009-6.009", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3875.wav", "onoffCaption": "sheep goat bleating at 1.313-3.313, 5.182-7.182", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3890.wav", "onoffCaption": "gunshot at 3.266-5.266, 5.827-7.827", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3921.wav", "onoffCaption": "sheep goat bleating at 2.646-4.646", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_50.wav", "onoffCaption": "tapping clicking clanking at 1.23-4.67", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_122.wav", "onoffCaption": "dog barking at 0.436-2.436, 4.526-6.526", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_170.wav", "onoffCaption": "sneeze at 0.528-3.736, 4.834-8.042", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_195.wav", "onoffCaption": "thump thud at 2.18-4.642, 5.753-8.215", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_218.wav", "onoffCaption": "door slamming at 2.198-4.198", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_303.wav", "onoffCaption": "cow mooing at 3.336-8.316 and dog barking at 7.061-9.061", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_351.wav", "onoffCaption": "gunshot at 0.026-2.026 and whistling at 0.049-2.924 and duck quacking at 2.674-4.674, 6.611-8.611", "frequencyCaption": "gunshot one times and whistling one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_526.wav", "onoffCaption": "duck quacking at 2.365-4.365, 5.347-7.347", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_574.wav", "onoffCaption": "door knocking at 2.41-6.026, 6.947-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_591.wav", "onoffCaption": "dog barking at 0.373-2.373 and door knocking at 4.578-8.728", "frequencyCaption": "dog barking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_707.wav", "onoffCaption": "sneeze at 0.383-4.912 and duck quacking at 6.614-8.614", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_755.wav", "onoffCaption": "thump thud at 0.247-4.697, 5.344-7.86", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_822.wav", "onoffCaption": "gunshot at 2.509-4.983, 7.46-9.934 and tapping clicking clanking at 3.058-6.498 and spraying at 3.322-4.103, 5.476-6.208, 7.466-8.116", "frequencyCaption": "gunshot two times and tapping clicking clanking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_895.wav", "onoffCaption": "door slamming at 2.693-4.217, 4.832-6.356 and woman laughing at 3.667-6.752", "frequencyCaption": "door slamming two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_939.wav", "onoffCaption": "tapping clicking clanking at 0.706-4.146 and sheep goat bleating at 2.877-4.877", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1016.wav", "onoffCaption": "whistling at 0.256-8.641 and woman laughing at 0.274-2.372 and gunshot at 3.743-5.743", "frequencyCaption": "whistling one times and woman laughing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1044.wav", "onoffCaption": "door knocking at 1.427-6.427 and thump thud at 2.18-6.63 and duck quacking at 4.935-6.935", "frequencyCaption": "door knocking one times and thump thud one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1237.wav", "onoffCaption": "sheep goat bleating at 1.542-3.542, 4.23-6.23 and thump thud at 3.688-6.188", "frequencyCaption": "sheep goat bleating two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1265.wav", "onoffCaption": "door slamming at 2.016-5.016", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1280.wav", "onoffCaption": "cat meowing at 0.119-1.479, 3.298-4.445, 5.477-6.752", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1412.wav", "onoffCaption": "duck quacking at 3.869-5.869, 7.87-9.87", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1440.wav", "onoffCaption": "whistling at 0.032-2.907, 5.337-8.212", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1509.wav", "onoffCaption": "sheep goat bleating at 2.97-6.89", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1633.wav", "onoffCaption": "door knocking at 0.593-3.13, 4.121-6.748, 7.286-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1661.wav", "onoffCaption": "door slamming at 0.077-1.096", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1683.wav", "onoffCaption": "door slamming at 0.183-1.574, 2.15-3.001, 4.15-6.068 and explosion at 7.733-10.0", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1684.wav", "onoffCaption": "train horn at 1.734-8.203", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1728.wav", "onoffCaption": "cow mooing at 3.139-6.121, 6.742-9.724", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1798.wav", "onoffCaption": "gunshot at 0.031-2.031, 3.728-5.728, 6.569-8.569", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1944.wav", "onoffCaption": "tapping clicking clanking at 2.026-5.466, 6.277-9.717", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3074.wav", "onoffCaption": "explosion at 0.526-2.59 and burping belching at 2.463-4.494, 5.45-7.481 and tapping clicking clanking at 4.892-8.332", "frequencyCaption": "explosion one times and burping belching two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3091.wav", "onoffCaption": "cow mooing at 0.669-5.649, 7.382-10.0 and door knocking at 1.429-3.781", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3207.wav", "onoffCaption": "woman laughing at 3.018-6.594, 7.428-9.674", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3252.wav", "onoffCaption": "woman laughing at 2.935-6.987, 7.556-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3255.wav", "onoffCaption": "cow mooing at 2.968-5.978 and spraying at 5.505-6.255, 7.07-7.82, 9.091-9.841", "frequencyCaption": "cow mooing one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3349.wav", "onoffCaption": "duck quacking at 4.337-6.337", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3422.wav", "onoffCaption": "door knocking at 0.853-4.603, 6.378-8.921", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3495.wav", "onoffCaption": "whistling at 0.504-2.733, 4.408-6.637", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3539.wav", "onoffCaption": "door slamming at 0.962-2.215, 3.921-5.174, 7.427-8.68 and thump thud at 2.092-6.542 and gunshot at 5.373-7.373", "frequencyCaption": "door slamming three times and thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3603.wav", "onoffCaption": "cat meowing at 1.309-2.88", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3604.wav", "onoffCaption": "sneeze at 2.882-5.199, 5.801-8.262", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3651.wav", "onoffCaption": "sheep goat bleating at 2.163-4.163, 5.115-7.115", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3718.wav", "onoffCaption": "door slamming at 0.076-1.591, 2.399-3.25, 3.84-5.131 and duck quacking at 2.94-4.94, 5.948-7.948", "frequencyCaption": "door slamming three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3926.wav", "onoffCaption": "door slamming at 1.11-2.634, 3.498-5.022, 6.331-7.855 and gunshot at 2.786-4.786, 6.387-8.387 and spraying at 7.305-8.305", "frequencyCaption": "door slamming three times and gunshot two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3974.wav", "onoffCaption": "tapping clicking clanking at 1.133-4.573, 5.176-7.376 and burping belching at 3.032-7.032", "frequencyCaption": "tapping clicking clanking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3991.wav", "onoffCaption": "dog barking at 0.342-2.342 and duck quacking at 6.157-8.157", "frequencyCaption": "dog barking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_57.wav", "onoffCaption": "gunshot at 1.054-3.054 and door slamming at 5.275-7.471", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_125.wav", "onoffCaption": "duck quacking at 1.445-3.445, 4.236-6.236, 7.055-9.055 and sneeze at 2.323-4.282", "frequencyCaption": "duck quacking three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_257.wav", "onoffCaption": "car horn honking at 2.061-6.461, 7.737-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_356.wav", "onoffCaption": "gunshot at 0.745-2.745 and cow mooing at 4.159-8.588", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_468.wav", "onoffCaption": "cow mooing at 3.409-6.378", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_472.wav", "onoffCaption": "spraying at 0.583-3.602", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_497.wav", "onoffCaption": "train horn at 0.289-4.409, 6.425-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_569.wav", "onoffCaption": "thump thud at 0.004-2.504, 4.066-6.566", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_573.wav", "onoffCaption": "tapping clicking clanking at 0.237-3.677, 5.378-7.737", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_596.wav", "onoffCaption": "thump thud at 1.237-4.284, 5.393-7.909", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_601.wav", "onoffCaption": "dog barking at 0.134-2.134 and burping belching at 5.233-10.0", "frequencyCaption": "dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_700.wav", "onoffCaption": "dog barking at 1.016-3.016, 5.055-7.055", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_838.wav", "onoffCaption": "car horn honking at 2.969-7.369", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_877.wav", "onoffCaption": "door knocking at 3.48-6.327, 6.917-8.966", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_888.wav", "onoffCaption": "duck quacking at 0.569-2.569, 3.902-5.902, 6.419-8.419", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_892.wav", "onoffCaption": "door slamming at 2.212-5.173, 6.664-7.642", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_976.wav", "onoffCaption": "train horn at 0.105-4.173, 5.738-9.806", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_989.wav", "onoffCaption": "cat meowing at 0.055-1.204, 3.495-4.507, 5.761-7.755", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_993.wav", "onoffCaption": "door knocking at 2.388-6.921 and sheep goat bleating at 2.785-6.081, 7.459-10.0", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1043.wav", "onoffCaption": "cat meowing at 0.138-1.709, 2.9-5.761 and door knocking at 4.303-6.683, 7.227-9.607", "frequencyCaption": "cat meowing two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1059.wav", "onoffCaption": "car horn honking at 0.031-4.431, 4.972-6.972, 7.512-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1142.wav", "onoffCaption": "tapping clicking clanking at 1.363-4.803, 5.875-8.084", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1158.wav", "onoffCaption": "door knocking at 0.327-2.827, 3.933-6.701 and car horn honking at 2.786-7.035", "frequencyCaption": "door knocking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1230.wav", "onoffCaption": "thump thud at 0.0-3.667, 6.045-9.712", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1331.wav", "onoffCaption": "gunshot at 0.021-2.021 and explosion at 0.033-2.762, 5.092-7.821", "frequencyCaption": "gunshot one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1415.wav", "onoffCaption": "door slamming at 0.772-1.577, 2.542-5.516 and sneeze at 1.092-2.592", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1514.wav", "onoffCaption": "door knocking at 0.852-5.291, 6.978-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1666.wav", "onoffCaption": "tapping clicking clanking at 0.654-4.094", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1699.wav", "onoffCaption": "burping belching at 0.365-2.396, 4.748-7.307", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1767.wav", "onoffCaption": "door slamming at 2.159-3.276, 4.526-7.487", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1782.wav", "onoffCaption": "sheep goat bleating at 0.888-2.888, 4.833-6.833 and train horn at 5.074-7.748", "frequencyCaption": "sheep goat bleating two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1810.wav", "onoffCaption": "sneeze at 0.452-1.698 and duck quacking at 0.778-2.778, 3.694-5.694, 6.436-8.436", "frequencyCaption": "sneeze one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1911.wav", "onoffCaption": "woman laughing at 2.363-5.463, 7.505-9.873", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3021.wav", "onoffCaption": "explosion at 0.157-5.157, 6.381-8.972", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3120.wav", "onoffCaption": "dog barking at 0.042-2.042 and woman laughing at 1.67-4.025", "frequencyCaption": "dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3175.wav", "onoffCaption": "burping belching at 3.834-7.093, 7.803-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3248.wav", "onoffCaption": "explosion at 0.563-3.291", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3353.wav", "onoffCaption": "dog barking at 1.301-4.222, 5.025-7.946 and cat meowing at 4.497-7.738", "frequencyCaption": "dog barking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3438.wav", "onoffCaption": "spraying at 1.37-2.151, 3.483-4.052, 4.616-5.467 and door knocking at 7.804-10.0", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3477.wav", "onoffCaption": "dog barking at 1.283-3.283, 5.184-7.184", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3488.wav", "onoffCaption": "burping belching at 0.015-3.559, 4.328-7.328, 7.932-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3492.wav", "onoffCaption": "door knocking at 0.811-5.321 and dog barking at 2.454-4.454 and sneeze at 4.318-6.001", "frequencyCaption": "door knocking one times and dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3576.wav", "onoffCaption": "thump thud at 0.352-4.802 and cat meowing at 7.774-9.722", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3589.wav", "onoffCaption": "duck quacking at 0.293-2.293, 3.357-5.357, 6.355-8.355", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3593.wav", "onoffCaption": "dog barking at 0.036-2.036, 3.013-5.013, 5.719-7.719", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3705.wav", "onoffCaption": "dog barking at 3.735-5.735", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3868.wav", "onoffCaption": "train horn at 0.332-4.772, 6.048-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3872.wav", "onoffCaption": "door slamming at 0.661-1.342, 3.734-4.415, 6.72-7.401", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3897.wav", "onoffCaption": "thump thud at 2.086-4.857", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3969.wav", "onoffCaption": "thump thud at 2.95-5.45, 6.105-8.605", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3973.wav", "onoffCaption": "car horn honking at 2.448-4.961, 6.029-8.542", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3996.wav", "onoffCaption": "burping belching at 2.047-4.082, 4.884-6.919, 7.802-9.837 and explosion at 2.284-7.284 and sneeze at 2.504-3.961, 4.923-6.38, 7.442-8.899", "frequencyCaption": "burping belching three times and explosion one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_18.wav", "onoffCaption": "thump thud at 0.402-2.63, 4.475-6.937", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_69.wav", "onoffCaption": "car horn honking at 1.273-5.114 and cat meowing at 1.976-3.517, 4.196-5.223", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_202.wav", "onoffCaption": "sheep goat bleating at 3.609-5.609, 7.145-9.145", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_273.wav", "onoffCaption": "tapping clicking clanking at 0.069-3.509, 4.304-6.615", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_296.wav", "onoffCaption": "door slamming at 1.008-2.549, 4.839-6.604, 7.937-9.328", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_319.wav", "onoffCaption": "door slamming at 0.803-1.781", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_368.wav", "onoffCaption": "burping belching at 0.832-3.593, 4.807-7.037, 7.772-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_427.wav", "onoffCaption": "cow mooing at 0.73-3.699", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_456.wav", "onoffCaption": "sneeze at 0.319-1.864, 3.011-4.556 and train horn at 7.979-10.0", "frequencyCaption": "sneeze two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_625.wav", "onoffCaption": "door slamming at 1.746-2.863, 3.942-5.466", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_654.wav", "onoffCaption": "tapping clicking clanking at 0.273-3.713, 4.322-7.189, 7.877-10.0", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_849.wav", "onoffCaption": "cow mooing at 2.184-5.482, 7.665-10.0 and woman laughing at 2.751-5.034", "frequencyCaption": "cow mooing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_923.wav", "onoffCaption": "tapping clicking clanking at 3.813-7.253, 7.927-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_952.wav", "onoffCaption": "gunshot at 0.715-2.885 and whistling at 0.92-9.305", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1098.wav", "onoffCaption": "cow mooing at 0.964-3.933, 4.576-7.273", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1117.wav", "onoffCaption": "sheep goat bleating at 0.773-2.773, 3.983-5.983, 6.891-9.811", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1166.wav", "onoffCaption": "cow mooing at 2.671-5.653, 7.472-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1183.wav", "onoffCaption": "door slamming at 2.346-4.346, 5.646-7.646", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1315.wav", "onoffCaption": "sneeze at 1.326-2.938 and door slamming at 6.273-7.251, 8.082-9.06", "frequencyCaption": "sneeze one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1364.wav", "onoffCaption": "car horn honking at 1.483-5.883, 7.161-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1381.wav", "onoffCaption": "whistling at 0.727-8.598 and sneeze at 2.712-4.0", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1530.wav", "onoffCaption": "tapping clicking clanking at 2.798-6.238 and cow mooing at 2.997-6.007, 7.616-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1541.wav", "onoffCaption": "sheep goat bleating at 0.579-2.579 and cat meowing at 3.673-4.684", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1580.wav", "onoffCaption": "woman laughing at 2.859-6.435, 7.53-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1629.wav", "onoffCaption": "dog barking at 0.638-2.638", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1658.wav", "onoffCaption": "spraying at 0.41-1.011, 3.235-3.836, 5.746-6.347 and thump thud at 6.685-9.147", "frequencyCaption": "spraying three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1732.wav", "onoffCaption": "sheep goat bleating at 0.343-2.343, 3.836-5.836", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1743.wav", "onoffCaption": "explosion at 2.23-5.102, 6.265-9.265", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1834.wav", "onoffCaption": "explosion at 3.121-5.993, 6.705-9.577", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1845.wav", "onoffCaption": "cow mooing at 1.29-4.259, 6.477-8.876", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3104.wav", "onoffCaption": "burping belching at 0.345-4.345 and duck quacking at 1.432-3.432", "frequencyCaption": "burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3151.wav", "onoffCaption": "cat meowing at 0.059-1.144", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3190.wav", "onoffCaption": "sneeze at 0.959-5.488, 6.743-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3289.wav", "onoffCaption": "cat meowing at 0.933-4.174 and door knocking at 1.443-3.818 and dog barking at 1.738-3.738", "frequencyCaption": "cat meowing one times and door knocking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3306.wav", "onoffCaption": "tapping clicking clanking at 0.901-4.341 and spraying at 2.023-3.718, 4.646-5.513 and gunshot at 7.421-9.421", "frequencyCaption": "tapping clicking clanking one times and spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3377.wav", "onoffCaption": "whistling at 1.018-3.027, 4.596-7.243 and sheep goat bleating at 1.491-3.491, 4.68-6.68", "frequencyCaption": "whistling two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3392.wav", "onoffCaption": "whistling at 2.377-5.352", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3449.wav", "onoffCaption": "car horn honking at 2.799-7.311 and spraying at 7.997-8.624", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3507.wav", "onoffCaption": "door knocking at 1.768-6.143, 7.28-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3523.wav", "onoffCaption": "cat meowing at 2.864-3.874, 4.793-6.329, 7.107-8.107", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3552.wav", "onoffCaption": "burping belching at 0.815-4.375, 5.148-8.069", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3721.wav", "onoffCaption": "dog barking at 0.194-2.194, 4.11-6.11 and sneeze at 2.888-4.847, 6.691-8.65", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3750.wav", "onoffCaption": "duck quacking at 1.107-3.107, 3.755-5.755, 7.819-9.819", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3827.wav", "onoffCaption": "door slamming at 3.427-4.718, 6.79-8.181 and burping belching at 5.419-8.419", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3856.wav", "onoffCaption": "sneeze at 2.238-4.197 and woman laughing at 6.756-9.361", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_26.wav", "onoffCaption": "whistling at 2.265-9.298", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_154.wav", "onoffCaption": "cow mooing at 0.083-3.065, 4.203-6.641, 7.573-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_226.wav", "onoffCaption": "cow mooing at 3.057-6.355", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_327.wav", "onoffCaption": "duck quacking at 2.283-4.283, 6.514-8.514", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_403.wav", "onoffCaption": "sneeze at 3.343-4.577 and thump thud at 7.198-9.969", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_419.wav", "onoffCaption": "explosion at 0.829-3.429 and spraying at 3.018-3.518, 4.092-4.592, 5.533-6.033", "frequencyCaption": "explosion one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_502.wav", "onoffCaption": "whistling at 0.199-8.584 and sneeze at 3.066-5.312", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_518.wav", "onoffCaption": "duck quacking at 0.924-2.924, 4.185-6.185, 7.586-9.586", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_670.wav", "onoffCaption": "tapping clicking clanking at 0.126-3.566, 4.832-8.272", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_695.wav", "onoffCaption": "thump thud at 0.106-4.024 and door slamming at 0.509-3.29", "frequencyCaption": "thump thud one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_771.wav", "onoffCaption": "whistling at 1.179-9.19", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_794.wav", "onoffCaption": "door slamming at 2.185-5.185", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_806.wav", "onoffCaption": "sheep goat bleating at 0.148-2.148, 3.428-5.428, 6.242-8.242 and dog barking at 1.14-3.14, 5.434-7.434 and spraying at 1.374-2.225", "frequencyCaption": "sheep goat bleating three times and dog barking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_907.wav", "onoffCaption": "door slamming at 0.681-1.486, 2.744-3.549, 4.715-5.52 and car horn honking at 2.057-6.457", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1028.wav", "onoffCaption": "door knocking at 0.558-3.326, 4.653-7.472", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1032.wav", "onoffCaption": "gunshot at 0.97-3.243, 5.737-8.01 and whistling at 1.394-5.878, 6.976-9.839 and dog barking at 6.431-8.431", "frequencyCaption": "gunshot two times and whistling two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1129.wav", "onoffCaption": "duck quacking at 1.938-3.938, 4.687-6.687", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1133.wav", "onoffCaption": "door knocking at 0.162-2.662 and gunshot at 6.396-8.396", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1241.wav", "onoffCaption": "door knocking at 0.891-5.593, 6.815-9.127", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1340.wav", "onoffCaption": "tapping clicking clanking at 0.427-3.867 and train horn at 6.281-9.161", "frequencyCaption": "tapping clicking clanking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1464.wav", "onoffCaption": "sneeze at 1.81-4.127, 5.404-7.323 and dog barking at 2.707-4.707, 6.039-8.039", "frequencyCaption": "sneeze two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1481.wav", "onoffCaption": "gunshot at 2.87-4.87, 6.198-8.198", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1565.wav", "onoffCaption": "explosion at 0.188-2.276, 3.785-5.873 and woman laughing at 1.189-4.577, 5.689-7.781", "frequencyCaption": "explosion two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1617.wav", "onoffCaption": "spraying at 4.123-6.707, 8.0-8.867", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1716.wav", "onoffCaption": "burping belching at 0.962-3.288, 5.002-7.328", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1861.wav", "onoffCaption": "duck quacking at 0.358-2.358, 4.545-6.545 and sneeze at 5.937-7.214", "frequencyCaption": "duck quacking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1884.wav", "onoffCaption": "train horn at 0.134-2.934 and door knocking at 5.52-7.895", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1960.wav", "onoffCaption": "whistling at 2.883-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1985.wav", "onoffCaption": "door knocking at 1.766-3.926, 4.914-7.074", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3050.wav", "onoffCaption": "thump thud at 2.535-6.985", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3223.wav", "onoffCaption": "sneeze at 1.918-3.082", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3239.wav", "onoffCaption": "car horn honking at 1.927-6.834, 7.521-10.0 and door knocking at 3.368-8.368", "frequencyCaption": "car horn honking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3276.wav", "onoffCaption": "thump thud at 0.898-3.398", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3322.wav", "onoffCaption": "door slamming at 3.403-5.403, 7.253-9.253", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3338.wav", "onoffCaption": "door knocking at 0.462-2.65 and car horn honking at 6.561-10.0", "frequencyCaption": "door knocking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3406.wav", "onoffCaption": "sneeze at 0.145-2.064 and thump thud at 4.654-7.116", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3620.wav", "onoffCaption": "train horn at 3.384-6.584", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3675.wav", "onoffCaption": "dog barking at 3.128-5.566, 7.331-9.769", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3690.wav", "onoffCaption": "sneeze at 0.221-1.515, 2.707-4.001, 4.655-5.949 and woman laughing at 4.131-6.736 and door slamming at 4.556-7.273", "frequencyCaption": "sneeze three times and woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3774.wav", "onoffCaption": "woman laughing at 0.469-3.523, 4.471-7.24", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3791.wav", "onoffCaption": "gunshot at 2.794-4.794", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3803.wav", "onoffCaption": "car horn honking at 3.31-5.775", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3819.wav", "onoffCaption": "dog barking at 1.059-3.059, 4.623-7.061", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3902.wav", "onoffCaption": "sneeze at 2.138-3.384, 4.942-6.188, 8.009-9.255", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3918.wav", "onoffCaption": "explosion at 0.429-3.301", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_73.wav", "onoffCaption": "cat meowing at 1.611-6.611", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_96.wav", "onoffCaption": "door knocking at 0.339-4.107, 5.594-8.094", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_101.wav", "onoffCaption": "spraying at 3.192-5.628, 6.912-9.348", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_140.wav", "onoffCaption": "burping belching at 0.155-2.385 and spraying at 1.861-3.594 and cat meowing at 3.753-5.941", "frequencyCaption": "burping belching one times and spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_269.wav", "onoffCaption": "car horn honking at 0.696-5.208, 7.272-9.272", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_372.wav", "onoffCaption": "spraying at 3.375-3.883, 6.226-6.734", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_397.wav", "onoffCaption": "sneeze at 3.122-4.579 and gunshot at 6.347-8.347", "frequencyCaption": "sneeze one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_516.wav", "onoffCaption": "explosion at 2.108-4.636, 5.554-8.082 and cat meowing at 2.607-3.651, 5.572-6.581", "frequencyCaption": "explosion two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_557.wav", "onoffCaption": "cow mooing at 0.457-3.467", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_724.wav", "onoffCaption": "sneeze at 2.444-5.554, 6.077-7.371", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_812.wav", "onoffCaption": "door slamming at 2.125-3.144, 5.228-6.128", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_853.wav", "onoffCaption": "tapping clicking clanking at 0.486-3.926, 5.27-8.038 and spraying at 1.615-2.615, 3.253-4.51, 5.128-7.256", "frequencyCaption": "tapping clicking clanking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_909.wav", "onoffCaption": "tapping clicking clanking at 0.702-4.142, 5.289-7.949", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_948.wav", "onoffCaption": "explosion at 3.11-6.95", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1067.wav", "onoffCaption": "thump thud at 0.112-3.159 and burping belching at 0.191-2.556, 4.378-6.804", "frequencyCaption": "thump thud one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1082.wav", "onoffCaption": "sneeze at 2.318-4.263, 5.955-7.9", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1199.wav", "onoffCaption": "dog barking at 1.241-6.878", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1214.wav", "onoffCaption": "sheep goat bleating at 2.417-4.417, 5.379-7.613 and cow mooing at 5.246-8.228", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1431.wav", "onoffCaption": "thump thud at 0.131-3.178, 5.554-8.167", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1470.wav", "onoffCaption": "spraying at 0.07-1.803, 2.542-3.623, 4.2-4.722", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1495.wav", "onoffCaption": "thump thud at 0.204-3.871 and door knocking at 5.99-10.0", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1642.wav", "onoffCaption": "woman laughing at 1.271-8.716", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1759.wav", "onoffCaption": "cat meowing at 3.734-5.044, 5.722-7.032, 7.879-9.189", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1935.wav", "onoffCaption": "door slamming at 0.387-1.068, 3.499-5.499, 6.392-8.392", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1974.wav", "onoffCaption": "train horn at 0.923-3.363, 4.373-6.429 and sneeze at 4.66-6.16", "frequencyCaption": "train horn two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1991.wav", "onoffCaption": "sneeze at 4.039-5.142, 5.76-7.288 and whistling at 5.589-7.598", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3005.wav", "onoffCaption": "door slamming at 2.507-3.831, 4.569-7.052, 8.333-9.233", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3044.wav", "onoffCaption": "door knocking at 0.268-4.778, 5.999-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3293.wav", "onoffCaption": "sneeze at 0.238-3.886 and train horn at 0.981-4.221, 6.222-8.722 and whistling at 1.282-4.257, 4.777-7.669", "frequencyCaption": "sneeze one times and train horn two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3388.wav", "onoffCaption": "cow mooing at 2.335-5.345, 7.12-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3412.wav", "onoffCaption": "spraying at 0.422-2.55, 3.472-5.6", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3453.wav", "onoffCaption": "dog barking at 0.309-2.309, 2.831-4.831, 5.437-7.437 and tapping clicking clanking at 2.146-5.586, 6.531-9.971", "frequencyCaption": "dog barking three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3509.wav", "onoffCaption": "tapping clicking clanking at 0.602-4.042, 4.582-6.631 and burping belching at 6.007-8.237 and car horn honking at 6.487-8.952", "frequencyCaption": "tapping clicking clanking two times and burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3548.wav", "onoffCaption": "burping belching at 1.491-4.75, 5.696-8.955 and gunshot at 1.925-4.018 and duck quacking at 3.075-5.075", "frequencyCaption": "burping belching two times and gunshot one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3916.wav", "onoffCaption": "burping belching at 0.422-2.453, 4.46-7.46 and explosion at 4.672-7.544", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3957.wav", "onoffCaption": "tapping clicking clanking at 0.344-3.784, 5.183-8.623 and dog barking at 3.175-5.175", "frequencyCaption": "tapping clicking clanking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_32.wav", "onoffCaption": "whistling at 1.978-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_67.wav", "onoffCaption": "sneeze at 1.634-3.947", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_82.wav", "onoffCaption": "sheep goat bleating at 2.09-4.09, 5.408-7.408 and door knocking at 3.653-6.055", "frequencyCaption": "sheep goat bleating two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_115.wav", "onoffCaption": "thump thud at 1.139-3.639, 4.976-7.438, 7.978-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_228.wav", "onoffCaption": "burping belching at 1.339-3.37, 4.917-7.12 and dog barking at 5.396-7.396, 7.946-9.946", "frequencyCaption": "burping belching two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_267.wav", "onoffCaption": "explosion at 0.424-2.517 and door knocking at 6.923-9.111", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_282.wav", "onoffCaption": "train horn at 2.266-4.746", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_298.wav", "onoffCaption": "door slamming at 0.022-0.703, 1.758-3.011", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_333.wav", "onoffCaption": "dog barking at 0.293-2.293 and door slamming at 4.141-4.946, 5.517-7.88, 8.508-9.189", "frequencyCaption": "dog barking one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_366.wav", "onoffCaption": "spraying at 2.509-3.017 and cat meowing at 6.585-8.169", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_383.wav", "onoffCaption": "door knocking at 0.352-3.727, 4.364-7.739", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_399.wav", "onoffCaption": "cow mooing at 2.658-5.627, 7.837-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_458.wav", "onoffCaption": "whistling at 1.057-9.068 and explosion at 4.286-7.158", "frequencyCaption": "whistling one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_543.wav", "onoffCaption": "cow mooing at 1.056-4.038, 5.843-8.812", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_631.wav", "onoffCaption": "gunshot at 2.915-4.915, 6.448-8.448", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_730.wav", "onoffCaption": "cat meowing at 3.413-4.562, 6.613-7.762", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_765.wav", "onoffCaption": "car horn honking at 0.945-4.786, 5.97-9.811", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_780.wav", "onoffCaption": "whistling at 0.72-8.09", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_847.wav", "onoffCaption": "burping belching at 1.931-9.099", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1026.wav", "onoffCaption": "tapping clicking clanking at 2.614-6.054 and gunshot at 2.933-4.933, 6.04-8.04 and door slamming at 3.285-4.402", "frequencyCaption": "tapping clicking clanking one times and gunshot two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1073.wav", "onoffCaption": "train horn at 2.629-5.949, 6.477-9.797", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1096.wav", "onoffCaption": "door slamming at 2.215-4.215 and whistling at 7.117-10.0", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1168.wav", "onoffCaption": "duck quacking at 0.384-2.384", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1200.wav", "onoffCaption": "tapping clicking clanking at 0.097-3.537, 4.07-6.583 and train horn at 3.145-6.679", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1255.wav", "onoffCaption": "woman laughing at 0.763-4.151 and cow mooing at 5.988-8.97", "frequencyCaption": "woman laughing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1425.wav", "onoffCaption": "gunshot at 0.046-2.046, 2.925-4.925, 5.487-7.487", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1603.wav", "onoffCaption": "sneeze at 0.369-1.533", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1656.wav", "onoffCaption": "dog barking at 0.624-2.624", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1718.wav", "onoffCaption": "woman laughing at 1.553-3.651", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1921.wav", "onoffCaption": "duck quacking at 0.196-2.196 and tapping clicking clanking at 4.003-7.443", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3011.wav", "onoffCaption": "door knocking at 0.914-3.484 and sheep goat bleating at 2.889-4.889, 6.023-8.023", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3237.wav", "onoffCaption": "door knocking at 2.281-4.53 and cow mooing at 7.956-10.0", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3262.wav", "onoffCaption": "cat meowing at 1.237-4.582, 5.9-7.26 and spraying at 2.83-3.562, 4.76-5.492, 7.812-8.544 and burping belching at 4.274-7.476", "frequencyCaption": "cat meowing two times and spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3278.wav", "onoffCaption": "duck quacking at 3.644-5.644, 7.565-9.565", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3287.wav", "onoffCaption": "sheep goat bleating at 0.66-2.66, 3.969-5.969", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3363.wav", "onoffCaption": "dog barking at 2.469-4.469, 4.984-6.984 and sneeze at 5.02-6.52", "frequencyCaption": "dog barking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3379.wav", "onoffCaption": "cow mooing at 2.657-5.667, 6.343-9.006", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3447.wav", "onoffCaption": "car horn honking at 0.089-3.308 and cow mooing at 0.324-4.753, 6.437-9.419", "frequencyCaption": "car horn honking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3634.wav", "onoffCaption": "cow mooing at 0.997-5.426, 6.53-10.0 and duck quacking at 2.647-4.647, 5.454-7.454", "frequencyCaption": "cow mooing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3661.wav", "onoffCaption": "dog barking at 1.926-4.326, 6.456-8.456", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3684.wav", "onoffCaption": "duck quacking at 2.617-4.617, 5.935-7.935", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3735.wav", "onoffCaption": "door knocking at 1.259-5.634 and tapping clicking clanking at 3.553-6.993", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3858.wav", "onoffCaption": "door knocking at 2.954-5.203, 5.759-8.008", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3943.wav", "onoffCaption": "spraying at 0.374-2.81, 4.757-5.507", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_28.wav", "onoffCaption": "thump thud at 1.425-4.472, 5.233-8.28", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_98.wav", "onoffCaption": "sneeze at 3.369-5.482 and sheep goat bleating at 7.725-9.725", "frequencyCaption": "sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_232.wav", "onoffCaption": "train horn at 2.722-9.191", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_329.wav", "onoffCaption": "woman laughing at 0.281-2.386, 3.414-5.66 and gunshot at 0.769-2.769", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_417.wav", "onoffCaption": "tapping clicking clanking at 0.114-3.554 and door slamming at 5.506-7.03, 8.007-9.007", "frequencyCaption": "tapping clicking clanking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_442.wav", "onoffCaption": "car horn honking at 0.262-3.175 and spraying at 2.001-3.763, 4.393-4.915", "frequencyCaption": "car horn honking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_559.wav", "onoffCaption": "door slamming at 0.382-2.84, 4.65-6.129, 7.21-9.436 and spraying at 1.394-2.144, 3.499-4.249, 5.971-6.721", "frequencyCaption": "door slamming three times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_664.wav", "onoffCaption": "car horn honking at 2.537-7.049", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_681.wav", "onoffCaption": "duck quacking at 3.168-5.168, 6.271-8.271", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_808.wav", "onoffCaption": "explosion at 3.702-7.542", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_913.wav", "onoffCaption": "explosion at 1.558-4.43, 5.523-8.276", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_946.wav", "onoffCaption": "duck quacking at 0.315-2.315, 4.783-6.783 and dog barking at 4.076-6.076", "frequencyCaption": "duck quacking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1069.wav", "onoffCaption": "train horn at 1.208-4.448, 5.077-7.082", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1127.wav", "onoffCaption": "car horn honking at 0.006-2.792 and spraying at 6.323-6.845, 7.424-7.946", "frequencyCaption": "car horn honking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1172.wav", "onoffCaption": "gunshot at 0.688-2.688, 4.069-6.069", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1197.wav", "onoffCaption": "train horn at 0.206-3.006, 4.395-7.195", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1301.wav", "onoffCaption": "cow mooing at 1.996-6.425", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1354.wav", "onoffCaption": "cow mooing at 1.681-4.663", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1524.wav", "onoffCaption": "dog barking at 0.558-2.558, 3.395-5.395, 6.074-8.074 and sheep goat bleating at 1.962-3.962", "frequencyCaption": "dog barking three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1571.wav", "onoffCaption": "thump thud at 0.702-2.93, 4.42-7.262 and explosion at 1.683-4.555 and burping belching at 2.168-5.427", "frequencyCaption": "thump thud two times and explosion one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1594.wav", "onoffCaption": "thump thud at 0.537-4.987, 6.972-9.172", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1619.wav", "onoffCaption": "dog barking at 2.648-5.086, 5.864-8.302 and woman laughing at 6.145-8.227", "frequencyCaption": "dog barking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1702.wav", "onoffCaption": "cat meowing at 2.721-6.066, 6.608-7.693, 8.471-9.618", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1757.wav", "onoffCaption": "gunshot at 1.51-3.51, 5.146-7.386 and thump thud at 2.06-4.56", "frequencyCaption": "gunshot two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1796.wav", "onoffCaption": "tapping clicking clanking at 0.181-3.621 and duck quacking at 2.364-4.364 and train horn at 7.722-10.0", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1820.wav", "onoffCaption": "cat meowing at 0.661-4.006, 4.781-6.047, 6.6-7.685 and woman laughing at 6.484-9.569", "frequencyCaption": "cat meowing three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1875.wav", "onoffCaption": "dog barking at 2.881-4.881 and cow mooing at 7.22-10.0", "frequencyCaption": "dog barking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1890.wav", "onoffCaption": "spraying at 0.707-1.648, 3.552-4.493, 6.603-7.544", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3110.wav", "onoffCaption": "cat meowing at 0.442-2.436 and duck quacking at 1.45-3.45, 3.969-5.969, 6.517-8.517", "frequencyCaption": "cat meowing one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3145.wav", "onoffCaption": "train horn at 0.327-6.796 and door knocking at 6.15-8.525", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3336.wav", "onoffCaption": "thump thud at 2.128-4.628, 5.384-8.155", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3347.wav", "onoffCaption": "explosion at 3.829-7.669", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3386.wav", "onoffCaption": "door knocking at 2.413-6.029, 6.842-9.217", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3408.wav", "onoffCaption": "explosion at 3.363-8.363", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3513.wav", "onoffCaption": "woman laughing at 1.221-5.273, 6.213-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3546.wav", "onoffCaption": "gunshot at 2.824-4.824", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3711.wav", "onoffCaption": "spraying at 0.063-0.585, 1.438-1.96, 3.397-3.919 and cow mooing at 1.311-4.293, 6.776-9.397", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3760.wav", "onoffCaption": "duck quacking at 0.042-2.042, 3.248-5.248, 6.411-8.411", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3785.wav", "onoffCaption": "tapping clicking clanking at 0.561-4.001, 6.477-9.917", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3817.wav", "onoffCaption": "spraying at 0.006-0.61", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3842.wav", "onoffCaption": "sneeze at 0.021-1.521, 2.134-3.634", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3959.wav", "onoffCaption": "burping belching at 1.098-4.098, 5.432-7.873", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_59.wav", "onoffCaption": "train horn at 1.154-5.484 and cat meowing at 8.009-9.021", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_164.wav", "onoffCaption": "burping belching at 0.151-4.151, 6.609-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_181.wav", "onoffCaption": "sneeze at 1.324-4.972, 6.303-9.951", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_216.wav", "onoffCaption": "thump thud at 0.386-2.886, 4.342-6.391 and door slamming at 0.64-1.779", "frequencyCaption": "thump thud two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_243.wav", "onoffCaption": "gunshot at 0.792-2.792, 4.535-6.535, 7.287-9.287", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_358.wav", "onoffCaption": "door knocking at 2.31-6.843", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_429.wav", "onoffCaption": "dog barking at 0.504-2.504, 3.032-5.032 and sheep goat bleating at 7.803-9.803", "frequencyCaption": "dog barking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_433.wav", "onoffCaption": "duck quacking at 0.816-2.816 and door knocking at 5.455-9.205", "frequencyCaption": "duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_466.wav", "onoffCaption": "duck quacking at 0.463-2.463, 3.794-5.794, 7.209-9.209", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_483.wav", "onoffCaption": "whistling at 3.824-6.799, 7.476-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_528.wav", "onoffCaption": "tapping clicking clanking at 2.682-6.122, 7.378-9.544", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_598.wav", "onoffCaption": "whistling at 0.363-2.592 and woman laughing at 6.028-8.311", "frequencyCaption": "whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_615.wav", "onoffCaption": "whistling at 0.199-2.428", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_640.wav", "onoffCaption": "thump thud at 0.872-4.79, 5.882-9.8", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_836.wav", "onoffCaption": "whistling at 1.112-6.226, 7.367-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_879.wav", "onoffCaption": "door knocking at 2.646-5.702, 7.465-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_937.wav", "onoffCaption": "dog barking at 4.0-6.0, 7.139-9.139", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_962.wav", "onoffCaption": "cat meowing at 0.143-5.143, 6.242-10.0 and car horn honking at 1.548-3.895", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_987.wav", "onoffCaption": "car horn honking at 0.823-4.477, 6.669-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1018.wav", "onoffCaption": "car horn honking at 0.255-4.767, 5.748-8.386", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1103.wav", "onoffCaption": "car horn honking at 3.214-7.536", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1156.wav", "onoffCaption": "whistling at 3.274-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1325.wav", "onoffCaption": "sneeze at 2.663-5.124, 6.447-8.908 and gunshot at 3.303-5.303, 6.37-8.37", "frequencyCaption": "sneeze two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1370.wav", "onoffCaption": "cow mooing at 0.829-5.258, 6.489-9.114", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1395.wav", "onoffCaption": "duck quacking at 2.874-4.874, 7.229-9.229", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1500.wav", "onoffCaption": "thump thud at 1.523-4.294 and burping belching at 3.024-6.283", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1555.wav", "onoffCaption": "whistling at 0.923-3.798, 5.507-7.736 and gunshot at 2.133-4.133", "frequencyCaption": "whistling two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1668.wav", "onoffCaption": "burping belching at 0.939-4.141 and car horn honking at 2.97-5.788, 7.057-9.875", "frequencyCaption": "burping belching one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1726.wav", "onoffCaption": "explosion at 2.333-7.333", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1773.wav", "onoffCaption": "woman laughing at 2.701-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1804.wav", "onoffCaption": "explosion at 0.379-4.379, 6.146-8.899 and cat meowing at 0.504-1.516 and whistling at 3.877-9.052", "frequencyCaption": "explosion two times and cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1851.wav", "onoffCaption": "car horn honking at 3.222-6.008, 7.231-9.239", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1950.wav", "onoffCaption": "whistling at 3.046-8.221 and sheep goat bleating at 7.003-9.003", "frequencyCaption": "whistling one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3060.wav", "onoffCaption": "door knocking at 1.078-5.911, 7.79-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3134.wav", "onoffCaption": "thump thud at 2.316-4.655, 6.441-8.89", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3161.wav", "onoffCaption": "thump thud at 1.467-4.514, 5.451-8.498", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3184.wav", "onoffCaption": "thump thud at 0.155-2.617, 4.858-7.467 and dog barking at 6.665-8.665", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3209.wav", "onoffCaption": "car horn honking at 0.261-4.102, 5.871-9.712", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3312.wav", "onoffCaption": "door knocking at 0.796-3.108 and thump thud at 6.247-10.0", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3436.wav", "onoffCaption": "door knocking at 2.451-6.219, 6.734-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3479.wav", "onoffCaption": "car horn honking at 0.4-2.865", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3537.wav", "onoffCaption": "duck quacking at 0.27-2.27 and car horn honking at 6.138-10.0", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3562.wav", "onoffCaption": "sneeze at 2.315-3.592, 5.084-6.372, 7.543-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3587.wav", "onoffCaption": "car horn honking at 1.174-3.687", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3744.wav", "onoffCaption": "tapping clicking clanking at 1.635-5.075, 5.921-9.361", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3829.wav", "onoffCaption": "duck quacking at 0.57-2.57, 3.69-5.69", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3833.wav", "onoffCaption": "duck quacking at 2.933-4.933, 6.082-8.082", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3866.wav", "onoffCaption": "cow mooing at 0.249-3.259 and sheep goat bleating at 2.129-4.129, 4.999-6.999", "frequencyCaption": "cow mooing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3883.wav", "onoffCaption": "whistling at 0.366-8.377", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3928.wav", "onoffCaption": "gunshot at 2.58-4.58, 6.96-8.96 and door knocking at 3.521-5.898", "frequencyCaption": "gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3998.wav", "onoffCaption": "car horn honking at 1.23-4.817, 7.133-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_16.wav", "onoffCaption": "woman laughing at 1.302-3.894, 4.898-7.49", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_43.wav", "onoffCaption": "sheep goat bleating at 1.381-3.381, 4.024-6.942, 7.88-9.88", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_131.wav", "onoffCaption": "train horn at 2.615-5.975", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_259.wav", "onoffCaption": "dog barking at 1.087-3.087, 5.583-7.583", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_317.wav", "onoffCaption": "door slamming at 0.245-2.01, 3.859-5.624 and explosion at 1.731-6.731", "frequencyCaption": "door slamming two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_342.wav", "onoffCaption": "woman laughing at 0.341-2.907, 4.104-6.67, 7.981-10.0 and train horn at 3.333-5.973", "frequencyCaption": "woman laughing three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_499.wav", "onoffCaption": "woman laughing at 0.549-3.634, 4.963-7.2", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_532.wav", "onoffCaption": "woman laughing at 2.264-4.856, 5.681-7.7", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_567.wav", "onoffCaption": "whistling at 0.302-8.052", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_582.wav", "onoffCaption": "sheep goat bleating at 0.147-2.147, 4.118-6.118", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_714.wav", "onoffCaption": "door slamming at 1.142-3.142, 5.231-7.231", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_741.wav", "onoffCaption": "explosion at 2.483-5.483, 5.99-8.99", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_863.wav", "onoffCaption": "spraying at 2.737-4.432, 5.492-7.187, 7.813-9.508", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_886.wav", "onoffCaption": "dog barking at 0.055-2.055, 3.364-5.364, 6.632-8.632 and sneeze at 2.661-4.978", "frequencyCaption": "dog barking three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_978.wav", "onoffCaption": "duck quacking at 3.758-5.758, 6.667-8.667", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1002.wav", "onoffCaption": "cat meowing at 2.9-5.804, 8.273-9.583", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1057.wav", "onoffCaption": "train horn at 0.616-3.056 and woman laughing at 2.221-5.502", "frequencyCaption": "train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1119.wav", "onoffCaption": "gunshot at 0.328-2.328, 3.264-5.264, 6.146-8.146", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1224.wav", "onoffCaption": "sheep goat bleating at 1.04-3.04", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1271.wav", "onoffCaption": "woman laughing at 3.563-6.658, 7.737-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1293.wav", "onoffCaption": "burping belching at 1.074-4.276, 5.175-7.481", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1294.wav", "onoffCaption": "dog barking at 0.663-2.663, 3.258-5.258 and explosion at 0.759-5.759 and sheep goat bleating at 1.279-3.279, 5.169-7.169", "frequencyCaption": "dog barking two times and explosion one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1388.wav", "onoffCaption": "whistling at 0.114-8.125 and gunshot at 1.695-3.695, 4.819-6.819 and explosion at 4.022-7.022", "frequencyCaption": "whistling one times and gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1401.wav", "onoffCaption": "thump thud at 0.379-2.841, 3.369-5.708, 6.811-9.748", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1454.wav", "onoffCaption": "door knocking at 2.104-4.941 and explosion at 5.967-10.0", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1627.wav", "onoffCaption": "thump thud at 0.558-3.058, 4.154-6.977", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1672.wav", "onoffCaption": "cat meowing at 2.314-7.314 and spraying at 5.588-8.048", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1697.wav", "onoffCaption": "woman laughing at 0.865-3.634, 4.773-7.73", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1769.wav", "onoffCaption": "tapping clicking clanking at 0.115-3.555 and duck quacking at 5.928-7.928 and dog barking at 6.357-8.795", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1905.wav", "onoffCaption": "thump thud at 2.283-4.783, 5.933-8.433 and sneeze at 7.176-9.095", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3035.wav", "onoffCaption": "whistling at 1.36-6.535", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3085.wav", "onoffCaption": "tapping clicking clanking at 2.948-6.388, 7.912-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3213.wav", "onoffCaption": "door slamming at 0.257-1.736 and whistling at 0.27-2.279, 3.595-6.47", "frequencyCaption": "door slamming one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3214.wav", "onoffCaption": "thump thud at 2.382-4.721, 5.845-8.607", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3246.wav", "onoffCaption": "tapping clicking clanking at 0.331-3.771 and cow mooing at 6.725-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3308.wav", "onoffCaption": "woman laughing at 0.073-5.112 and duck quacking at 0.649-2.649", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3463.wav", "onoffCaption": "dog barking at 0.095-2.095, 3.294-5.294, 6.96-8.96", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3486.wav", "onoffCaption": "sheep goat bleating at 0.432-3.432, 4.38-6.38", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3578.wav", "onoffCaption": "dog barking at 0.095-2.495, 3.497-5.935, 7.9-9.9", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3610.wav", "onoffCaption": "explosion at 0.757-3.757, 5.449-8.449", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3642.wav", "onoffCaption": "door knocking at 0.246-2.746, 5.236-7.736", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3645.wav", "onoffCaption": "tapping clicking clanking at 1.919-5.359", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3759.wav", "onoffCaption": "door knocking at 2.919-5.649, 6.984-9.048", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3899.wav", "onoffCaption": "woman laughing at 2.366-5.135, 6.523-9.277", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3932.wav", "onoffCaption": "woman laughing at 0.413-2.696, 3.687-6.456, 7.42-9.645", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3967.wav", "onoffCaption": "woman laughing at 0.22-2.312, 3.688-5.78", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3982.wav", "onoffCaption": "whistling at 0.296-5.796, 6.441-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_11.wav", "onoffCaption": "spraying at 0.048-0.675, 2.079-3.163 and explosion at 6.554-10.0", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_44.wav", "onoffCaption": "sheep goat bleating at 0.24-2.24, 3.277-5.277, 7.394-9.394", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_163.wav", "onoffCaption": "duck quacking at 0.386-2.386, 3.148-5.148 and sheep goat bleating at 3.949-5.949", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_186.wav", "onoffCaption": "door slamming at 0.044-1.161 and burping belching at 1.036-4.316", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_310.wav", "onoffCaption": "sneeze at 1.983-4.386, 5.995-7.495", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_345.wav", "onoffCaption": "spraying at 1.07-3.462 and explosion at 4.971-9.971", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_535.wav", "onoffCaption": "woman laughing at 2.988-5.343, 6.095-8.681", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_608.wav", "onoffCaption": "explosion at 2.457-5.329", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_713.wav", "onoffCaption": "duck quacking at 0.237-2.237, 3.979-5.979, 7.934-9.934", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_746.wav", "onoffCaption": "train horn at 0.123-4.453, 5.099-9.429", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_831.wav", "onoffCaption": "dog barking at 0.341-2.341, 2.889-4.889, 6.503-8.503", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_881.wav", "onoffCaption": "tapping clicking clanking at 0.683-4.123 and thump thud at 7.869-10.0", "frequencyCaption": "tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1005.wav", "onoffCaption": "door knocking at 1.384-7.444 and explosion at 5.109-8.109", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1050.wav", "onoffCaption": "explosion at 2.575-4.639, 5.936-8.753", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1223.wav", "onoffCaption": "sheep goat bleating at 1.401-3.401, 5.028-7.028", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1276.wav", "onoffCaption": "gunshot at 2.945-4.945, 6.205-8.706", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1338.wav", "onoffCaption": "tapping clicking clanking at 0.123-3.563, 4.576-8.016 and sheep goat bleating at 6.188-8.188", "frequencyCaption": "tapping clicking clanking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1453.wav", "onoffCaption": "train horn at 2.239-5.119, 5.847-8.727", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1548.wav", "onoffCaption": "thump thud at 0.102-3.149, 3.878-6.469", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1620.wav", "onoffCaption": "gunshot at 0.078-2.552 and thump thud at 0.946-4.864", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1675.wav", "onoffCaption": "door knocking at 2.094-4.469, 5.66-8.035", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1690.wav", "onoffCaption": "door slamming at 1.034-3.517 and spraying at 7.495-8.099", "frequencyCaption": "door slamming one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1957.wav", "onoffCaption": "sneeze at 1.883-3.59, 4.837-6.544, 8.083-9.79", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3067.wav", "onoffCaption": "thump thud at 0.31-2.81 and burping belching at 6.33-9.091", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3082.wav", "onoffCaption": "cow mooing at 0.949-3.959, 5.766-7.948", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3199.wav", "onoffCaption": "thump thud at 1.786-4.833, 6.026-9.073", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3241.wav", "onoffCaption": "sheep goat bleating at 2.492-4.492, 5.259-7.259, 7.956-9.956", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3431.wav", "onoffCaption": "woman laughing at 0.558-3.258, 3.954-6.039, 6.909-9.007 and whistling at 2.627-7.111", "frequencyCaption": "woman laughing three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3617.wav", "onoffCaption": "train horn at 0.479-6.948, 7.832-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3935.wav", "onoffCaption": "sneeze at 1.497-3.958, 4.871-6.197", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_136.wav", "onoffCaption": "thump thud at 0.761-3.532 and whistling at 5.796-10.0", "frequencyCaption": "thump thud one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_244.wav", "onoffCaption": "spraying at 0.842-1.709 and door slamming at 2.747-3.247", "frequencyCaption": "spraying one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_461.wav", "onoffCaption": "cat meowing at 2.228-3.978 and explosion at 3.017-5.024, 6.181-8.188 and cow mooing at 5.927-8.909", "frequencyCaption": "cat meowing one times and explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_484.wav", "onoffCaption": "car horn honking at 1.94-5.781", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_560.wav", "onoffCaption": "dog barking at 2.223-4.223, 5.485-7.485", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_585.wav", "onoffCaption": "spraying at 3.005-3.609, 4.518-6.646, 7.52-9.215", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_612.wav", "onoffCaption": "dog barking at 0.012-2.012, 4.228-6.228, 6.945-8.945", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_864.wav", "onoffCaption": "explosion at 1.183-4.183 and woman laughing at 7.821-10.0", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_965.wav", "onoffCaption": "dog barking at 2.009-4.009, 5.762-7.762 and explosion at 2.108-4.861, 5.845-8.598", "frequencyCaption": "dog barking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_980.wav", "onoffCaption": "cat meowing at 0.306-1.446, 2.819-3.959", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1151.wav", "onoffCaption": "sneeze at 0.567-3.181 and explosion at 1.075-6.075", "frequencyCaption": "sneeze one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1406.wav", "onoffCaption": "sheep goat bleating at 1.684-3.684, 4.252-6.252, 7.758-9.758", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1507.wav", "onoffCaption": "cat meowing at 0.787-2.328, 4.605-5.915 and burping belching at 1.025-6.149, 7.732-9.857", "frequencyCaption": "cat meowing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1791.wav", "onoffCaption": "duck quacking at 0.18-2.18 and cat meowing at 0.29-3.531, 4.935-7.682", "frequencyCaption": "duck quacking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1803.wav", "onoffCaption": "whistling at 2.821-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1819.wav", "onoffCaption": "sneeze at 0.143-2.389, 2.913-4.832, 5.475-7.588", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1902.wav", "onoffCaption": "dog barking at 2.405-4.405, 6.495-8.495", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1918.wav", "onoffCaption": "burping belching at 3.973-6.973", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3028.wav", "onoffCaption": "sneeze at 0.602-1.879", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3032.wav", "onoffCaption": "woman laughing at 1.827-5.215, 6.908-9.263", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3129.wav", "onoffCaption": "spraying at 1.764-2.828, 4.27-5.334 and door knocking at 6.965-9.502", "frequencyCaption": "spraying two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3133.wav", "onoffCaption": "woman laughing at 1.628-3.836, 5.97-8.178", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3340.wav", "onoffCaption": "door slamming at 0.33-2.526 and cow mooing at 6.04-9.022", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3464.wav", "onoffCaption": "dog barking at 3.184-5.184, 6.269-8.269", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3481.wav", "onoffCaption": "explosion at 1.66-4.528, 6.089-8.957", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3565.wav", "onoffCaption": "cow mooing at 0.862-3.872 and spraying at 7.02-7.624, 9.258-9.862", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3580.wav", "onoffCaption": "whistling at 0.212-7.611", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3716.wav", "onoffCaption": "thump thud at 3.076-7.526", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3861.wav", "onoffCaption": "whistling at 0.139-7.889", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3884.wav", "onoffCaption": "cat meowing at 3.525-8.525", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3960.wav", "onoffCaption": "cow mooing at 3.13-6.099, 6.661-9.63", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3985.wav", "onoffCaption": "thump thud at 1.04-4.958", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_108.wav", "onoffCaption": "cow mooing at 0.556-4.985 and train horn at 0.71-8.91 and woman laughing at 2.098-4.466, 5.962-8.33", "frequencyCaption": "cow mooing one times and train horn one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_179.wav", "onoffCaption": "gunshot at 2.115-4.115, 5.617-7.617", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_211.wav", "onoffCaption": "burping belching at 0.658-6.602, 7.664-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_434.wav", "onoffCaption": "explosion at 2.431-5.16 and train horn at 7.14-10.0", "frequencyCaption": "explosion one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_647.wav", "onoffCaption": "sheep goat bleating at 1.751-3.751", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_709.wav", "onoffCaption": "cow mooing at 0.149-3.447, 4.752-7.721", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_930.wav", "onoffCaption": "duck quacking at 2.399-4.399, 5.505-7.505 and woman laughing at 3.444-5.526, 6.964-9.448 and dog barking at 4.535-6.535", "frequencyCaption": "duck quacking two times and woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_941.wav", "onoffCaption": "cow mooing at 0.379-5.359, 7.153-9.591", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1104.wav", "onoffCaption": "door slamming at 0.251-2.016, 2.613-4.378, 5.27-7.035 and gunshot at 2.093-4.093, 5.224-7.243 and sheep goat bleating at 2.203-4.203, 6.075-8.075", "frequencyCaption": "door slamming three times and gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1190.wav", "onoffCaption": "woman laughing at 0.835-3.035 and spraying at 6.648-7.223", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1239.wav", "onoffCaption": "woman laughing at 0.599-3.185, 4.736-7.341", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1289.wav", "onoffCaption": "car horn honking at 3.261-6.047, 7.282-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1322.wav", "onoffCaption": "duck quacking at 3.022-5.022, 6.303-8.303", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1377.wav", "onoffCaption": "sheep goat bleating at 0.256-2.256 and sneeze at 4.422-6.105, 7.36-9.043", "frequencyCaption": "sheep goat bleating one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1392.wav", "onoffCaption": "car horn honking at 0.604-3.117, 5.468-7.981", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1449.wav", "onoffCaption": "explosion at 2.052-4.313, 6.672-8.933", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1552.wav", "onoffCaption": "burping belching at 0.716-3.716 and duck quacking at 5.025-7.025, 7.592-9.592", "frequencyCaption": "burping belching one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1721.wav", "onoffCaption": "cow mooing at 1.991-4.973, 5.974-8.943", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1774.wav", "onoffCaption": "door slamming at 1.63-2.863, 4.075-6.075, 6.681-9.11", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1827.wav", "onoffCaption": "woman laughing at 3.327-5.919", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1856.wav", "onoffCaption": "sneeze at 2.355-4.969", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3098.wav", "onoffCaption": "car horn honking at 1.715-6.037", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3117.wav", "onoffCaption": "sneeze at 0.541-2.604, 4.626-6.689 and sheep goat bleating at 2.158-4.158, 5.39-7.39", "frequencyCaption": "sneeze two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3166.wav", "onoffCaption": "spraying at 2.954-3.805", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3183.wav", "onoffCaption": "sheep goat bleating at 0.991-2.991, 3.74-5.74, 6.273-8.273", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3315.wav", "onoffCaption": "duck quacking at 0.019-2.019, 3.177-5.177, 6.468-8.468", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3530.wav", "onoffCaption": "woman laughing at 2.851-5.443", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3541.wav", "onoffCaption": "spraying at 0.143-0.793, 1.737-2.588, 3.538-4.27 and train horn at 5.735-10.0", "frequencyCaption": "spraying three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3658.wav", "onoffCaption": "cat meowing at 0.634-2.529, 3.868-5.763 and explosion at 3.879-6.879, 7.905-10.0", "frequencyCaption": "cat meowing two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_3743.wav", "onoffCaption": "thump thud at 2.308-6.758", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3834.wav", "onoffCaption": "sneeze at 1.737-4.397 and woman laughing at 6.52-9.106", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3845.wav", "onoffCaption": "tapping clicking clanking at 3.386-6.826 and door slamming at 6.172-8.63", "frequencyCaption": "tapping clicking clanking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_260.wav", "onoffCaption": "car horn honking at 0.679-3.497, 4.414-6.414, 7.4-9.747 and thump thud at 0.765-3.104", "frequencyCaption": "car horn honking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_285.wav", "onoffCaption": "cat meowing at 2.662-3.674", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_410.wav", "onoffCaption": "sneeze at 2.922-4.918", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_445.wav", "onoffCaption": "burping belching at 2.349-4.443 and sheep goat bleating at 7.468-9.468", "frequencyCaption": "burping belching one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_636.wav", "onoffCaption": "thump thud at 0.085-3.752, 4.485-8.152", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_778.wav", "onoffCaption": "cat meowing at 0.073-1.968, 4.277-5.463", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_914.wav", "onoffCaption": "cat meowing at 2.239-3.266, 3.834-4.861, 5.407-6.434", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1120.wav", "onoffCaption": "sheep goat bleating at 3.176-5.176, 6.418-8.418", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1175.wav", "onoffCaption": "spraying at 0.156-2.284", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1248.wav", "onoffCaption": "woman laughing at 0.652-3.447, 4.195-6.99 and burping belching at 2.568-6.591, 7.652-10.0", "frequencyCaption": "woman laughing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1306.wav", "onoffCaption": "woman laughing at 0.786-3.154, 5.194-7.562 and spraying at 1.837-2.412, 4.389-7.386 and explosion at 5.217-8.211", "frequencyCaption": "woman laughing two times and spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1353.wav", "onoffCaption": "woman laughing at 2.632-4.832 and tapping clicking clanking at 7.798-10.0", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1438.wav", "onoffCaption": "dog barking at 0.127-5.764 and tapping clicking clanking at 7.741-10.0", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1488.wav", "onoffCaption": "woman laughing at 0.084-3.365, 4.111-7.392", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1523.wav", "onoffCaption": "cat meowing at 3.057-4.628, 5.333-6.904, 8.188-9.759", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1576.wav", "onoffCaption": "woman laughing at 0.377-2.602", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1593.wav", "onoffCaption": "duck quacking at 2.419-4.419, 5.123-7.123, 7.708-9.708", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1705.wav", "onoffCaption": "gunshot at 0.188-2.188, 3.006-5.006, 6.0-8.0 and cow mooing at 2.515-5.813, 6.954-9.655", "frequencyCaption": "gunshot three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1750.wav", "onoffCaption": "thump thud at 0.121-4.039, 5.53-8.03", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1872.wav", "onoffCaption": "cat meowing at 0.257-1.267 and train horn at 5.374-7.814", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1897.wav", "onoffCaption": "cow mooing at 3.107-8.087", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1969.wav", "onoffCaption": "duck quacking at 0.218-2.218", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3059.wav", "onoffCaption": "burping belching at 0.083-2.286", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3142.wav", "onoffCaption": "woman laughing at 0.185-2.885, 4.611-7.311 and burping belching at 2.382-5.822 and duck quacking at 5.807-7.807", "frequencyCaption": "woman laughing two times and burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3364.wav", "onoffCaption": "door slamming at 1.137-3.333, 4.292-6.488, 7.604-9.8", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3381.wav", "onoffCaption": "duck quacking at 2.558-4.558 and dog barking at 3.089-5.089, 6.993-8.993", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3514.wav", "onoffCaption": "sneeze at 1.47-6.47, 7.74-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3629.wav", "onoffCaption": "duck quacking at 2.888-4.888, 7.107-9.107", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3699.wav", "onoffCaption": "woman laughing at 0.287-3.568, 4.142-6.944, 7.835-9.884", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3732.wav", "onoffCaption": "cow mooing at 0.364-3.662, 4.801-7.605", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3782.wav", "onoffCaption": "spraying at 0.427-2.887, 4.194-5.451", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3810.wav", "onoffCaption": "thump thud at 2.015-5.682, 7.857-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_35.wav", "onoffCaption": "thump thud at 3.337-5.676, 7.377-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_60.wav", "onoffCaption": "cat meowing at 0.035-1.619 and explosion at 0.384-3.256, 3.9-6.772 and spraying at 3.01-5.47", "frequencyCaption": "cat meowing one times and explosion two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_147.wav", "onoffCaption": "gunshot at 0.792-2.792, 3.335-5.335", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_235.wav", "onoffCaption": "cat meowing at 0.778-1.787, 2.553-3.565 and gunshot at 2.067-4.067, 6.151-8.151", "frequencyCaption": "cat meowing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_334.wav", "onoffCaption": "duck quacking at 3.078-5.078", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_361.wav", "onoffCaption": "car horn honking at 2.942-5.289, 5.9-8.413", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_384.wav", "onoffCaption": "duck quacking at 1.589-3.589, 4.206-6.206", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_511.wav", "onoffCaption": "woman laughing at 0.627-2.91 and thump thud at 6.446-9.217", "frequencyCaption": "woman laughing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_663.wav", "onoffCaption": "whistling at 0.402-4.886 and dog barking at 2.751-4.751, 6.775-8.775 and door slamming at 8.642-9.142", "frequencyCaption": "whistling one times and dog barking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_679.wav", "onoffCaption": "sneeze at 0.263-2.326, 3.042-5.105, 6.243-8.306", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_686.wav", "onoffCaption": "cat meowing at 0.012-3.357, 5.746-7.478", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_737.wav", "onoffCaption": "thump thud at 3.081-7.531", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_762.wav", "onoffCaption": "car horn honking at 3.016-7.528", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_787.wav", "onoffCaption": "sneeze at 0.267-2.506, 3.642-5.881 and cat meowing at 1.741-6.101", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_815.wav", "onoffCaption": "dog barking at 0.838-2.838, 4.348-6.348 and woman laughing at 5.774-7.982", "frequencyCaption": "dog barking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1021.wav", "onoffCaption": "explosion at 2.342-5.342, 5.92-8.92 and burping belching at 2.438-5.438", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1074.wav", "onoffCaption": "burping belching at 2.126-4.233, 5.913-8.02", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1207.wav", "onoffCaption": "sheep goat bleating at 0.427-2.427, 3.863-5.863, 6.741-8.741", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1252.wav", "onoffCaption": "thump thud at 2.779-5.55, 6.397-8.569", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1349.wav", "onoffCaption": "burping belching at 3.22-5.251, 5.977-8.943", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1477.wav", "onoffCaption": "door knocking at 0.762-5.058", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1492.wav", "onoffCaption": "train horn at 0.546-3.706", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1539.wav", "onoffCaption": "whistling at 0.69-5.174, 5.721-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1589.wav", "onoffCaption": "cow mooing at 1.966-4.935, 6.103-9.072", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1604.wav", "onoffCaption": "sneeze at 0.952-5.481, 6.461-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1651.wav", "onoffCaption": "thump thud at 0.031-4.481, 5.06-7.831 and cow mooing at 1.949-6.929 and gunshot at 2.458-4.458, 5.594-7.594", "frequencyCaption": "thump thud two times and cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1868.wav", "onoffCaption": "dog barking at 2.736-4.736, 5.262-7.262", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1973.wav", "onoffCaption": "sheep goat bleating at 0.478-2.478", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1996.wav", "onoffCaption": "sheep goat bleating at 0.337-2.337, 4.071-6.071", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3043.wav", "onoffCaption": "duck quacking at 1.343-3.343 and door slamming at 6.658-8.576", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3158.wav", "onoffCaption": "spraying at 2.079-3.841, 4.379-6.141, 7.114-8.876", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3230.wav", "onoffCaption": "car horn honking at 0.711-5.111 and burping belching at 7.655-10.0", "frequencyCaption": "car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3265.wav", "onoffCaption": "gunshot at 0.842-2.861, 3.712-5.731, 6.886-8.905", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3280.wav", "onoffCaption": "thump thud at 0.098-4.548 and train horn at 1.037-4.807, 5.772-8.572", "frequencyCaption": "thump thud one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3331.wav", "onoffCaption": "explosion at 0.213-3.213, 4.115-6.28 and gunshot at 7.473-9.473", "frequencyCaption": "explosion two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3415.wav", "onoffCaption": "tapping clicking clanking at 1.663-5.103 and thump thud at 1.668-4.168, 5.582-8.082", "frequencyCaption": "tapping clicking clanking one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3633.wav", "onoffCaption": "cow mooing at 0.348-4.777", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3666.wav", "onoffCaption": "whistling at 3.006-7.49", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3683.wav", "onoffCaption": "tapping clicking clanking at 0.355-3.795, 6.028-8.443", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3728.wav", "onoffCaption": "door knocking at 0.613-2.993 and sheep goat bleating at 1.374-3.374, 4.594-6.594", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3767.wav", "onoffCaption": "cow mooing at 1.682-4.651, 6.711-9.68", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3798.wav", "onoffCaption": "sneeze at 2.406-6.054, 7.169-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3911.wav", "onoffCaption": "explosion at 3.713-6.713, 7.614-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_5000.wav", "onoffCaption": "woman laughing at 2.643-4.868, 7.346-9.571", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_85.wav", "onoffCaption": "thump thud at 0.092-4.542, 5.186-9.636", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_112.wav", "onoffCaption": "door slamming at 2.357-3.474, 4.082-6.0", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_175.wav", "onoffCaption": "whistling at 0.475-8.713", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_190.wav", "onoffCaption": "burping belching at 2.79-5.969", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_248.wav", "onoffCaption": "whistling at 0.668-3.643, 6.008-8.983", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_306.wav", "onoffCaption": "whistling at 2.191-9.535", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_353.wav", "onoffCaption": "explosion at 1.105-6.105, 7.952-9.971", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_438.wav", "onoffCaption": "sheep goat bleating at 1.945-3.945", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_523.wav", "onoffCaption": "explosion at 2.55-5.141, 5.651-8.242", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_544.wav", "onoffCaption": "gunshot at 1.072-3.242, 3.961-6.002, 7.293-9.293", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_705.wav", "onoffCaption": "train horn at 0.471-5.329", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_750.wav", "onoffCaption": "cow mooing at 2.139-5.149, 5.874-8.836", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_827.wav", "onoffCaption": "tapping clicking clanking at 0.111-3.551, 5.819-8.222 and whistling at 3.405-8.58 and explosion at 5.426-8.154", "frequencyCaption": "tapping clicking clanking two times and whistling one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_840.wav", "onoffCaption": "gunshot at 0.056-2.056 and whistling at 4.934-10.0", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1013.wav", "onoffCaption": "door slamming at 0.266-0.766, 1.741-3.659, 4.964-6.964 and explosion at 1.158-6.158", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1091.wav", "onoffCaption": "explosion at 0.702-3.302", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1108.wav", "onoffCaption": "door knocking at 0.744-3.047, 4.096-6.399", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1235.wav", "onoffCaption": "explosion at 0.214-2.967", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1260.wav", "onoffCaption": "train horn at 1.145-3.945 and explosion at 6.009-8.609", "frequencyCaption": "train horn one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1285.wav", "onoffCaption": "thump thud at 0.028-2.799 and gunshot at 0.633-2.633, 4.968-6.968", "frequencyCaption": "thump thud one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1422.wav", "onoffCaption": "whistling at 1.541-6.716", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1445.wav", "onoffCaption": "gunshot at 2.192-4.192, 6.123-8.123", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1636.wav", "onoffCaption": "cow mooing at 0.692-3.674, 4.245-7.07 and explosion at 1.708-4.576, 6.62-9.614 and door slamming at 7.173-8.312", "frequencyCaption": "cow mooing two times and explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1663.wav", "onoffCaption": "thump thud at 1.73-3.958, 5.73-8.425", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1686.wav", "onoffCaption": "spraying at 1.212-3.648, 4.274-6.71, 7.857-10.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1778.wav", "onoffCaption": "dog barking at 2.24-4.24 and explosion at 6.124-10.0", "frequencyCaption": "dog barking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1926.wav", "onoffCaption": "whistling at 3.395-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1941.wav", "onoffCaption": "whistling at 0.213-7.868 and gunshot at 3.4-5.4", "frequencyCaption": "whistling one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3016.wav", "onoffCaption": "cow mooing at 2.11-5.408", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3071.wav", "onoffCaption": "car horn honking at 0.372-2.837, 5.051-7.551", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3094.wav", "onoffCaption": "gunshot at 2.875-4.875", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3202.wav", "onoffCaption": "duck quacking at 0.628-2.628, 3.29-5.29, 6.192-8.192", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3257.wav", "onoffCaption": "car horn honking at 0.974-3.321, 5.028-7.375 and train horn at 3.119-5.759, 6.861-9.328", "frequencyCaption": "car horn honking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3319.wav", "onoffCaption": "sneeze at 3.802-5.036, 6.027-8.488", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3427.wav", "onoffCaption": "spraying at 0.506-2.966, 4.117-5.85, 7.285-9.895 and woman laughing at 1.012-4.588", "frequencyCaption": "spraying three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3440.wav", "onoffCaption": "car horn honking at 1.313-5.713, 7.865-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3497.wav", "onoffCaption": "duck quacking at 1.589-3.589, 4.21-6.21, 7.047-9.047", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3601.wav", "onoffCaption": "spraying at 0.049-1.133 and burping belching at 5.237-8.237", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3654.wav", "onoffCaption": "sneeze at 2.713-4.325", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3838.wav", "onoffCaption": "train horn at 2.023-7.738", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3923.wav", "onoffCaption": "gunshot at 0.039-2.209, 3.821-5.821", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3944.wav", "onoffCaption": "sheep goat bleating at 0.704-2.704, 4.835-6.835 and cat meowing at 1.039-2.252, 3.829-5.245", "frequencyCaption": "sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_48.wav", "onoffCaption": "door knocking at 1.798-3.925, 6.24-8.864", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_52.wav", "onoffCaption": "cat meowing at 0.243-5.243 and whistling at 1.921-4.15, 5.629-8.103", "frequencyCaption": "cat meowing one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_120.wav", "onoffCaption": "duck quacking at 2.114-4.114, 4.766-6.766, 7.77-9.77", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_252.wav", "onoffCaption": "gunshot at 0.236-2.236, 4.669-6.71", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_349.wav", "onoffCaption": "gunshot at 2.895-4.895, 7.074-9.074 and duck quacking at 3.758-5.758", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_422.wav", "onoffCaption": "spraying at 2.662-5.054, 5.777-8.169", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_477.wav", "onoffCaption": "dog barking at 2.616-4.616, 5.85-7.85", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_488.wav", "onoffCaption": "thump thud at 0.294-4.744 and explosion at 2.961-5.833", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_492.wav", "onoffCaption": "burping belching at 3.015-6.515, 7.914-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_539.wav", "onoffCaption": "duck quacking at 0.626-2.626 and gunshot at 0.647-2.647 and sheep goat bleating at 6.015-8.015", "frequencyCaption": "duck quacking one times and gunshot one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_576.wav", "onoffCaption": "cow mooing at 0.017-2.986", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_589.wav", "onoffCaption": "sneeze at 0.227-2.34", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_593.wav", "onoffCaption": "spraying at 0.736-1.386, 2.43-4.89 and burping belching at 2.475-5.734 and car horn honking at 3.11-7.51", "frequencyCaption": "spraying two times and burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_604.wav", "onoffCaption": "sneeze at 0.064-2.381, 4.864-7.181 and woman laughing at 1.889-4.658, 5.873-8.11", "frequencyCaption": "sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_868.wav", "onoffCaption": "duck quacking at 1.107-3.107, 5.382-7.382", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_872.wav", "onoffCaption": "sheep goat bleating at 1.92-4.92", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_897.wav", "onoffCaption": "woman laughing at 2.093-7.132, 7.959-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_969.wav", "onoffCaption": "thump thud at 0.204-3.871, 4.373-6.774 and duck quacking at 0.295-2.295, 4.429-6.429", "frequencyCaption": "thump thud two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_973.wav", "onoffCaption": "woman laughing at 2.152-4.636, 7.017-9.501", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_996.wav", "onoffCaption": "door knocking at 1.037-3.127 and train horn at 4.574-7.054", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1009.wav", "onoffCaption": "explosion at 0.047-5.047, 6.829-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1046.wav", "onoffCaption": "thump thud at 2.605-7.055", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1112.wav", "onoffCaption": "dog barking at 0.813-3.251", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1147.wav", "onoffCaption": "door slamming at 0.008-0.986, 2.768-3.746 and burping belching at 6.192-9.192", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1334.wav", "onoffCaption": "burping belching at 2.724-9.892", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1410.wav", "onoffCaption": "spraying at 2.671-3.298, 4.313-4.94, 5.78-6.407", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1511.wav", "onoffCaption": "cow mooing at 2.906-5.875, 6.784-9.753", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1544.wav", "onoffCaption": "tapping clicking clanking at 0.031-3.471, 4.249-7.055", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1679.wav", "onoffCaption": "whistling at 0.412-5.912", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1762.wav", "onoffCaption": "sneeze at 0.149-5.149", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1787.wav", "onoffCaption": "tapping clicking clanking at 3.695-7.135", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1815.wav", "onoffCaption": "cat meowing at 2.792-3.978, 4.638-6.388", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1840.wav", "onoffCaption": "whistling at 0.127-7.877 and duck quacking at 0.626-2.626, 4.514-6.514", "frequencyCaption": "whistling one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1914.wav", "onoffCaption": "duck quacking at 0.36-2.36 and spraying at 5.948-6.68, 7.206-7.938, 9.061-9.793", "frequencyCaption": "duck quacking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3024.wav", "onoffCaption": "whistling at 1.51-4.385, 4.941-7.136 and sheep goat bleating at 4.617-6.617", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3125.wav", "onoffCaption": "cow mooing at 0.525-3.535 and sneeze at 7.273-10.0", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3170.wav", "onoffCaption": "thump thud at 2.4-6.775 and cow mooing at 3.142-7.571", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3195.wav", "onoffCaption": "duck quacking at 3.632-5.632, 6.157-8.157", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3356.wav", "onoffCaption": "dog barking at 0.004-2.004 and sneeze at 0.554-2.867", "frequencyCaption": "dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3468.wav", "onoffCaption": "tapping clicking clanking at 1.654-5.094, 6.11-9.55", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3472.wav", "onoffCaption": "thump thud at 0.086-3.133, 4.191-7.238 and cow mooing at 2.851-6.149, 7.782-10.0", "frequencyCaption": "thump thud two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3526.wav", "onoffCaption": "burping belching at 2.329-5.329, 5.967-8.967 and spraying at 3.37-5.498", "frequencyCaption": "burping belching two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3569.wav", "onoffCaption": "woman laughing at 3.645-6.717", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3573.wav", "onoffCaption": "tapping clicking clanking at 0.078-3.518 and woman laughing at 2.07-4.438", "frequencyCaption": "tapping clicking clanking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3596.wav", "onoffCaption": "gunshot at 0.42-2.42, 3.855-5.855 and whistling at 0.821-3.796, 5.538-7.962", "frequencyCaption": "gunshot two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3700.wav", "onoffCaption": "thump thud at 2.566-6.484", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3822.wav", "onoffCaption": "tapping clicking clanking at 0.026-3.466, 4.038-7.478 and sneeze at 0.281-2.594 and whistling at 0.577-8.588", "frequencyCaption": "tapping clicking clanking two times and sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3877.wav", "onoffCaption": "sheep goat bleating at 2.689-5.985, 7.249-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3888.wav", "onoffCaption": "gunshot at 0.011-2.011, 4.076-6.076 and woman laughing at 2.145-4.353, 5.515-7.798 and burping belching at 3.798-6.321, 7.51-9.581", "frequencyCaption": "gunshot two times and woman laughing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3892.wav", "onoffCaption": "whistling at 0.003-5.178, 6.434-10.0 and gunshot at 0.281-2.281, 4.664-6.664", "frequencyCaption": "whistling two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3939.wav", "onoffCaption": "door knocking at 2.256-5.103, 6.83-8.99", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3976.wav", "onoffCaption": "sheep goat bleating at 4.316-6.316, 7.624-9.624", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3989.wav", "onoffCaption": "duck quacking at 0.753-2.753, 4.201-6.201", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3993.wav", "onoffCaption": "sheep goat bleating at 0.539-3.619, 5.059-8.139", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_39.wav", "onoffCaption": "explosion at 0.263-2.327, 4.015-6.079", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_89.wav", "onoffCaption": "sheep goat bleating at 1.399-3.399, 4.145-6.145, 6.832-9.296", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_207.wav", "onoffCaption": "train horn at 0.007-4.188 and cat meowing at 0.737-2.714", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_276.wav", "onoffCaption": "dog barking at 0.676-2.676 and sheep goat bleating at 1.893-3.893, 4.654-7.441", "frequencyCaption": "dog barking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_293.wav", "onoffCaption": "spraying at 3.538-4.795", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_338.wav", "onoffCaption": "tapping clicking clanking at 3.456-6.896", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_388.wav", "onoffCaption": "car horn honking at 0.262-4.662, 6.687-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_406.wav", "onoffCaption": "sheep goat bleating at 1.038-3.038, 4.725-6.725, 7.491-9.491", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_453.wav", "onoffCaption": "tapping clicking clanking at 0.529-3.969, 5.003-7.612", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_548.wav", "onoffCaption": "burping belching at 3.305-8.305", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_620.wav", "onoffCaption": "spraying at 1.449-1.971, 3.051-3.573, 4.44-4.962", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_651.wav", "onoffCaption": "explosion at 0.475-5.396, 6.197-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_819.wav", "onoffCaption": "cow mooing at 0.503-3.485, 4.402-7.384 and dog barking at 3.328-5.328, 6.489-8.489", "frequencyCaption": "cow mooing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_902.wav", "onoffCaption": "explosion at 0.945-5.945, 7.669-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_926.wav", "onoffCaption": "woman laughing at 1.915-5.196, 6.014-9.295", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_957.wav", "onoffCaption": "spraying at 3.156-3.783, 4.327-4.896, 5.659-6.743", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1078.wav", "onoffCaption": "dog barking at 2.572-4.572", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1136.wav", "onoffCaption": "sneeze at 1.816-3.761 and woman laughing at 6.146-9.722", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1163.wav", "onoffCaption": "thump thud at 0.616-3.663, 4.265-6.765 and cat meowing at 1.658-6.658, 7.892-8.904", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1186.wav", "onoffCaption": "tapping clicking clanking at 2.584-6.024, 7.048-9.723", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1310.wav", "onoffCaption": "cat meowing at 0.11-1.122, 1.672-3.404, 4.651-7.141", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1345.wav", "onoffCaption": "tapping clicking clanking at 0.045-3.485, 4.395-7.835", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1361.wav", "onoffCaption": "whistling at 2.651-8.484", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1384.wav", "onoffCaption": "burping belching at 1.138-3.959 and explosion at 7.473-10.0", "frequencyCaption": "burping belching one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1535.wav", "onoffCaption": "cat meowing at 0.403-1.616, 2.396-3.951", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1560.wav", "onoffCaption": "dog barking at 0.055-2.055, 2.803-4.803, 6.123-8.123", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1585.wav", "onoffCaption": "car horn honking at 1.4-5.649", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1608.wav", "onoffCaption": "tapping clicking clanking at 0.833-4.273 and sneeze at 1.635-2.738", "frequencyCaption": "tapping clicking clanking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1713.wav", "onoffCaption": "burping belching at 0.208-3.708 and train horn at 5.339-8.219", "frequencyCaption": "burping belching one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1737.wav", "onoffCaption": "woman laughing at 2.723-4.948, 6.995-9.6", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1746.wav", "onoffCaption": "cat meowing at 2.19-4.167, 5.515-7.651", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1831.wav", "onoffCaption": "gunshot at 2.959-4.959, 6.448-8.448", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1864.wav", "onoffCaption": "duck quacking at 2.461-4.461, 5.492-7.492 and woman laughing at 3.466-6.261", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1881.wav", "onoffCaption": "sheep goat bleating at 2.941-4.941, 5.649-7.649", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3101.wav", "onoffCaption": "door slamming at 0.408-2.628, 4.108-6.328 and woman laughing at 2.274-4.474 and door knocking at 3.199-6.752, 7.928-10.0", "frequencyCaption": "door slamming two times and woman laughing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3154.wav", "onoffCaption": "gunshot at 2.635-4.635", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3218.wav", "onoffCaption": "gunshot at 0.374-2.374 and sneeze at 5.184-7.798", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3269.wav", "onoffCaption": "woman laughing at 2.804-5.573", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3303.wav", "onoffCaption": "train horn at 0.168-3.408, 4.333-7.573", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3372.wav", "onoffCaption": "whistling at 2.433-5.408, 7.427-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3397.wav", "onoffCaption": "train horn at 3.047-8.791 and spraying at 3.625-4.229, 6.445-7.445", "frequencyCaption": "train horn one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3419.wav", "onoffCaption": "duck quacking at 0.225-2.225 and dog barking at 1.184-3.184, 4.339-6.339", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3502.wav", "onoffCaption": "sneeze at 0.887-3.501, 4.361-6.975 and door slamming at 2.409-3.214, 5.124-7.905", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3557.wav", "onoffCaption": "thump thud at 1.144-5.519, 6.289-8.744 and sheep goat bleating at 2.984-6.904, 7.408-9.604", "frequencyCaption": "thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3724.wav", "onoffCaption": "duck quacking at 0.417-2.417, 3.645-5.645", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3755.wav", "onoffCaption": "dog barking at 1.109-3.109, 4.896-6.896, 7.76-9.76 and duck quacking at 3.606-5.606", "frequencyCaption": "dog barking three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3794.wav", "onoffCaption": "whistling at 0.449-3.324, 4.92-7.795", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3806.wav", "onoffCaption": "car horn honking at 2.923-5.27, 6.695-9.042", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3853.wav", "onoffCaption": "car horn honking at 2.237-6.749, 7.707-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3948.wav", "onoffCaption": "cow mooing at 2.363-7.343 and gunshot at 6.261-8.501", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_23.wav", "onoffCaption": "sheep goat bleating at 2.024-4.024, 4.681-6.681, 7.699-9.699", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_76.wav", "onoffCaption": "cow mooing at 0.517-5.497", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_151.wav", "onoffCaption": "burping belching at 0.578-4.447 and door slamming at 2.144-3.468", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_223.wav", "onoffCaption": "thump thud at 0.521-3.292 and sneeze at 2.031-3.285", "frequencyCaption": "thump thud one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_239.wav", "onoffCaption": "sheep goat bleating at 0.693-2.693, 4.197-6.197", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_289.wav", "onoffCaption": "cat meowing at 0.229-1.25, 2.111-3.132, 4.459-5.48 and woman laughing at 2.584-5.684, 7.136-9.504", "frequencyCaption": "cat meowing three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_322.wav", "onoffCaption": "door slamming at 1.672-4.551, 7.01-8.401 and thump thud at 2.44-4.779", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_377.wav", "onoffCaption": "whistling at 2.592-5.567, 6.838-9.813", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_392.wav", "onoffCaption": "door slamming at 0.562-2.69, 3.661-5.789, 7.08-9.208", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_507.wav", "onoffCaption": "duck quacking at 2.858-4.858", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_675.wav", "onoffCaption": "woman laughing at 2.855-5.063, 5.692-8.111", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_690.wav", "onoffCaption": "thump thud at 2.745-4.973, 7.024-9.486", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_721.wav", "onoffCaption": "car horn honking at 0.361-4.761, 7.058-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_774.wav", "onoffCaption": "burping belching at 2.738-4.832 and cow mooing at 7.019-10.0", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_791.wav", "onoffCaption": "train horn at 0.204-2.671, 3.684-6.181 and whistling at 0.286-8.671 and burping belching at 5.965-8.195", "frequencyCaption": "train horn two times and whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_803.wav", "onoffCaption": "dog barking at 1.213-3.213, 5.389-7.389 and explosion at 3.633-6.505, 7.382-10.0", "frequencyCaption": "dog barking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_918.wav", "onoffCaption": "door knocking at 1.957-4.084, 6.555-8.682", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1037.wav", "onoffCaption": "cow mooing at 0.323-3.292, 4.556-7.542", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1211.wav", "onoffCaption": "train horn at 0.755-4.515, 5.812-9.572", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1244.wav", "onoffCaption": "gunshot at 0.224-2.698 and woman laughing at 6.812-9.912", "frequencyCaption": "gunshot one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1461.wav", "onoffCaption": "duck quacking at 0.523-2.523, 4.907-6.907 and woman laughing at 1.227-3.421", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1484.wav", "onoffCaption": "burping belching at 2.002-8.979 and tapping clicking clanking at 3.197-6.637", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1612.wav", "onoffCaption": "whistling at 2.42-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1647.wav", "onoffCaption": "gunshot at 0.059-2.059, 2.889-4.889, 5.928-7.928 and sneeze at 0.303-1.986, 2.839-4.133, 5.618-8.191", "frequencyCaption": "gunshot three times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1709.wav", "onoffCaption": "dog barking at 3.101-5.101", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1965.wav", "onoffCaption": "spraying at 3.478-4.329 and thump thud at 6.947-10.0", "frequencyCaption": "spraying one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1980.wav", "onoffCaption": "car horn honking at 3.67-6.135, 7.16-9.625", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3055.wav", "onoffCaption": "thump thud at 0.803-3.85", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3226.wav", "onoffCaption": "whistling at 2.624-7.108, 7.973-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3273.wav", "onoffCaption": "cow mooing at 3.464-6.446", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3296.wav", "onoffCaption": "sheep goat bleating at 2.22-4.22, 5.571-7.571", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3327.wav", "onoffCaption": "sheep goat bleating at 0.979-2.979, 5.085-7.085", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3368.wav", "onoffCaption": "car horn honking at 2.877-6.531, 7.446-9.989", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3403.wav", "onoffCaption": "door knocking at 2.848-5.472, 7.49-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3518.wav", "onoffCaption": "cat meowing at 0.029-1.924, 3.136-5.031", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3625.wav", "onoffCaption": "whistling at 1.721-4.696, 5.425-8.4", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3670.wav", "onoffCaption": "whistling at 1.871-9.882", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3695.wav", "onoffCaption": "explosion at 2.44-5.993 and sheep goat bleating at 4.285-6.285", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3771.wav", "onoffCaption": "explosion at 2.574-5.574, 6.762-9.762", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3907.wav", "onoffCaption": "train horn at 0.688-2.825 and door slamming at 1.492-2.47 and sneeze at 5.398-7.01, 7.662-9.274", "frequencyCaption": "train horn one times and door slamming one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_24.wav", "onoffCaption": "door slamming at 0.185-2.098, 3.22-5.133, 7.597-9.51", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_93.wav", "onoffCaption": "woman laughing at 3.523-6.804", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_104.wav", "onoffCaption": "car horn honking at 1.439-3.786, 5.888-8.792", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_156.wav", "onoffCaption": "cow mooing at 0.403-3.413 and cat meowing at 1.774-6.774", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_449.wav", "onoffCaption": "cat meowing at 0.961-2.377 and car horn honking at 1.892-5.067", "frequencyCaption": "cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_500.wav", "onoffCaption": "train horn at 0.129-3.489, 5.32-7.76", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_552.wav", "onoffCaption": "duck quacking at 1.62-3.62, 4.619-6.619 and cow mooing at 1.955-4.965, 5.474-8.484", "frequencyCaption": "duck quacking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_668.wav", "onoffCaption": "door knocking at 0.117-3.617", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_773.wav", "onoffCaption": "woman laughing at 2.284-6.336", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_796.wav", "onoffCaption": "door slamming at 0.302-1.475, 2.628-3.801, 5.758-6.931 and car horn honking at 2.659-7.566", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_804.wav", "onoffCaption": "sneeze at 1.835-4.91, 5.489-8.564", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_856.wav", "onoffCaption": "cow mooing at 0.081-5.061 and explosion at 1.242-3.833, 4.45-6.514, 7.335-10.0", "frequencyCaption": "cow mooing one times and explosion three times"}
+{"filepath": "data/multi_event_train/syn_1030.wav", "onoffCaption": "cat meowing at 0.829-1.84, 3.044-4.055, 4.829-5.84 and gunshot at 2.193-4.193, 4.87-6.87, 7.81-9.81", "frequencyCaption": "cat meowing three times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1062.wav", "onoffCaption": "whistling at 0.568-8.579 and train horn at 1.664-6.566, 7.957-10.0", "frequencyCaption": "whistling one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_1087.wav", "onoffCaption": "train horn at 2.12-5.52, 6.427-9.827 and thump thud at 2.993-6.911", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1179.wav", "onoffCaption": "thump thud at 0.225-2.725, 4.517-7.017 and train horn at 2.293-5.653, 6.302-9.662", "frequencyCaption": "thump thud two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_1434.wav", "onoffCaption": "cat meowing at 2.858-7.858", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1466.wav", "onoffCaption": "train horn at 0.155-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1483.wav", "onoffCaption": "sneeze at 1.223-2.906 and sheep goat bleating at 1.495-3.495 and dog barking at 2.465-4.465", "frequencyCaption": "sneeze one times and sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1598.wav", "onoffCaption": "door slamming at 0.59-1.707, 3.721-5.045", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1879.wav", "onoffCaption": "woman laughing at 0.565-8.01 and gunshot at 2.025-4.025", "frequencyCaption": "woman laughing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1930.wav", "onoffCaption": "train horn at 0.044-4.484 and thump thud at 0.511-2.85, 4.648-6.987 and sneeze at 3.344-7.4", "frequencyCaption": "train horn one times and thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1962.wav", "onoffCaption": "explosion at 2.685-5.685, 6.915-9.787", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1987.wav", "onoffCaption": "burping belching at 0.826-6.427, 6.931-9.038 and dog barking at 1.463-3.463, 4.507-6.507 and thump thud at 4.019-6.519", "frequencyCaption": "burping belching two times and dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3000.wav", "onoffCaption": "sneeze at 2.175-4.421, 6.906-9.152", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3052.wav", "onoffCaption": "duck quacking at 3.237-5.237", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3149.wav", "onoffCaption": "whistling at 0.327-8.712", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3221.wav", "onoffCaption": "whistling at 0.692-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3404.wav", "onoffCaption": "door knocking at 1.875-6.314", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3456.wav", "onoffCaption": "explosion at 0.557-3.557, 4.707-7.707 and gunshot at 0.902-2.902, 3.819-6.059, 7.052-9.052", "frequencyCaption": "explosion two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3677.wav", "onoffCaption": "woman laughing at 2.965-5.063", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3849.wav", "onoffCaption": "car horn honking at 0.921-5.828, 6.938-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3900.wav", "onoffCaption": "gunshot at 0.379-2.379, 3.498-5.498, 6.33-8.33", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3952.wav", "onoffCaption": "tapping clicking clanking at 0.053-3.493, 5.295-8.735 and door slamming at 3.514-4.687, 5.589-6.762, 7.398-8.571", "frequencyCaption": "tapping clicking clanking two times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_71.wav", "onoffCaption": "car horn honking at 0.054-2.054, 2.567-5.48, 6.163-8.547 and dog barking at 1.396-3.396", "frequencyCaption": "car horn honking three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_94.wav", "onoffCaption": "whistling at 0.664-2.673 and dog barking at 5.547-7.547", "frequencyCaption": "whistling one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_103.wav", "onoffCaption": "train horn at 3.383-6.917, 7.644-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_119.wav", "onoffCaption": "explosion at 1.17-6.17", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_224.wav", "onoffCaption": "explosion at 0.69-3.443, 4.949-7.678", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_271.wav", "onoffCaption": "sheep goat bleating at 0.252-2.252, 3.168-5.863, 7.483-9.483", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_294.wav", "onoffCaption": "dog barking at 4.009-6.009, 6.713-8.713", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_325.wav", "onoffCaption": "train horn at 2.454-4.934", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_370.wav", "onoffCaption": "woman laughing at 0.539-3.611", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_395.wav", "onoffCaption": "explosion at 0.154-5.154, 6.077-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_454.wav", "onoffCaption": "cat meowing at 0.659-1.744, 2.58-3.665 and car horn honking at 6.169-10.0", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_555.wav", "onoffCaption": "train horn at 2.939-6.139, 7.128-9.199", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_627.wav", "onoffCaption": "sneeze at 0.592-2.299, 3.698-5.783", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_726.wav", "onoffCaption": "sneeze at 1.533-3.061, 4.94-6.468", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_769.wav", "onoffCaption": "burping belching at 3.011-6.19, 7.094-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_851.wav", "onoffCaption": "duck quacking at 0.562-2.562, 4.942-6.942", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_950.wav", "onoffCaption": "sheep goat bleating at 0.278-2.278, 3.82-6.577", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1065.wav", "onoffCaption": "burping belching at 0.199-2.293, 3.025-5.119, 6.426-8.52", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_1080.wav", "onoffCaption": "woman laughing at 2.234-8.968", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1164.wav", "onoffCaption": "door knocking at 2.967-5.319, 6.283-8.635", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1181.wav", "onoffCaption": "cat meowing at 0.605-2.793, 3.637-5.597, 7.0-8.011", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1216.wav", "onoffCaption": "whistling at 0.333-7.988", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1243.wav", "onoffCaption": "cat meowing at 0.456-1.596 and sheep goat bleating at 0.866-2.866 and sneeze at 1.641-3.186, 5.174-7.413, 8.442-9.534", "frequencyCaption": "cat meowing one times and sheep goat bleating one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1259.wav", "onoffCaption": "explosion at 0.006-2.094, 3.253-5.341", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1317.wav", "onoffCaption": "door knocking at 0.43-3.539 and duck quacking at 2.429-4.429", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1342.wav", "onoffCaption": "train horn at 2.787-6.147, 7.196-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1358.wav", "onoffCaption": "burping belching at 1.915-3.95, 4.703-6.738", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1429.wav", "onoffCaption": "sneeze at 2.999-4.091, 5.749-7.988 and door slamming at 5.615-8.332", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1433.wav", "onoffCaption": "thump thud at 2.715-6.382, 7.011-9.511", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1528.wav", "onoffCaption": "car horn honking at 3.34-5.84, 6.745-9.245", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1532.wav", "onoffCaption": "door slamming at 0.251-1.089, 2.484-3.322, 5.322-6.16", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1615.wav", "onoffCaption": "car horn honking at 3.507-7.907", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1640.wav", "onoffCaption": "thump thud at 2.133-4.633", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1714.wav", "onoffCaption": "cow mooing at 2.052-7.032 and cat meowing at 3.252-6.493 and burping belching at 3.449-5.48, 6.955-9.955", "frequencyCaption": "cow mooing one times and cat meowing one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1741.wav", "onoffCaption": "cat meowing at 3.699-4.708, 7.085-8.094, 8.677-9.686", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1836.wav", "onoffCaption": "whistling at 1.123-3.998, 6.213-9.088", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1937.wav", "onoffCaption": "explosion at 0.473-5.473, 7.065-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3007.wav", "onoffCaption": "spraying at 3.438-4.379, 5.961-8.545", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3106.wav", "onoffCaption": "thump thud at 0.143-4.518 and burping belching at 7.45-9.485", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3274.wav", "onoffCaption": "sheep goat bleating at 2.531-4.531, 5.953-8.314", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3291.wav", "onoffCaption": "cat meowing at 2.873-4.768, 5.368-7.308, 8.311-9.866", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3375.wav", "onoffCaption": "gunshot at 1.097-3.097, 3.963-5.963, 7.832-9.832 and sneeze at 3.728-5.016", "frequencyCaption": "gunshot three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3390.wav", "onoffCaption": "train horn at 1.323-4.643", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3451.wav", "onoffCaption": "door knocking at 1.209-3.299 and spraying at 3.065-3.916, 5.992-6.896", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3550.wav", "onoffCaption": "car horn honking at 1.828-4.341", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3622.wav", "onoffCaption": "door knocking at 3.698-8.137", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3638.wav", "onoffCaption": "dog barking at 3.178-5.178, 7.122-9.122", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3688.wav", "onoffCaption": "explosion at 0.51-2.512, 3.826-6.694, 7.756-10.0 and dog barking at 1.669-3.669, 6.087-8.087 and tapping clicking clanking at 4.081-7.521", "frequencyCaption": "explosion three times and dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3692.wav", "onoffCaption": "cat meowing at 0.796-2.691, 3.33-5.225", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3723.wav", "onoffCaption": "door knocking at 0.865-4.633, 5.792-9.56", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3739.wav", "onoffCaption": "explosion at 1.129-4.969, 5.71-8.318 and sneeze at 5.171-7.488, 8.553-9.656", "frequencyCaption": "explosion two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3789.wav", "onoffCaption": "train horn at 1.777-5.977, 7.19-10.0 and whistling at 3.008-8.183", "frequencyCaption": "train horn two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3793.wav", "onoffCaption": "woman laughing at 0.114-2.397, 4.314-6.412", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3854.wav", "onoffCaption": "sheep goat bleating at 0.091-2.091", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3955.wav", "onoffCaption": "burping belching at 0.1-2.303 and sneeze at 0.938-2.226, 2.785-4.073, 5.144-6.432 and spraying at 2.713-3.235", "frequencyCaption": "burping belching one times and sneeze three times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_255.wav", "onoffCaption": "duck quacking at 2.246-4.246 and car horn honking at 6.104-8.104", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_401.wav", "onoffCaption": "explosion at 1.188-4.188, 6.166-8.234 and door slamming at 6.603-7.284", "frequencyCaption": "explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_495.wav", "onoffCaption": "tapping clicking clanking at 1.277-4.717 and spraying at 6.204-7.268, 9.239-9.814", "frequencyCaption": "tapping clicking clanking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_603.wav", "onoffCaption": "door slamming at 0.31-1.21", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_672.wav", "onoffCaption": "sheep goat bleating at 2.385-5.705, 7.194-9.194", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_697.wav", "onoffCaption": "car horn honking at 2.257-5.17, 6.998-9.911", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_718.wav", "onoffCaption": "explosion at 0.272-3.272, 3.864-6.864", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_905.wav", "onoffCaption": "dog barking at 0.109-2.109, 2.928-4.928, 6.585-8.585", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_974.wav", "onoffCaption": "cow mooing at 0.23-3.24, 3.796-6.806", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1131.wav", "onoffCaption": "thump thud at 3.383-7.301", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1228.wav", "onoffCaption": "spraying at 2.431-3.181, 4.888-5.829, 7.392-9.784", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1333.wav", "onoffCaption": "tapping clicking clanking at 0.156-3.596", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1499.wav", "onoffCaption": "door knocking at 0.679-2.9, 4.216-6.437, 7.091-9.312", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1567.wav", "onoffCaption": "spraying at 1.396-2.653", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1582.wav", "onoffCaption": "tapping clicking clanking at 0.881-4.321, 5.587-9.027 and cow mooing at 0.881-5.861", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1765.wav", "onoffCaption": "gunshot at 3.529-5.529, 7.826-9.826", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1780.wav", "onoffCaption": "whistling at 0.548-3.523", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1863.wav", "onoffCaption": "dog barking at 0.423-2.423 and whistling at 6.542-10.0", "frequencyCaption": "dog barking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1886.wav", "onoffCaption": "duck quacking at 3.169-5.169 and train horn at 7.932-10.0", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1909.wav", "onoffCaption": "gunshot at 3.317-5.317, 7.169-9.169", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1978.wav", "onoffCaption": "door knocking at 1.1-5.61 and gunshot at 3.2-5.2, 6.765-9.271", "frequencyCaption": "door knocking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3039.wav", "onoffCaption": "thump thud at 0.21-2.71, 3.351-5.851, 6.52-9.02", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_3048.wav", "onoffCaption": "tapping clicking clanking at 2.359-5.799, 7.591-9.639", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3122.wav", "onoffCaption": "door knocking at 0.12-3.24 and duck quacking at 0.249-2.249, 2.872-4.872 and cow mooing at 0.975-3.944", "frequencyCaption": "door knocking one times and duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3153.wav", "onoffCaption": "spraying at 2.568-3.3, 4.971-5.703", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3320.wav", "onoffCaption": "door slamming at 0.339-3.12, 5.228-6.603 and car horn honking at 1.221-6.128", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3351.wav", "onoffCaption": "whistling at 0.082-2.091, 2.634-4.643, 5.173-7.182", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_3505.wav", "onoffCaption": "gunshot at 0.828-2.828, 3.51-5.551, 6.26-8.26", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3574.wav", "onoffCaption": "dog barking at 0.732-2.732, 3.583-5.583, 6.4-8.4", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3707.wav", "onoffCaption": "thump thud at 2.729-5.776, 7.199-9.97", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3776.wav", "onoffCaption": "sneeze at 0.514-3.174, 3.942-6.345, 7.027-8.527 and duck quacking at 3.844-5.844, 6.531-8.531", "frequencyCaption": "sneeze three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3801.wav", "onoffCaption": "woman laughing at 0.433-3.72, 4.347-7.634", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3895.wav", "onoffCaption": "tapping clicking clanking at 0.957-4.397, 6.045-8.62", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_168.wav", "onoffCaption": "gunshot at 0.302-2.302, 3.007-5.007, 5.65-7.65 and sneeze at 1.729-4.804, 5.308-7.304", "frequencyCaption": "gunshot three times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_200.wav", "onoffCaption": "thump thud at 0.027-4.402, 5.833-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_425.wav", "onoffCaption": "whistling at 0.103-7.853", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_470.wav", "onoffCaption": "dog barking at 0.337-2.337 and cat meowing at 1.04-2.051, 2.899-4.443, 4.964-6.15 and gunshot at 2.741-4.741", "frequencyCaption": "dog barking one times and cat meowing three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_656.wav", "onoffCaption": "explosion at 2.865-7.865", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_820.wav", "onoffCaption": "burping belching at 2.113-7.714", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_921.wav", "onoffCaption": "tapping clicking clanking at 3.542-6.982", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_991.wav", "onoffCaption": "gunshot at 3.601-5.642, 7.694-9.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1115.wav", "onoffCaption": "tapping clicking clanking at 1.078-4.518 and cat meowing at 2.764-4.381, 6.623-8.24", "frequencyCaption": "tapping clicking clanking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1140.wav", "onoffCaption": "car horn honking at 0.279-3.192, 5.549-8.462", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1298.wav", "onoffCaption": "cat meowing at 1.184-2.459, 3.186-4.461", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1366.wav", "onoffCaption": "door slamming at 0.202-1.717, 2.995-4.51, 5.792-7.307 and spraying at 1.267-2.524 and woman laughing at 1.332-3.526, 5.309-7.503", "frequencyCaption": "door slamming three times and spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1383.wav", "onoffCaption": "cow mooing at 2.663-5.961", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1458.wav", "onoffCaption": "door knocking at 2.411-6.85", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1516.wav", "onoffCaption": "sneeze at 2.129-4.442, 6.085-8.893", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1543.wav", "onoffCaption": "car horn honking at 1.994-4.341, 6.759-9.106", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1730.wav", "onoffCaption": "door slamming at 0.66-1.913, 3.512-5.512 and spraying at 7.065-7.666", "frequencyCaption": "door slamming two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1812.wav", "onoffCaption": "sneeze at 0.01-2.134, 4.32-6.444", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1847.wav", "onoffCaption": "train horn at 2.659-6.419", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3076.wav", "onoffCaption": "cow mooing at 0.452-5.432, 7.721-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3089.wav", "onoffCaption": "explosion at 2.8-5.4, 7.474-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3177.wav", "onoffCaption": "thump thud at 1.086-4.133", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3192.wav", "onoffCaption": "woman laughing at 3.884-6.239, 6.838-9.193", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3304.wav", "onoffCaption": "train horn at 2.616-6.816, 7.46-9.94", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3420.wav", "onoffCaption": "door knocking at 0.306-3.153, 5.11-7.174 and car horn honking at 4.113-7.608", "frequencyCaption": "door knocking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3521.wav", "onoffCaption": "tapping clicking clanking at 1.082-4.522, 5.042-7.335", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3591.wav", "onoffCaption": "door knocking at 2.49-5.546 and car horn honking at 7.149-10.0", "frequencyCaption": "door knocking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3649.wav", "onoffCaption": "dog barking at 0.503-3.823, 4.345-7.665", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3752.wav", "onoffCaption": "sheep goat bleating at 1.849-3.849, 6.339-8.339 and duck quacking at 2.367-4.367, 5.96-7.96", "frequencyCaption": "sheep goat bleating two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3825.wav", "onoffCaption": "tapping clicking clanking at 0.088-3.528, 4.234-7.674 and sneeze at 0.459-2.455, 3.231-5.593, 6.22-7.394", "frequencyCaption": "tapping clicking clanking two times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3870.wav", "onoffCaption": "dog barking at 0.159-2.159 and burping belching at 3.641-10.0", "frequencyCaption": "dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_55.wav", "onoffCaption": "car horn honking at 0.516-3.302, 4.306-7.092", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_127.wav", "onoffCaption": "explosion at 0.637-2.701, 3.39-5.454, 6.295-8.359", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_172.wav", "onoffCaption": "woman laughing at 0.259-2.678, 3.762-6.181, 7.293-9.712", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_197.wav", "onoffCaption": "sheep goat bleating at 1.284-3.284, 3.836-5.836, 7.193-9.193", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_301.wav", "onoffCaption": "door knocking at 3.503-6.57", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_354.wav", "onoffCaption": "explosion at 2.583-6.136, 6.737-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_524.wav", "onoffCaption": "duck quacking at 0.427-2.427, 3.177-5.177, 6.113-8.113 and burping belching at 2.667-5.667, 6.53-9.53", "frequencyCaption": "duck quacking three times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_571.wav", "onoffCaption": "cow mooing at 0.128-3.138, 4.344-7.354 and dog barking at 4.593-6.593", "frequencyCaption": "cow mooing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_594.wav", "onoffCaption": "cat meowing at 2.648-3.66, 4.641-5.653", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_619.wav", "onoffCaption": "whistling at 0.822-7.241", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_702.wav", "onoffCaption": "duck quacking at 0.54-2.54", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_757.wav", "onoffCaption": "cow mooing at 0.723-3.705, 4.306-6.958 and train horn at 1.18-7.649", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_875.wav", "onoffCaption": "duck quacking at 1.336-3.336, 5.54-7.54 and car horn honking at 1.58-6.487, 7.594-9.594 and gunshot at 2.079-4.079, 5.251-7.251", "frequencyCaption": "duck quacking two times and car horn honking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_890.wav", "onoffCaption": "sneeze at 3.586-5.505, 6.712-8.631", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1014.wav", "onoffCaption": "train horn at 0.202-2.357 and spraying at 6.382-6.882", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1041.wav", "onoffCaption": "cat meowing at 0.344-1.353 and sheep goat bleating at 4.842-6.842, 7.989-9.989", "frequencyCaption": "cat meowing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1232.wav", "onoffCaption": "explosion at 0.237-5.237", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1267.wav", "onoffCaption": "car horn honking at 0.041-3.882, 5.425-7.938 and spraying at 2.028-2.895", "frequencyCaption": "car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1282.wav", "onoffCaption": "gunshot at 1.822-3.822 and cow mooing at 2.568-7.548", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1329.wav", "onoffCaption": "door knocking at 0.595-3.651, 4.593-6.72", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1399.wav", "onoffCaption": "sneeze at 0.117-3.765, 4.445-5.537 and explosion at 0.46-3.46 and car horn honking at 1.044-3.044", "frequencyCaption": "sneeze two times and explosion one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1417.wav", "onoffCaption": "sheep goat bleating at 1.462-3.462, 4.727-6.727 and tapping clicking clanking at 2.217-5.657, 7.32-10.0", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1442.wav", "onoffCaption": "tapping clicking clanking at 0.277-3.717, 5.707-9.147", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1559.wav", "onoffCaption": "thump thud at 0.397-4.064, 5.612-9.279", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1631.wav", "onoffCaption": "car horn honking at 0.189-2.702", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1664.wav", "onoffCaption": "train horn at 0.002-4.07", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1808.wav", "onoffCaption": "woman laughing at 2.822-5.876, 6.644-9.698", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1913.wav", "onoffCaption": "sneeze at 1.993-3.239, 4.458-6.571, 7.242-9.645", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1946.wav", "onoffCaption": "burping belching at 2.326-6.662 and door slamming at 4.842-5.82", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3023.wav", "onoffCaption": "door slamming at 0.184-2.312, 3.655-5.783, 6.912-9.04", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3093.wav", "onoffCaption": "woman laughing at 0.977-3.569", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3138.wav", "onoffCaption": "train horn at 2.088-7.165 and explosion at 2.61-5.61 and door slamming at 3.402-4.655", "frequencyCaption": "train horn one times and explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3188.wav", "onoffCaption": "sneeze at 0.033-1.321, 2.162-4.776, 5.79-7.402", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3205.wav", "onoffCaption": "thump thud at 1.166-4.213, 4.916-7.242 and gunshot at 6.167-8.167", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3250.wav", "onoffCaption": "duck quacking at 0.323-2.323, 4.398-6.398", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3475.wav", "onoffCaption": "woman laughing at 0.214-2.806, 3.565-5.773", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3490.wav", "onoffCaption": "spraying at 0.003-0.784, 2.076-4.204, 5.949-6.457", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3606.wav", "onoffCaption": "cow mooing at 0.694-3.676, 5.839-8.821", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3653.wav", "onoffCaption": "thump thud at 0.056-3.974, 5.313-9.231", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3748.wav", "onoffCaption": "door slamming at 0.261-1.74", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3924.wav", "onoffCaption": "woman laughing at 0.478-3.532, 5.351-8.05", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3971.wav", "onoffCaption": "sheep goat bleating at 1.131-3.131 and thump thud at 4.915-9.365 and dog barking at 6.582-8.582", "frequencyCaption": "sheep goat bleating one times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3994.wav", "onoffCaption": "burping belching at 1.116-3.739, 5.597-8.418", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_14.wav", "onoffCaption": "gunshot at 1.876-3.876, 5.278-7.278 and explosion at 2.328-5.057", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_133.wav", "onoffCaption": "cow mooing at 1.851-6.831", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_166.wav", "onoffCaption": "car horn honking at 1.533-3.88, 6.373-8.72 and tapping clicking clanking at 3.047-6.487", "frequencyCaption": "car horn honking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_183.wav", "onoffCaption": "spraying at 0.373-1.314, 2.792-3.733 and cat meowing at 2.599-3.611", "frequencyCaption": "spraying two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_315.wav", "onoffCaption": "door slamming at 0.256-1.489, 2.931-4.164", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_340.wav", "onoffCaption": "door slamming at 2.894-5.323 and cow mooing at 7.457-10.0", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_530.wav", "onoffCaption": "whistling at 0.753-9.138", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_565.wav", "onoffCaption": "cow mooing at 0.66-3.629 and woman laughing at 7.388-10.0", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_580.wav", "onoffCaption": "door slamming at 3.578-6.578, 7.636-9.832", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_658.wav", "onoffCaption": "explosion at 2.357-7.357", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_716.wav", "onoffCaption": "dog barking at 0.108-2.108, 2.668-4.668, 5.421-7.421 and gunshot at 1.186-3.186, 5.192-7.192 and door knocking at 4.129-6.193", "frequencyCaption": "dog barking three times and gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_743.wav", "onoffCaption": "cow mooing at 0.853-3.835, 5.646-8.205", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_834.wav", "onoffCaption": "duck quacking at 2.186-4.186, 5.978-7.978", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_861.wav", "onoffCaption": "thump thud at 2.521-6.439, 7.515-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1000.wav", "onoffCaption": "gunshot at 2.613-4.613, 6.416-8.416 and door knocking at 3.178-5.399, 6.317-8.686", "frequencyCaption": "gunshot two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1055.wav", "onoffCaption": "door slamming at 1.103-2.22, 4.078-5.195, 7.409-8.526", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1226.wav", "onoffCaption": "dog barking at 1.16-3.16", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1273.wav", "onoffCaption": "whistling at 0.636-2.865", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1296.wav", "onoffCaption": "whistling at 0.45-2.679", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1368.wav", "onoffCaption": "door knocking at 2.254-4.656, 6.714-8.841", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1456.wav", "onoffCaption": "tapping clicking clanking at 1.369-4.809", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1518.wav", "onoffCaption": "explosion at 0.098-5.098 and tapping clicking clanking at 0.832-4.272", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1625.wav", "onoffCaption": "cat meowing at 0.696-2.446, 3.021-4.771, 5.412-7.162", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1670.wav", "onoffCaption": "burping belching at 2.796-4.919, 5.517-7.64", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1681.wav", "onoffCaption": "cow mooing at 1.387-4.356, 6.831-9.813", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1695.wav", "onoffCaption": "gunshot at 0.724-2.724, 4.717-6.717 and duck quacking at 1.185-3.185, 4.117-6.117, 7.028-9.028", "frequencyCaption": "gunshot two times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1849.wav", "onoffCaption": "thump thud at 0.642-5.092, 5.855-7.909", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1907.wav", "onoffCaption": "explosion at 1.151-6.151", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1952.wav", "onoffCaption": "door knocking at 0.422-3.975 and duck quacking at 1.2-3.2 and car horn honking at 6.424-9.242", "frequencyCaption": "door knocking one times and duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3037.wav", "onoffCaption": "duck quacking at 0.241-2.241, 3.858-5.858 and thump thud at 4.667-6.895, 7.566-10.0", "frequencyCaption": "duck quacking two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3062.wav", "onoffCaption": "woman laughing at 2.849-5.049", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3087.wav", "onoffCaption": "duck quacking at 0.031-2.031, 3.164-5.164 and burping belching at 7.688-10.0", "frequencyCaption": "duck quacking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3179.wav", "onoffCaption": "gunshot at 1.377-3.377, 4.054-6.054", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3211.wav", "onoffCaption": "sheep goat bleating at 0.633-4.553, 5.305-7.305 and tapping clicking clanking at 1.74-5.18, 7.248-10.0", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3244.wav", "onoffCaption": "cow mooing at 2.773-6.071, 7.377-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3434.wav", "onoffCaption": "duck quacking at 3.054-5.054, 7.036-9.036", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3461.wav", "onoffCaption": "door slamming at 2.061-4.842, 6.857-9.638", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3484.wav", "onoffCaption": "duck quacking at 0.197-2.197, 2.792-4.792, 5.774-7.774 and thump thud at 3.437-6.484", "frequencyCaption": "duck quacking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3612.wav", "onoffCaption": "whistling at 0.85-5.334, 7.349-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3647.wav", "onoffCaption": "dog barking at 0.531-3.851, 5.617-7.617 and train horn at 0.823-3.303, 4.388-6.934 and tapping clicking clanking at 2.164-5.604", "frequencyCaption": "dog barking two times and train horn two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3709.wav", "onoffCaption": "train horn at 0.105-2.242, 2.805-4.942, 5.66-7.797", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_3930.wav", "onoffCaption": "door knocking at 0.868-4.421", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3965.wav", "onoffCaption": "woman laughing at 0.226-3.614, 4.453-7.841 and cow mooing at 3.081-6.05 and door slamming at 3.829-5.829", "frequencyCaption": "woman laughing two times and cow mooing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3980.wav", "onoffCaption": "door slamming at 0.164-3.138 and sheep goat bleating at 0.911-2.911, 5.018-7.018 and car horn honking at 2.298-7.205", "frequencyCaption": "door slamming one times and sheep goat bleating two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_41.wav", "onoffCaption": "tapping clicking clanking at 2.541-5.981", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_129.wav", "onoffCaption": "dog barking at 1.053-3.053, 4.927-6.927 and cat meowing at 4.681-9.041 and sneeze at 6.661-8.273", "frequencyCaption": "dog barking two times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_199.wav", "onoffCaption": "woman laughing at 0.853-7.587 and door knocking at 2.12-4.21 and door slamming at 6.068-7.185, 7.926-9.043", "frequencyCaption": "woman laughing one times and door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_214.wav", "onoffCaption": "sneeze at 2.226-3.838, 4.85-6.395", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_241.wav", "onoffCaption": "explosion at 0.978-5.978", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_431.wav", "onoffCaption": "spraying at 2.906-3.847, 4.661-5.602, 6.824-7.765", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_464.wav", "onoffCaption": "door knocking at 0.881-3.728, 4.455-7.302 and door slamming at 1.699-4.578, 5.851-6.656 and explosion at 2.481-6.481", "frequencyCaption": "door knocking two times and door slamming two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_481.wav", "onoffCaption": "gunshot at 2.466-4.466, 5.127-7.127 and door knocking at 3.505-8.207 and train horn at 3.535-6.935", "frequencyCaption": "gunshot two times and door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_617.wav", "onoffCaption": "spraying at 2.92-3.57, 5.893-6.543, 8.392-9.042", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_642.wav", "onoffCaption": "gunshot at 2.277-4.277, 5.537-7.537 and cat meowing at 6.26-7.26", "frequencyCaption": "gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_759.wav", "onoffCaption": "door slamming at 3.234-5.46 and gunshot at 5.625-7.625", "frequencyCaption": "door slamming one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_884.wav", "onoffCaption": "spraying at 0.199-0.699, 1.763-2.263, 2.935-3.435 and cat meowing at 0.762-2.333, 2.93-4.501", "frequencyCaption": "spraying three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_935.wav", "onoffCaption": "sheep goat bleating at 2.538-4.538, 6.466-8.466", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_960.wav", "onoffCaption": "sneeze at 0.1-3.21, 4.615-6.7", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_985.wav", "onoffCaption": "burping belching at 3.179-6.358, 7.834-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1101.wav", "onoffCaption": "sneeze at 0.201-2.44, 3.115-5.028, 5.81-7.895 and gunshot at 1.528-3.528", "frequencyCaption": "sneeze three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1154.wav", "onoffCaption": "spraying at 0.02-1.104 and whistling at 0.446-5.946", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1170.wav", "onoffCaption": "spraying at 2.656-4.418, 6.913-8.17", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1269.wav", "onoffCaption": "duck quacking at 0.627-2.627, 3.358-5.358", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1327.wav", "onoffCaption": "burping belching at 0.01-2.135, 3.537-5.662, 6.649-8.774 and door knocking at 1.426-3.586", "frequencyCaption": "burping belching three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1372.wav", "onoffCaption": "dog barking at 2.736-4.736, 6.873-8.873", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1397.wav", "onoffCaption": "cat meowing at 0.314-1.898, 2.63-4.214", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1403.wav", "onoffCaption": "explosion at 1.981-4.734, 6.313-9.313", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1419.wav", "onoffCaption": "train horn at 0.012-4.012 and spraying at 6.3-7.381", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1502.wav", "onoffCaption": "duck quacking at 0.082-2.082 and car horn honking at 0.559-4.054 and spraying at 5.939-8.523", "frequencyCaption": "duck quacking one times and car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1526.wav", "onoffCaption": "thump thud at 0.761-2.989, 4.754-6.982", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1557.wav", "onoffCaption": "train horn at 0.528-8.728", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1724.wav", "onoffCaption": "door slamming at 0.416-1.267, 2.1-2.951, 4.414-5.265", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1771.wav", "onoffCaption": "thump thud at 0.394-2.622, 3.385-5.613, 6.502-8.73", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1794.wav", "onoffCaption": "door knocking at 0.846-3.576, 5.727-8.351", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1806.wav", "onoffCaption": "explosion at 1.47-6.47 and door slamming at 1.716-4.497, 5.429-6.908 and burping belching at 3.429-7.429", "frequencyCaption": "explosion one times and door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1853.wav", "onoffCaption": "cow mooing at 0.434-3.416, 3.954-6.789, 7.523-9.86", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_1948.wav", "onoffCaption": "train horn at 1.774-6.851 and tapping clicking clanking at 3.442-6.882", "frequencyCaption": "train horn one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3078.wav", "onoffCaption": "sneeze at 0.026-1.483 and thump thud at 3.614-6.114, 7.183-10.0", "frequencyCaption": "sneeze one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3136.wav", "onoffCaption": "train horn at 0.866-5.768, 7.973-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3163.wav", "onoffCaption": "car horn honking at 0.131-2.478", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3186.wav", "onoffCaption": "spraying at 1.654-3.416, 4.687-7.123 and cat meowing at 2.384-5.625, 7.118-10.0", "frequencyCaption": "spraying two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3310.wav", "onoffCaption": "train horn at 1.864-5.398 and door knocking at 4.175-6.239, 6.937-9.001", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3345.wav", "onoffCaption": "door knocking at 3.904-8.176", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3535.wav", "onoffCaption": "gunshot at 1.061-3.061, 4.558-6.558 and duck quacking at 3.044-5.044", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3560.wav", "onoffCaption": "car horn honking at 0.198-4.71, 6.293-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3585.wav", "onoffCaption": "cat meowing at 2.877-4.017", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3608.wav", "onoffCaption": "cow mooing at 2.655-5.953 and sneeze at 7.997-10.0", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3713.wav", "onoffCaption": "whistling at 0.531-5.015, 7.0-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3746.wav", "onoffCaption": "car horn honking at 0.427-3.213", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3831.wav", "onoffCaption": "sheep goat bleating at 2.297-4.297, 6.101-8.101", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3864.wav", "onoffCaption": "sneeze at 0.203-1.703", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3881.wav", "onoffCaption": "sneeze at 0.434-2.117, 2.96-4.643, 5.251-6.934", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_158.wav", "onoffCaption": "sheep goat bleating at 0.22-2.22, 4.379-6.379, 7.198-9.198", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_230.wav", "onoffCaption": "thump thud at 0.17-2.509, 4.2-6.539", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_265.wav", "onoffCaption": "burping belching at 0.549-2.58, 3.304-5.335 and cow mooing at 1.658-4.64, 5.717-8.699", "frequencyCaption": "burping belching two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_280.wav", "onoffCaption": "sheep goat bleating at 3.352-5.352, 6.27-9.27", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_331.wav", "onoffCaption": "woman laughing at 0.703-2.986, 4.385-6.668 and burping belching at 3.892-6.892", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_415.wav", "onoffCaption": "gunshot at 0.476-2.476 and cat meowing at 0.509-2.503, 3.064-4.648, 5.727-7.962", "frequencyCaption": "gunshot one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_440.wav", "onoffCaption": "duck quacking at 0.184-2.184, 3.112-5.112", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_633.wav", "onoffCaption": "cat meowing at 0.287-1.434, 2.463-3.61", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_666.wav", "onoffCaption": "cow mooing at 2.642-5.94", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_683.wav", "onoffCaption": "door slamming at 2.21-3.601", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_699.wav", "onoffCaption": "door slamming at 2.043-4.269, 4.82-6.211, 8.048-9.572", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_728.wav", "onoffCaption": "train horn at 2.321-6.389, 7.291-9.903", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_767.wav", "onoffCaption": "duck quacking at 0.926-2.926, 4.515-6.515 and cat meowing at 1.965-3.925", "frequencyCaption": "duck quacking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_798.wav", "onoffCaption": "whistling at 0.489-6.899", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_911.wav", "onoffCaption": "cat meowing at 0.48-1.666 and whistling at 0.61-10.0", "frequencyCaption": "cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_944.wav", "onoffCaption": "burping belching at 1.37-5.393, 6.183-9.183", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1125.wav", "onoffCaption": "spraying at 2.52-3.371, 4.122-6.582, 7.329-7.956", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1195.wav", "onoffCaption": "door knocking at 0.96-3.2", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1218.wav", "onoffCaption": "tapping clicking clanking at 1.668-5.108", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1257.wav", "onoffCaption": "sneeze at 1.556-2.844, 3.917-6.002 and duck quacking at 2.435-4.435, 5.674-7.674 and spraying at 3.075-3.659, 5.829-6.413", "frequencyCaption": "sneeze two times and duck quacking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1303.wav", "onoffCaption": "car horn honking at 1.45-3.797 and spraying at 2.235-3.086 and dog barking at 6.931-9.852", "frequencyCaption": "car horn honking one times and spraying one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1356.wav", "onoffCaption": "door knocking at 3.378-6.931", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1468.wav", "onoffCaption": "cat meowing at 0.276-1.378, 2.979-4.245 and gunshot at 7.143-9.143", "frequencyCaption": "cat meowing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1573.wav", "onoffCaption": "dog barking at 0.216-2.216, 3.355-5.355, 6.148-8.148 and spraying at 0.276-0.845 and tapping clicking clanking at 5.304-8.744", "frequencyCaption": "dog barking three times and spraying one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1596.wav", "onoffCaption": "cat meowing at 2.368-3.554, 4.623-7.014", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1601.wav", "onoffCaption": "door knocking at 0.127-2.149 and cow mooing at 3.366-6.376", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1700.wav", "onoffCaption": "whistling at 1.146-8.764", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1755.wav", "onoffCaption": "sneeze at 0.305-1.892, 4.262-6.375", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1822.wav", "onoffCaption": "train horn at 1.537-4.417, 6.155-9.035", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1877.wav", "onoffCaption": "gunshot at 2.109-4.109 and whistling at 7.924-10.0", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1892.wav", "onoffCaption": "explosion at 0.321-3.377", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1939.wav", "onoffCaption": "door knocking at 0.279-3.126, 3.663-6.51", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1989.wav", "onoffCaption": "sheep goat bleating at 0.042-2.042 and woman laughing at 4.477-7.069", "frequencyCaption": "sheep goat bleating one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3009.wav", "onoffCaption": "sheep goat bleating at 0.509-2.509", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3112.wav", "onoffCaption": "woman laughing at 3.607-6.212 and duck quacking at 5.845-7.845", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3147.wav", "onoffCaption": "whistling at 2.305-6.789, 7.817-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3334.wav", "onoffCaption": "spraying at 0.947-1.697, 3.096-3.846, 4.38-5.13 and whistling at 7.494-10.0", "frequencyCaption": "spraying three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3361.wav", "onoffCaption": "dog barking at 0.659-2.659, 4.075-6.075 and cow mooing at 2.011-6.991, 7.547-10.0", "frequencyCaption": "dog barking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3384.wav", "onoffCaption": "gunshot at 0.991-2.991 and spraying at 3.058-4.753", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3511.wav", "onoffCaption": "cat meowing at 0.223-1.232, 1.827-2.836 and whistling at 0.758-8.413", "frequencyCaption": "cat meowing two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3544.wav", "onoffCaption": "tapping clicking clanking at 1.273-4.713, 5.387-8.827 and car horn honking at 1.592-3.939 and sheep goat bleating at 1.733-3.733", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3679.wav", "onoffCaption": "spraying at 0.11-0.737, 1.279-1.787, 3.007-5.467 and door knocking at 7.869-10.0", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3686.wav", "onoffCaption": "explosion at 1.044-6.044, 6.567-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3737.wav", "onoffCaption": "sheep goat bleating at 0.425-2.425, 4.108-6.108 and door slamming at 7.848-8.748", "frequencyCaption": "sheep goat bleating two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3762.wav", "onoffCaption": "duck quacking at 2.729-4.729", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3787.wav", "onoffCaption": "burping belching at 2.836-6.095, 7.143-9.467", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3815.wav", "onoffCaption": "cow mooing at 2.141-5.151 and explosion at 2.372-7.372", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3840.wav", "onoffCaption": "duck quacking at 1.918-3.918, 5.313-7.313", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_30.wav", "onoffCaption": "duck quacking at 1.581-3.581, 5.379-7.379", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_65.wav", "onoffCaption": "gunshot at 0.859-2.859", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_80.wav", "onoffCaption": "sneeze at 0.482-2.478, 3.427-5.423, 7.453-9.449 and explosion at 0.882-3.754, 4.584-6.904", "frequencyCaption": "sneeze three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_117.wav", "onoffCaption": "explosion at 1.659-4.715, 6.7-9.756", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_142.wav", "onoffCaption": "spraying at 0.225-0.733, 1.499-2.007 and dog barking at 0.699-4.019, 4.823-8.143 and door knocking at 0.984-4.734, 5.73-7.951", "frequencyCaption": "spraying two times and dog barking two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_364.wav", "onoffCaption": "car horn honking at 0.766-3.692, 5.467-8.228", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_381.wav", "onoffCaption": "thump thud at 0.034-3.952, 6.217-8.717 and gunshot at 1.044-3.085", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_514.wav", "onoffCaption": "cow mooing at 0.534-3.516 and door slamming at 3.555-4.574, 6.074-7.093", "frequencyCaption": "cow mooing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_541.wav", "onoffCaption": "car horn honking at 2.908-6.749", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_629.wav", "onoffCaption": "burping belching at 1.02-6.144, 6.845-9.4", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_732.wav", "onoffCaption": "explosion at 0.038-2.126 and thump thud at 0.211-2.982, 4.408-6.482", "frequencyCaption": "explosion one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_782.wav", "onoffCaption": "thump thud at 1.036-3.375, 4.184-6.955 and cat meowing at 2.618-5.859, 7.36-10.0", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_810.wav", "onoffCaption": "dog barking at 0.258-2.258, 4.132-6.132, 7.99-9.99", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_845.wav", "onoffCaption": "thump thud at 2.152-6.07", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1024.wav", "onoffCaption": "woman laughing at 0.196-2.277, 4.543-6.624", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1071.wav", "onoffCaption": "door slamming at 0.205-2.205, 3.528-5.528, 6.22-7.473 and duck quacking at 2.613-4.613", "frequencyCaption": "door slamming three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1094.wav", "onoffCaption": "sheep goat bleating at 2.335-4.335, 5.743-7.743", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1202.wav", "onoffCaption": "cat meowing at 0.067-1.076", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1319.wav", "onoffCaption": "train horn at 0.993-3.873, 4.733-7.613", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1472.wav", "onoffCaption": "thump thud at 0.435-4.353 and sheep goat bleating at 1.679-3.679, 5.085-7.085", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1497.wav", "onoffCaption": "door slamming at 0.326-2.326, 4.646-6.646", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1569.wav", "onoffCaption": "cow mooing at 0.08-3.09, 4.676-7.686 and door slamming at 7.197-9.197", "frequencyCaption": "cow mooing two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1654.wav", "onoffCaption": "tapping clicking clanking at 3.172-6.612", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1838.wav", "onoffCaption": "sheep goat bleating at 1.288-4.608, 6.25-9.57 and duck quacking at 2.365-4.365", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1888.wav", "onoffCaption": "tapping clicking clanking at 0.995-4.435 and cow mooing at 2.244-5.226, 6.414-9.383", "frequencyCaption": "tapping clicking clanking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1923.wav", "onoffCaption": "gunshot at 1.011-3.011, 4.239-6.239, 7.057-9.057", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1976.wav", "onoffCaption": "cat meowing at 2.801-4.795, 5.356-7.544, 8.821-9.961", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1993.wav", "onoffCaption": "woman laughing at 2.007-4.089, 6.48-8.562", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3013.wav", "onoffCaption": "sheep goat bleating at 2.166-4.166 and cow mooing at 4.473-7.442 and gunshot at 6.86-8.86", "frequencyCaption": "sheep goat bleating one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3046.wav", "onoffCaption": "sheep goat bleating at 0.979-2.979 and door slamming at 5.08-6.604, 7.268-9.697", "frequencyCaption": "sheep goat bleating one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3108.wav", "onoffCaption": "spraying at 0.384-2.512, 4.662-7.003 and sneeze at 0.575-2.821", "frequencyCaption": "spraying two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3235.wav", "onoffCaption": "cat meowing at 0.814-1.825", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3260.wav", "onoffCaption": "cat meowing at 0.62-2.56, 3.135-5.075, 6.248-8.188", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3285.wav", "onoffCaption": "gunshot at 0.533-2.533", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3410.wav", "onoffCaption": "duck quacking at 0.698-2.698, 4.798-6.798", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3445.wav", "onoffCaption": "cow mooing at 1.175-4.185, 4.703-7.672 and door knocking at 6.169-8.549", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3636.wav", "onoffCaption": "train horn at 1.218-3.873, 5.846-8.552", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3663.wav", "onoffCaption": "explosion at 2.042-4.049", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3778.wav", "onoffCaption": "train horn at 2.917-6.237, 7.741-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3914.wav", "onoffCaption": "thump thud at 0.368-2.868, 3.643-6.143, 7.238-9.738", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_3941.wav", "onoffCaption": "car horn honking at 0.289-5.196, 7.372-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_37.wav", "onoffCaption": "burping belching at 0.202-5.202, 6.756-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_62.wav", "onoffCaption": "door slamming at 0.077-0.758, 1.309-3.309 and car horn honking at 7.208-10.0", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_87.wav", "onoffCaption": "cat meowing at 3.989-5.929, 7.639-9.389", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_110.wav", "onoffCaption": "woman laughing at 1.572-3.654, 4.928-7.01", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_145.wav", "onoffCaption": "cat meowing at 2.595-3.808", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_336.wav", "onoffCaption": "spraying at 2.725-3.725", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_408.wav", "onoffCaption": "spraying at 0.198-0.979, 2.793-3.42, 4.839-6.572", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_513.wav", "onoffCaption": "thump thud at 0.607-4.525, 5.536-9.454 and woman laughing at 3.351-7.403", "frequencyCaption": "thump thud two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_546.wav", "onoffCaption": "burping belching at 0.499-2.597", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_760.wav", "onoffCaption": "car horn honking at 2.233-6.555, 7.446-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_785.wav", "onoffCaption": "sheep goat bleating at 0.214-2.214 and tapping clicking clanking at 4.133-7.573", "frequencyCaption": "sheep goat bleating one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_817.wav", "onoffCaption": "burping belching at 0.67-2.768, 5.205-7.33", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_842.wav", "onoffCaption": "explosion at 3.225-6.225", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_959.wav", "onoffCaption": "door knocking at 0.278-2.581, 4.559-6.961, 7.965-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1023.wav", "onoffCaption": "spraying at 0.154-1.005, 3.277-4.972", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1076.wav", "onoffCaption": "woman laughing at 0.237-7.249, 7.811-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1093.wav", "onoffCaption": "spraying at 0.32-1.052, 3.287-5.415 and burping belching at 6.898-10.0", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1138.wav", "onoffCaption": "car horn honking at 0.532-2.532, 3.585-6.576, 7.794-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1188.wav", "onoffCaption": "cow mooing at 3.604-8.033", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1250.wav", "onoffCaption": "sheep goat bleating at 3.127-5.127, 5.942-7.942", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1420.wav", "onoffCaption": "door slamming at 3.092-4.07, 5.862-6.713 and thump thud at 5.014-7.514", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1427.wav", "onoffCaption": "tapping clicking clanking at 3.687-7.127", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1475.wav", "onoffCaption": "dog barking at 2.524-4.524 and duck quacking at 2.891-4.891, 5.41-7.41", "frequencyCaption": "dog barking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1490.wav", "onoffCaption": "woman laughing at 0.474-3.269", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1606.wav", "onoffCaption": "train horn at 3.052-6.372", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1924.wav", "onoffCaption": "woman laughing at 0.787-3.353, 4.339-6.905 and door knocking at 2.189-5.957", "frequencyCaption": "woman laughing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1971.wav", "onoffCaption": "duck quacking at 0.795-2.795, 4.693-6.693", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1994.wav", "onoffCaption": "burping belching at 0.079-2.11", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3014.wav", "onoffCaption": "duck quacking at 2.193-4.193, 4.922-6.922, 7.799-9.799", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3041.wav", "onoffCaption": "door slamming at 1.161-2.926 and whistling at 7.103-10.0", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3232.wav", "onoffCaption": "cat meowing at 2.11-4.05, 5.92-7.86", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3267.wav", "onoffCaption": "thump thud at 0.009-4.459, 6.04-8.228", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3329.wav", "onoffCaption": "thump thud at 0.967-4.014, 6.031-9.078", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3417.wav", "onoffCaption": "spraying at 0.093-1.034", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3442.wav", "onoffCaption": "gunshot at 0.47-2.47, 4.029-6.029 and cow mooing at 4.86-7.842", "frequencyCaption": "gunshot two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3559.wav", "onoffCaption": "burping belching at 0.082-2.176, 4.473-7.294 and woman laughing at 0.393-2.676", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3631.wav", "onoffCaption": "sneeze at 3.473-4.761, 6.665-9.279", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3664.wav", "onoffCaption": "tapping clicking clanking at 2.176-5.616, 6.881-9.797", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3681.wav", "onoffCaption": "tapping clicking clanking at 0.021-3.461", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3808.wav", "onoffCaption": "car horn honking at 0.792-3.139 and woman laughing at 1.428-3.783, 4.399-7.212", "frequencyCaption": "car horn honking one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3913.wav", "onoffCaption": "train horn at 0.685-4.045", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3946.wav", "onoffCaption": "car horn honking at 0.998-3.911 and thump thud at 2.478-6.396", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_78.wav", "onoffCaption": "door slamming at 2.514-3.494, 4.288-5.268, 6.573-7.553 and thump thud at 3.927-6.698", "frequencyCaption": "door slamming three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_237.wav", "onoffCaption": "dog barking at 1.736-3.736, 5.817-7.817 and door slamming at 3.506-4.006, 4.74-5.24", "frequencyCaption": "dog barking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_262.wav", "onoffCaption": "explosion at 0.161-2.879", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_278.wav", "onoffCaption": "whistling at 0.888-3.863, 5.403-8.378 and spraying at 1.588-3.98", "frequencyCaption": "whistling two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_287.wav", "onoffCaption": "train horn at 2.767-6.887", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_363.wav", "onoffCaption": "car horn honking at 0.079-2.897, 4.356-7.174 and thump thud at 5.249-7.588", "frequencyCaption": "car horn honking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_379.wav", "onoffCaption": "cat meowing at 0.541-3.445 and sheep goat bleating at 3.471-5.471, 7.835-9.835 and burping belching at 5.151-8.651", "frequencyCaption": "cat meowing one times and sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_386.wav", "onoffCaption": "tapping clicking clanking at 3.089-6.529", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_447.wav", "onoffCaption": "door knocking at 1.369-7.429 and cat meowing at 1.777-3.725, 5.36-7.308", "frequencyCaption": "door knocking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_634.wav", "onoffCaption": "sheep goat bleating at 3.048-5.048, 6.235-8.355 and woman laughing at 4.232-6.348, 7.658-9.852", "frequencyCaption": "sheep goat bleating two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_661.wav", "onoffCaption": "car horn honking at 0.639-5.039, 6.08-8.118", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_684.wav", "onoffCaption": "whistling at 1.082-8.737", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_735.wav", "onoffCaption": "thump thud at 2.444-4.672, 5.658-7.997 and gunshot at 2.509-4.528", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_858.wav", "onoffCaption": "woman laughing at 1.337-3.62 and spraying at 6.255-7.339", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_916.wav", "onoffCaption": "duck quacking at 0.154-2.154, 3.43-5.43", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_943.wav", "onoffCaption": "thump thud at 0.617-4.992", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1089.wav", "onoffCaption": "duck quacking at 0.838-2.838, 4.658-6.658 and sheep goat bleating at 3.276-5.276", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1177.wav", "onoffCaption": "door slamming at 0.485-2.398, 4.742-7.703", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1192.wav", "onoffCaption": "burping belching at 3.097-7.097", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1205.wav", "onoffCaption": "door slamming at 1.534-2.434, 3.424-4.324 and burping belching at 6.945-9.945", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1304.wav", "onoffCaption": "sneeze at 0.136-2.382, 3.132-5.378, 6.294-8.54 and cat meowing at 3.431-5.391, 6.289-8.237", "frequencyCaption": "sneeze three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1351.wav", "onoffCaption": "door slamming at 0.002-1.002, 2.305-4.305", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1521.wav", "onoffCaption": "sheep goat bleating at 0.099-2.099 and whistling at 0.517-8.267", "frequencyCaption": "sheep goat bleating one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1591.wav", "onoffCaption": "cow mooing at 0.046-3.015, 4.496-7.465, 7.986-10.0 and door slamming at 0.486-1.719", "frequencyCaption": "cow mooing three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1649.wav", "onoffCaption": "sheep goat bleating at 1.382-3.382 and duck quacking at 6.325-8.325", "frequencyCaption": "sheep goat bleating one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1653.wav", "onoffCaption": "duck quacking at 2.236-4.236", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1707.wav", "onoffCaption": "train horn at 0.201-3.001 and car horn honking at 0.44-2.44", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1748.wav", "onoffCaption": "car horn honking at 1.705-5.292 and cow mooing at 2.18-7.16", "frequencyCaption": "car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1752.wav", "onoffCaption": "burping belching at 1.373-5.242", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1825.wav", "onoffCaption": "door knocking at 2.788-5.037, 6.324-8.552", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1870.wav", "onoffCaption": "duck quacking at 2.886-4.886, 7.079-9.079", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3115.wav", "onoffCaption": "train horn at 1.259-4.499, 5.745-8.4", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3140.wav", "onoffCaption": "whistling at 0.617-2.626, 3.818-6.71 and door knocking at 2.508-5.132, 6.443-9.067", "frequencyCaption": "whistling two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3228.wav", "onoffCaption": "duck quacking at 1.877-3.877", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3282.wav", "onoffCaption": "cat meowing at 0.19-1.745, 3.887-5.442 and woman laughing at 7.712-9.81", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3298.wav", "onoffCaption": "door slamming at 3.537-6.511", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3333.wav", "onoffCaption": "door slamming at 0.342-1.515, 2.303-3.452, 4.241-6.437", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3366.wav", "onoffCaption": "car horn honking at 3.033-7.282", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3383.wav", "onoffCaption": "dog barking at 0.353-5.99 and door slamming at 5.231-7.594", "frequencyCaption": "dog barking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3399.wav", "onoffCaption": "woman laughing at 1.997-5.284 and car horn honking at 4.179-8.02", "frequencyCaption": "woman laughing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3458.wav", "onoffCaption": "gunshot at 2.555-4.555, 6.02-8.02", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3516.wav", "onoffCaption": "thump thud at 1.147-3.918 and sneeze at 5.679-9.327", "frequencyCaption": "thump thud one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3543.wav", "onoffCaption": "car horn honking at 1.374-5.028, 5.948-8.323", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3730.wav", "onoffCaption": "duck quacking at 2.802-4.802, 5.763-7.763", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3765.wav", "onoffCaption": "cow mooing at 2.874-5.843 and cat meowing at 6.157-7.184", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3780.wav", "onoffCaption": "sheep goat bleating at 1.923-5.563, 7.028-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3847.wav", "onoffCaption": "train horn at 2.831-7.733 and car horn honking at 3.775-8.024", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_213.wav", "onoffCaption": "sheep goat bleating at 4.077-6.077 and whistling at 4.832-7.807 and woman laughing at 5.713-8.132", "frequencyCaption": "sheep goat bleating one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_246.wav", "onoffCaption": "gunshot at 1.206-3.206", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_308.wav", "onoffCaption": "thump thud at 2.127-4.898, 7.007-9.507", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_412.wav", "onoffCaption": "cat meowing at 0.117-1.129 and woman laughing at 2.319-4.435", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_463.wav", "onoffCaption": "sneeze at 1.924-3.027, 3.851-4.954 and explosion at 7.605-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_486.wav", "onoffCaption": "tapping clicking clanking at 2.505-5.945, 7.953-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_509.wav", "onoffCaption": "thump thud at 1.212-4.879, 5.485-8.191", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_578.wav", "onoffCaption": "explosion at 1.569-4.569 and door slamming at 6.423-8.852", "frequencyCaption": "explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_610.wav", "onoffCaption": "tapping clicking clanking at 1.008-4.448, 6.821-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_645.wav", "onoffCaption": "gunshot at 2.581-4.581, 5.606-7.606", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_899.wav", "onoffCaption": "door slamming at 1.574-2.965, 4.028-5.419, 6.017-7.408", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_967.wav", "onoffCaption": "explosion at 2.942-7.863", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_982.wav", "onoffCaption": "tapping clicking clanking at 3.045-6.485", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1039.wav", "onoffCaption": "gunshot at 1.37-3.37, 4.286-6.286, 7.048-9.048", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1048.wav", "onoffCaption": "woman laughing at 0.025-3.11", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1122.wav", "onoffCaption": "thump thud at 3.834-8.284", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1153.wav", "onoffCaption": "sneeze at 0.825-2.437, 4.345-5.957 and dog barking at 2.113-4.513", "frequencyCaption": "sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1320.wav", "onoffCaption": "cow mooing at 0.004-3.302", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1375.wav", "onoffCaption": "explosion at 3.847-6.375, 7.628-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1390.wav", "onoffCaption": "sheep goat bleating at 0.762-2.762, 4.683-6.683", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1505.wav", "onoffCaption": "tapping clicking clanking at 0.221-3.661, 5.651-8.267", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1574.wav", "onoffCaption": "cow mooing at 0.572-3.582, 5.412-8.311", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1638.wav", "onoffCaption": "cow mooing at 1.479-4.777, 5.348-8.33 and duck quacking at 4.231-6.231, 7.487-9.487", "frequencyCaption": "cow mooing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1688.wav", "onoffCaption": "cat meowing at 0.14-1.184 and burping belching at 4.809-6.916", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1723.wav", "onoffCaption": "cow mooing at 0.814-3.824, 5.162-8.172", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1776.wav", "onoffCaption": "train horn at 2.256-6.324 and tapping clicking clanking at 2.356-5.796, 6.811-9.098", "frequencyCaption": "train horn one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1793.wav", "onoffCaption": "sheep goat bleating at 0.474-3.77, 4.683-7.979", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1801.wav", "onoffCaption": "thump thud at 1.867-4.329, 5.915-8.377", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1895.wav", "onoffCaption": "thump thud at 3.141-6.188", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3131.wav", "onoffCaption": "sheep goat bleating at 0.574-2.574, 5.046-7.046 and burping belching at 0.963-4.523", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3259.wav", "onoffCaption": "cat meowing at 1.523-4.553", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3317.wav", "onoffCaption": "tapping clicking clanking at 2.77-6.21", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3342.wav", "onoffCaption": "thump thud at 0.556-3.603, 4.335-7.382", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3499.wav", "onoffCaption": "thump thud at 0.149-3.816", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3567.wav", "onoffCaption": "woman laughing at 1.141-3.246", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3582.wav", "onoffCaption": "train horn at 2.172-4.639, 6.039-8.506 and thump thud at 7.093-9.555", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3714.wav", "onoffCaption": "spraying at 1.752-2.356, 3.004-3.945, 4.761-7.197", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3741.wav", "onoffCaption": "thump thud at 2.691-6.609", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3812.wav", "onoffCaption": "woman laughing at 1.805-4.005, 5.087-7.287", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3863.wav", "onoffCaption": "sneeze at 0.002-1.328, 2.29-5.202 and gunshot at 7.546-9.546", "frequencyCaption": "sneeze two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3886.wav", "onoffCaption": "whistling at 0.26-6.542, 7.991-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3909.wav", "onoffCaption": "woman laughing at 1.825-3.923, 4.569-6.667, 7.773-9.871", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3978.wav", "onoffCaption": "cat meowing at 0.014-1.598, 3.728-5.312", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_13.wav", "onoffCaption": "train horn at 1.267-6.125", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_161.wav", "onoffCaption": "duck quacking at 2.251-4.251, 5.882-7.882", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_184.wav", "onoffCaption": "door slamming at 1.129-4.09, 5.351-8.312", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_209.wav", "onoffCaption": "cow mooing at 0.235-3.217, 4.196-7.178 and duck quacking at 0.314-2.314, 3.528-5.528, 6.184-8.184", "frequencyCaption": "cow mooing two times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_312.wav", "onoffCaption": "door slamming at 2.09-4.548, 5.452-7.91", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_436.wav", "onoffCaption": "sneeze at 0.157-1.84, 2.888-4.801, 5.695-6.859 and explosion at 1.858-6.452", "frequencyCaption": "sneeze three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_537.wav", "onoffCaption": "thump thud at 0.307-4.682, 5.581-9.956", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_744.wav", "onoffCaption": "car horn honking at 0.921-3.834, 6.306-9.219", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_829.wav", "onoffCaption": "thump thud at 1.369-5.036, 5.978-8.234", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_833.wav", "onoffCaption": "thump thud at 0.767-3.106, 4.114-6.453, 7.09-9.429", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_866.wav", "onoffCaption": "sheep goat bleating at 0.53-4.45", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_928.wav", "onoffCaption": "tapping clicking clanking at 1.238-4.678, 6.197-8.498", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_932.wav", "onoffCaption": "woman laughing at 1.002-4.074, 6.289-9.361", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1007.wav", "onoffCaption": "door slamming at 1.093-3.093, 4.313-6.313", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1106.wav", "onoffCaption": "car horn honking at 3.019-6.194, 7.499-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1274.wav", "onoffCaption": "duck quacking at 1.809-3.809 and whistling at 5.361-10.0", "frequencyCaption": "duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1291.wav", "onoffCaption": "train horn at 0.326-3.726, 4.95-7.162", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1451.wav", "onoffCaption": "thump thud at 0.034-3.952, 4.674-7.174 and spraying at 3.911-5.644, 6.476-8.912", "frequencyCaption": "thump thud two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1550.wav", "onoffCaption": "cow mooing at 2.579-5.589, 7.075-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1622.wav", "onoffCaption": "door slamming at 0.324-2.782, 3.662-6.12, 7.194-9.652", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1739.wav", "onoffCaption": "train horn at 0.383-10.0 and sneeze at 3.175-4.421, 5.109-6.355, 7.415-8.661", "frequencyCaption": "train horn one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1854.wav", "onoffCaption": "spraying at 0.11-0.685, 1.718-2.622, 3.534-4.266", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1955.wav", "onoffCaption": "spraying at 2.243-2.818 and explosion at 2.78-5.78", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3030.wav", "onoffCaption": "door knocking at 0.001-2.848, 4.27-7.117", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3065.wav", "onoffCaption": "car horn honking at 2.783-6.624, 7.492-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3080.wav", "onoffCaption": "sheep goat bleating at 0.185-3.825", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3164.wav", "onoffCaption": "door slamming at 3.034-5.492, 6.137-7.678", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3181.wav", "onoffCaption": "thump thud at 0.103-4.553 and cat meowing at 1.125-3.102", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3216.wav", "onoffCaption": "whistling at 3.804-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3429.wav", "onoffCaption": "door knocking at 0.197-3.317, 5.04-7.55", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3433.wav", "onoffCaption": "sheep goat bleating at 1.432-3.432, 5.775-7.775", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3466.wav", "onoffCaption": "tapping clicking clanking at 0.407-3.847, 5.443-8.883", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3528.wav", "onoffCaption": "tapping clicking clanking at 3.215-6.655, 7.623-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3532.wav", "onoffCaption": "gunshot at 1.231-3.231, 4.12-6.12 and explosion at 7.291-10.0", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3640.wav", "onoffCaption": "sneeze at 0.411-2.524 and dog barking at 0.973-2.973 and spraying at 5.929-6.556, 7.556-8.183, 9.228-9.855", "frequencyCaption": "sneeze one times and dog barking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3836.wav", "onoffCaption": "cat meowing at 0.111-1.666, 2.677-4.232, 6.241-7.796", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3937.wav", "onoffCaption": "car horn honking at 0.116-3.029, 3.973-6.427, 7.604-9.604", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_46.wav", "onoffCaption": "duck quacking at 1.528-3.528", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_134.wav", "onoffCaption": "sheep goat bleating at 3.009-7.729", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_222.wav", "onoffCaption": "spraying at 0.061-0.569, 1.14-2.081", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_339.wav", "onoffCaption": "car horn honking at 1.248-3.248, 3.785-5.785 and spraying at 1.875-2.939 and gunshot at 3.353-5.353", "frequencyCaption": "car horn honking two times and spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_347.wav", "onoffCaption": "door knocking at 2.775-5.895, 6.592-9.712", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_407.wav", "onoffCaption": "duck quacking at 1.025-3.025, 4.357-6.357, 7.309-9.309", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_479.wav", "onoffCaption": "car horn honking at 2.446-6.033", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_562.wav", "onoffCaption": "door knocking at 2.335-4.462", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_587.wav", "onoffCaption": "duck quacking at 2.07-4.07, 5.856-7.856", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_674.wav", "onoffCaption": "burping belching at 0.475-2.573, 3.598-6.116, 6.89-9.015", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_691.wav", "onoffCaption": "explosion at 0.294-4.888, 6.62-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_711.wav", "onoffCaption": "woman laughing at 2.479-4.716, 6.449-8.729 and car horn honking at 3.55-6.063", "frequencyCaption": "woman laughing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_883.wav", "onoffCaption": "woman laughing at 1.491-3.975, 4.989-7.473", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_998.wav", "onoffCaption": "tapping clicking clanking at 0.324-3.764 and car horn honking at 6.065-8.851", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1052.wav", "onoffCaption": "woman laughing at 1.457-8.902", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1149.wav", "onoffCaption": "car horn honking at 0.002-4.909, 5.614-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1221.wav", "onoffCaption": "sneeze at 0.416-4.916 and cat meowing at 6.14-8.328", "frequencyCaption": "sneeze one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1344.wav", "onoffCaption": "cow mooing at 0.37-3.352 and train horn at 6.332-9.866", "frequencyCaption": "cow mooing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1404.wav", "onoffCaption": "cow mooing at 0.174-3.156, 3.958-6.94, 7.572-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_1609.wav", "onoffCaption": "tapping clicking clanking at 3.452-6.892", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1677.wav", "onoffCaption": "tapping clicking clanking at 0.138-3.578 and gunshot at 1.678-3.678, 4.883-7.389", "frequencyCaption": "tapping clicking clanking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1692.wav", "onoffCaption": "sneeze at 2.938-4.169 and thump thud at 6.382-9.153", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1712.wav", "onoffCaption": "explosion at 4.023-7.023", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1789.wav", "onoffCaption": "cat meowing at 2.803-4.387, 5.33-6.516 and cow mooing at 4.765-7.775", "frequencyCaption": "cat meowing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1880.wav", "onoffCaption": "spraying at 1.107-3.543", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1900.wav", "onoffCaption": "cat meowing at 0.685-2.256 and duck quacking at 5.166-7.166", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3243.wav", "onoffCaption": "dog barking at 2.647-4.647", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3326.wav", "onoffCaption": "door knocking at 1.874-6.17 and dog barking at 2.187-4.187", "frequencyCaption": "door knocking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3358.wav", "onoffCaption": "thump thud at 0.013-2.241", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3483.wav", "onoffCaption": "sneeze at 1.519-3.98, 5.122-7.583 and door slamming at 5.232-7.595, 8.514-9.747", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3598.wav", "onoffCaption": "train horn at 3.652-7.652", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3615.wav", "onoffCaption": "cat meowing at 2.269-3.84, 5.6-7.171, 7.761-9.332", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3770.wav", "onoffCaption": "spraying at 2.216-3.463, 4.491-7.075", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3795.wav", "onoffCaption": "whistling at 0.521-9.132", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3807.wav", "onoffCaption": "door knocking at 0.858-2.948 and gunshot at 1.961-4.131", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3879.wav", "onoffCaption": "duck quacking at 0.504-2.504, 3.187-5.187", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3962.wav", "onoffCaption": "door slamming at 0.085-0.936 and dog barking at 0.218-2.218", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3987.wav", "onoffCaption": "woman laughing at 0.024-2.616, 4.505-6.534 and cow mooing at 1.114-4.124, 5.734-8.744 and explosion at 1.707-4.575", "frequencyCaption": "woman laughing two times and cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_38.wav", "onoffCaption": "whistling at 2.329-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_77.wav", "onoffCaption": "whistling at 0.632-8.382", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_88.wav", "onoffCaption": "explosion at 0.507-5.507, 6.588-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_105.wav", "onoffCaption": "sheep goat bleating at 2.711-4.711, 5.831-7.831", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_277.wav", "onoffCaption": "sheep goat bleating at 3.395-5.395, 6.002-8.002 and door knocking at 4.127-8.277", "frequencyCaption": "sheep goat bleating two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_292.wav", "onoffCaption": "woman laughing at 2.321-4.437, 6.796-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_389.wav", "onoffCaption": "woman laughing at 2.815-5.234", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_448.wav", "onoffCaption": "burping belching at 2.701-5.701, 6.788-9.788", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_452.wav", "onoffCaption": "cow mooing at 1.714-5.012, 7.457-9.998 and train horn at 1.894-4.374", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_549.wav", "onoffCaption": "train horn at 2.855-7.932 and dog barking at 5.735-7.735", "frequencyCaption": "train horn one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_553.wav", "onoffCaption": "door knocking at 0.301-2.489, 3.653-5.841", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_621.wav", "onoffCaption": "thump thud at 3.315-6.086, 7.481-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_818.wav", "onoffCaption": "door knocking at 0.557-2.684 and spraying at 5.889-7.064, 8.823-9.998", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_903.wav", "onoffCaption": "car horn honking at 0.806-4.393", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_956.wav", "onoffCaption": "door slamming at 0.088-1.066, 1.694-2.532, 3.141-5.141", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1063.wav", "onoffCaption": "woman laughing at 2.638-5.733, 6.69-8.916", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1079.wav", "onoffCaption": "cat meowing at 0.552-2.688, 5.012-6.159, 7.267-10.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1137.wav", "onoffCaption": "whistling at 2.697-7.181, 7.71-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1162.wav", "onoffCaption": "door slamming at 0.678-3.678, 4.346-5.495 and train horn at 7.049-10.0", "frequencyCaption": "door slamming two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1178.wav", "onoffCaption": "thump thud at 3.23-7.148", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1187.wav", "onoffCaption": "dog barking at 1.943-3.943, 4.593-6.593, 7.488-9.497", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1311.wav", "onoffCaption": "spraying at 0.006-1.768, 3.268-5.03 and door knocking at 0.879-3.342, 4.739-7.202 and door slamming at 0.879-2.403, 2.917-4.441", "frequencyCaption": "spraying two times and door knocking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1435.wav", "onoffCaption": "burping belching at 1.147-3.473 and cow mooing at 1.605-4.574", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1534.wav", "onoffCaption": "tapping clicking clanking at 2.592-6.032, 7.042-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1561.wav", "onoffCaption": "sheep goat bleating at 0.655-2.655, 4.69-6.69", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1584.wav", "onoffCaption": "woman laughing at 0.521-3.575 and tapping clicking clanking at 2.057-5.497, 7.773-10.0", "frequencyCaption": "woman laughing one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1747.wav", "onoffCaption": "door knocking at 0.364-3.201", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1830.wav", "onoffCaption": "whistling at 1.244-4.119", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1865.wav", "onoffCaption": "door slamming at 3.362-4.841", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3001.wav", "onoffCaption": "door slamming at 0.226-1.399, 2.054-4.054", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3100.wav", "onoffCaption": "woman laughing at 0.453-2.937 and dog barking at 1.239-3.239 and cat meowing at 5.498-7.438", "frequencyCaption": "woman laughing one times and dog barking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3155.wav", "onoffCaption": "tapping clicking clanking at 1.428-4.868, 5.411-8.851 and dog barking at 2.667-4.667, 5.798-7.798", "frequencyCaption": "tapping clicking clanking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3268.wav", "onoffCaption": "door slamming at 0.163-2.944, 4.553-5.531", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3373.wav", "onoffCaption": "door knocking at 0.756-5.458, 6.45-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3396.wav", "onoffCaption": "spraying at 0.33-2.79, 3.695-6.155 and whistling at 0.506-8.345", "frequencyCaption": "spraying two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3418.wav", "onoffCaption": "tapping clicking clanking at 2.902-6.342, 7.418-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3503.wav", "onoffCaption": "door knocking at 0.979-3.603, 5.118-7.245", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3556.wav", "onoffCaption": "whistling at 1.284-6.459, 7.504-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3725.wav", "onoffCaption": "sheep goat bleating at 0.533-2.533, 4.782-7.45 and spraying at 3.404-4.308", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3848.wav", "onoffCaption": "door slamming at 0.005-2.133 and duck quacking at 0.54-2.54, 3.892-5.892", "frequencyCaption": "door slamming one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3852.wav", "onoffCaption": "spraying at 0.81-1.31 and duck quacking at 3.213-5.213", "frequencyCaption": "spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3949.wav", "onoffCaption": "door slamming at 2.945-5.726, 6.358-9.139", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3953.wav", "onoffCaption": "door slamming at 1.582-3.778 and gunshot at 1.671-3.671, 4.86-6.86, 7.927-9.927 and explosion at 1.95-5.077, 5.781-8.781", "frequencyCaption": "door slamming one times and gunshot three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_22.wav", "onoffCaption": "gunshot at 0.208-2.208, 3.513-5.513", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_92.wav", "onoffCaption": "train horn at 1.487-4.287 and car horn honking at 6.554-10.0", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_121.wav", "onoffCaption": "dog barking at 0.074-2.074, 4.268-6.268", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_150.wav", "onoffCaption": "cat meowing at 3.961-5.236, 7.028-8.612 and sheep goat bleating at 5.34-7.34", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_238.wav", "onoffCaption": "cat meowing at 1.32-3.26, 5.223-6.489", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_288.wav", "onoffCaption": "whistling at 2.216-6.7", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_323.wav", "onoffCaption": "woman laughing at 2.454-5.842", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_376.wav", "onoffCaption": "tapping clicking clanking at 0.29-3.73, 6.02-9.46", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_393.wav", "onoffCaption": "burping belching at 2.451-4.777, 5.569-7.604 and cow mooing at 2.659-5.957, 6.533-9.831", "frequencyCaption": "burping belching two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_506.wav", "onoffCaption": "train horn at 0.082-2.756", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_577.wav", "onoffCaption": "spraying at 3.475-4.05, 6.365-7.429", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_720.wav", "onoffCaption": "train horn at 0.504-3.904, 6.135-8.272 and thump thud at 3.277-6.048, 6.725-9.496", "frequencyCaption": "train horn two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_775.wav", "onoffCaption": "cat meowing at 2.974-3.985, 4.638-5.649, 6.693-7.704", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_790.wav", "onoffCaption": "explosion at 0.948-2.95, 3.893-6.893", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_802.wav", "onoffCaption": "woman laughing at 0.011-4.063", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_857.wav", "onoffCaption": "woman laughing at 1.419-4.214", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_896.wav", "onoffCaption": "door slamming at 2.591-5.565 and cat meowing at 8.569-9.879", "frequencyCaption": "door slamming one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_919.wav", "onoffCaption": "explosion at 0.047-3.047 and train horn at 0.453-3.853, 4.62-6.757 and sheep goat bleating at 5.955-7.955", "frequencyCaption": "explosion one times and train horn two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1036.wav", "onoffCaption": "thump thud at 0.983-5.433, 7.61-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1086.wav", "onoffCaption": "door slamming at 0.124-1.273 and explosion at 0.735-2.823, 4.138-6.226, 6.896-8.984", "frequencyCaption": "door slamming one times and explosion three times"}
+{"filepath": "data/multi_event_train/syn_1210.wav", "onoffCaption": "spraying at 0.162-1.246, 1.853-2.937, 3.553-4.637 and burping belching at 0.281-7.449", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1245.wav", "onoffCaption": "whistling at 2.827-5.056, 6.187-8.416", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1411.wav", "onoffCaption": "sneeze at 2.848-5.087", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1460.wav", "onoffCaption": "thump thud at 0.766-3.537, 4.514-6.887", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1485.wav", "onoffCaption": "train horn at 0.635-3.115, 4.967-7.447 and whistling at 2.075-4.95 and dog barking at 2.827-4.827, 5.893-7.893", "frequencyCaption": "train horn two times and whistling one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1613.wav", "onoffCaption": "thump thud at 1.417-3.917, 6.277-9.048", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1646.wav", "onoffCaption": "burping belching at 0.959-4.239, 6.326-8.357", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1708.wav", "onoffCaption": "gunshot at 3.501-5.501, 6.563-8.563", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1931.wav", "onoffCaption": "tapping clicking clanking at 2.988-6.428 and whistling at 3.277-5.286, 5.868-8.31", "frequencyCaption": "tapping clicking clanking one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1964.wav", "onoffCaption": "cat meowing at 2.184-3.916 and spraying at 5.222-6.126, 7.621-8.525", "frequencyCaption": "cat meowing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1981.wav", "onoffCaption": "sneeze at 2.341-4.955 and spraying at 3.88-4.784, 6.693-9.277", "frequencyCaption": "sneeze one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3054.wav", "onoffCaption": "whistling at 3.252-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3227.wav", "onoffCaption": "burping belching at 1.558-5.894", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3272.wav", "onoffCaption": "burping belching at 0.208-2.534, 3.348-5.674", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3297.wav", "onoffCaption": "train horn at 0.472-3.792 and duck quacking at 5.278-7.278, 7.887-9.887", "frequencyCaption": "train horn one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3369.wav", "onoffCaption": "spraying at 0.138-0.989, 2.815-5.251 and cat meowing at 0.541-5.541, 6.668-8.204 and car horn honking at 5.496-7.961", "frequencyCaption": "spraying two times and cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3402.wav", "onoffCaption": "dog barking at 0.304-2.304, 3.917-5.917 and spraying at 3.578-4.228, 4.758-5.408 and duck quacking at 4.149-6.149", "frequencyCaption": "dog barking two times and spraying two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3457.wav", "onoffCaption": "car horn honking at 0.049-4.449 and door slamming at 2.436-4.436, 6.293-7.466", "frequencyCaption": "car horn honking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3519.wav", "onoffCaption": "woman laughing at 0.968-4.356, 6.084-8.682", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3624.wav", "onoffCaption": "door knocking at 1.229-4.845 and cat meowing at 1.724-2.999 and spraying at 7.504-8.679", "frequencyCaption": "door knocking one times and cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3671.wav", "onoffCaption": "door knocking at 3.586-5.898, 6.835-9.147", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3694.wav", "onoffCaption": "dog barking at 1.084-3.084", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3906.wav", "onoffCaption": "gunshot at 1.193-3.193, 5.106-7.106", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3977.wav", "onoffCaption": "dog barking at 0.384-2.822 and train horn at 5.374-7.854", "frequencyCaption": "dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_53.wav", "onoffCaption": "gunshot at 1.059-3.059", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_174.wav", "onoffCaption": "car horn honking at 2.435-5.93, 6.754-9.219", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_191.wav", "onoffCaption": "dog barking at 1.037-3.037, 4.049-6.049, 7.259-9.259", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_206.wav", "onoffCaption": "door slamming at 0.219-1.368, 2.751-3.9 and sheep goat bleating at 2.449-7.329", "frequencyCaption": "door slamming two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_249.wav", "onoffCaption": "spraying at 1.489-3.949, 5.232-5.74, 6.275-7.025 and explosion at 3.381-8.302", "frequencyCaption": "spraying three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_307.wav", "onoffCaption": "gunshot at 0.873-2.873, 4.967-6.967 and duck quacking at 4.095-6.095, 7.153-9.153", "frequencyCaption": "gunshot two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_352.wav", "onoffCaption": "door knocking at 0.553-2.774, 3.711-5.932", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_439.wav", "onoffCaption": "car horn honking at 3.541-7.195", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_489.wav", "onoffCaption": "burping belching at 0.236-3.236, 3.906-6.906, 7.783-10.0 and sheep goat bleating at 0.829-4.149", "frequencyCaption": "burping belching three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_522.wav", "onoffCaption": "sneeze at 1.585-3.824, 4.766-7.005, 7.533-9.772", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_592.wav", "onoffCaption": "car horn honking at 3.597-6.523", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_650.wav", "onoffCaption": "burping belching at 0.679-3.679, 4.496-7.496", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_704.wav", "onoffCaption": "spraying at 0.353-0.853, 2.125-2.625, 4.729-5.229 and train horn at 2.949-7.149", "frequencyCaption": "spraying three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_751.wav", "onoffCaption": "dog barking at 0.236-2.236, 2.934-4.934, 6.462-8.462 and cat meowing at 0.323-1.598", "frequencyCaption": "dog barking three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_826.wav", "onoffCaption": "cat meowing at 1.916-3.46 and tapping clicking clanking at 5.548-8.988", "frequencyCaption": "cat meowing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_873.wav", "onoffCaption": "burping belching at 0.036-3.905, 5.109-7.574 and whistling at 0.109-2.118, 2.672-4.681", "frequencyCaption": "burping belching two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_968.wav", "onoffCaption": "burping belching at 2.876-5.876, 6.506-8.876 and door slamming at 8.343-9.343", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1012.wav", "onoffCaption": "cow mooing at 2.858-7.838", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1047.wav", "onoffCaption": "door knocking at 0.765-3.495 and cat meowing at 6.322-8.217", "frequencyCaption": "door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1109.wav", "onoffCaption": "door knocking at 0.19-2.317, 3.942-6.775", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1234.wav", "onoffCaption": "woman laughing at 0.092-2.792, 4.844-7.544", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1261.wav", "onoffCaption": "dog barking at 0.85-2.85, 3.551-5.551 and door knocking at 1.585-4.694, 5.258-8.367 and spraying at 2.752-4.88", "frequencyCaption": "dog barking two times and door knocking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1284.wav", "onoffCaption": "dog barking at 0.114-2.114, 3.412-5.412 and cat meowing at 2.154-4.342 and gunshot at 2.395-4.395, 6.662-8.662", "frequencyCaption": "dog barking two times and cat meowing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1360.wav", "onoffCaption": "door knocking at 2.493-6.255, 7.002-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1444.wav", "onoffCaption": "gunshot at 0.506-2.506 and burping belching at 0.555-5.555", "frequencyCaption": "gunshot one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1637.wav", "onoffCaption": "tapping clicking clanking at 4.251-7.691", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1662.wav", "onoffCaption": "tapping clicking clanking at 0.128-3.568, 4.812-7.009, 7.776-9.841 and door slamming at 1.132-3.59, 5.164-7.622", "frequencyCaption": "tapping clicking clanking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1687.wav", "onoffCaption": "woman laughing at 0.062-7.507", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1736.wav", "onoffCaption": "train horn at 0.102-3.502, 4.222-6.522", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1779.wav", "onoffCaption": "cat meowing at 0.502-3.532, 5.329-8.359", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1915.wav", "onoffCaption": "train horn at 0.937-3.577, 5.909-8.278", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1940.wav", "onoffCaption": "cat meowing at 3.102-4.251 and sheep goat bleating at 7.5-9.5", "frequencyCaption": "cat meowing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3025.wav", "onoffCaption": "train horn at 0.202-3.002, 5.381-8.181", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3070.wav", "onoffCaption": "spraying at 2.319-2.946, 5.07-5.697", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3095.wav", "onoffCaption": "door knocking at 1.645-4.182, 6.269-9.116", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3203.wav", "onoffCaption": "train horn at 2.208-5.978 and door slamming at 7.84-9.231", "frequencyCaption": "train horn one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3219.wav", "onoffCaption": "gunshot at 2.327-4.327, 6.446-8.446", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3256.wav", "onoffCaption": "cat meowing at 2.573-4.144, 6.409-7.98 and sneeze at 3.459-4.713", "frequencyCaption": "cat meowing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3318.wav", "onoffCaption": "car horn honking at 0.37-4.77 and sneeze at 5.499-6.663, 7.188-8.352", "frequencyCaption": "car horn honking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3426.wav", "onoffCaption": "spraying at 2.304-4.888", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3473.wav", "onoffCaption": "train horn at 3.563-9.623", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3496.wav", "onoffCaption": "door knocking at 1.616-6.318", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3568.wav", "onoffCaption": "spraying at 2.545-3.545, 4.18-5.18, 6.037-7.037", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3600.wav", "onoffCaption": "sneeze at 1.585-3.498, 4.742-7.001", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3655.wav", "onoffCaption": "gunshot at 1.525-3.655, 4.353-6.353", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3839.wav", "onoffCaption": "sheep goat bleating at 1.587-4.883, 6.121-8.121 and door knocking at 2.008-4.383, 5.019-7.248 and duck quacking at 5.02-7.02", "frequencyCaption": "sheep goat bleating two times and door knocking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3889.wav", "onoffCaption": "sheep goat bleating at 0.727-4.647, 5.179-7.179", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3922.wav", "onoffCaption": "tapping clicking clanking at 2.51-5.95, 6.743-9.167", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3992.wav", "onoffCaption": "car horn honking at 0.694-4.943, 6.084-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_49.wav", "onoffCaption": "cat meowing at 2.928-4.545 and tapping clicking clanking at 6.788-10.0", "frequencyCaption": "cat meowing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_169.wav", "onoffCaption": "car horn honking at 3.743-8.65", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_253.wav", "onoffCaption": "car horn honking at 0.476-4.063, 6.176-9.023 and door slamming at 3.689-5.885", "frequencyCaption": "car horn honking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_348.wav", "onoffCaption": "duck quacking at 1.696-3.696 and car horn honking at 6.827-10.0", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_423.wav", "onoffCaption": "sneeze at 0.039-1.984, 3.045-5.358 and duck quacking at 0.331-2.331, 4.696-6.696 and cat meowing at 4.868-6.758, 7.943-9.833", "frequencyCaption": "sneeze two times and duck quacking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_424.wav", "onoffCaption": "dog barking at 2.088-4.088, 5.769-7.769", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_476.wav", "onoffCaption": "cow mooing at 2.637-5.619, 6.636-9.618", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_493.wav", "onoffCaption": "train horn at 2.842-6.082 and thump thud at 4.328-6.79", "frequencyCaption": "train horn one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_538.wav", "onoffCaption": "door slamming at 3.054-4.357 and sneeze at 7.84-9.785", "frequencyCaption": "door slamming one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_588.wav", "onoffCaption": "sneeze at 0.202-3.85, 5.004-6.096", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_605.wav", "onoffCaption": "woman laughing at 3.674-6.26", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_869.wav", "onoffCaption": "gunshot at 3.09-5.09, 7.337-9.337", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_920.wav", "onoffCaption": "tapping clicking clanking at 0.304-3.744, 5.854-9.294", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_927.wav", "onoffCaption": "whistling at 2.837-7.321", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_972.wav", "onoffCaption": "whistling at 0.085-9.75 and cow mooing at 0.316-4.745, 6.432-8.861 and spraying at 0.327-1.059", "frequencyCaption": "whistling one times and cow mooing two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_997.wav", "onoffCaption": "cat meowing at 0.354-1.456, 2.312-3.414 and cow mooing at 1.639-6.068, 7.489-10.0 and duck quacking at 2.356-4.356", "frequencyCaption": "cat meowing two times and cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1008.wav", "onoffCaption": "gunshot at 2.084-4.324 and sneeze at 2.735-4.974", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1113.wav", "onoffCaption": "explosion at 3.046-7.046 and thump thud at 4.891-8.558", "frequencyCaption": "explosion one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1114.wav", "onoffCaption": "car horn honking at 2.766-6.261, 7.379-9.379", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1146.wav", "onoffCaption": "sneeze at 3.579-5.896, 6.434-9.281", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1335.wav", "onoffCaption": "explosion at 1.32-4.873 and sheep goat bleating at 1.995-3.995", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1385.wav", "onoffCaption": "cat meowing at 0.38-1.924, 3.984-5.528, 6.641-8.185", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1459.wav", "onoffCaption": "whistling at 0.017-8.361", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1510.wav", "onoffCaption": "tapping clicking clanking at 1.596-5.036", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1542.wav", "onoffCaption": "car horn honking at 2.759-7.159, 7.739-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1545.wav", "onoffCaption": "thump thud at 1.587-4.358 and car horn honking at 7.312-9.777", "frequencyCaption": "thump thud one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1678.wav", "onoffCaption": "explosion at 0.267-3.82, 5.934-8.806", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1763.wav", "onoffCaption": "whistling at 2.661-5.536, 6.975-9.85", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1786.wav", "onoffCaption": "cow mooing at 0.537-3.835 and thump thud at 5.986-8.757", "frequencyCaption": "cow mooing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1814.wav", "onoffCaption": "woman laughing at 0.504-2.741, 3.614-5.818, 7.061-9.416", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_1841.wav", "onoffCaption": "burping belching at 0.979-5.002", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1846.wav", "onoffCaption": "door knocking at 3.081-5.171, 6.785-8.875", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3088.wav", "onoffCaption": "woman laughing at 1.437-8.882", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3124.wav", "onoffCaption": "sheep goat bleating at 1.512-3.512, 4.025-6.025", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3171.wav", "onoffCaption": "whistling at 0.029-5.427, 6.409-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3176.wav", "onoffCaption": "woman laughing at 2.821-5.104 and spraying at 5.22-6.301, 7.497-8.578", "frequencyCaption": "woman laughing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3193.wav", "onoffCaption": "thump thud at 0.666-3.166 and sneeze at 1.27-5.326, 6.655-8.041", "frequencyCaption": "thump thud one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3194.wav", "onoffCaption": "woman laughing at 2.365-5.437, 6.472-8.84 and spraying at 3.387-4.468, 6.773-7.854", "frequencyCaption": "woman laughing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3302.wav", "onoffCaption": "burping belching at 2.505-4.612, 5.841-7.872", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3357.wav", "onoffCaption": "cow mooing at 0.093-3.075 and explosion at 7.184-10.0", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3469.wav", "onoffCaption": "sheep goat bleating at 3.297-5.297, 7.545-9.545", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3520.wav", "onoffCaption": "cow mooing at 0.447-3.745", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3527.wav", "onoffCaption": "cow mooing at 0.083-3.381", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3572.wav", "onoffCaption": "door knocking at 2.958-7.397 and door slamming at 4.054-4.994", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3597.wav", "onoffCaption": "door slamming at 0.711-2.839, 4.986-6.377", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3701.wav", "onoffCaption": "explosion at 1.073-3.393, 4.548-6.868, 7.945-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_3754.wav", "onoffCaption": "car horn honking at 0.048-4.448, 5.584-7.585 and spraying at 2.765-3.849", "frequencyCaption": "car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3823.wav", "onoffCaption": "door slamming at 2.956-5.917 and dog barking at 4.289-6.289", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3824.wav", "onoffCaption": "explosion at 0.787-3.787, 4.651-7.651 and sheep goat bleating at 3.192-5.192, 5.983-8.134", "frequencyCaption": "explosion two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3876.wav", "onoffCaption": "burping belching at 0.142-2.507, 3.011-5.376", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3893.wav", "onoffCaption": "duck quacking at 0.014-2.014, 3.379-5.379, 7.391-9.391", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3938.wav", "onoffCaption": "spraying at 1.836-4.228, 5.119-6.183, 6.916-9.044", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3988.wav", "onoffCaption": "door slamming at 2.053-4.416 and burping belching at 7.773-10.0", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_201.wav", "onoffCaption": "duck quacking at 0.022-2.022, 3.195-5.195, 5.934-7.934 and door slamming at 0.033-1.15", "frequencyCaption": "duck quacking three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_471.wav", "onoffCaption": "sneeze at 0.419-1.583 and explosion at 3.872-5.965, 7.719-10.0", "frequencyCaption": "sneeze one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_494.wav", "onoffCaption": "cat meowing at 1.089-2.644, 5.032-6.587 and thump thud at 4.146-6.485 and sheep goat bleating at 5.667-7.667", "frequencyCaption": "cat meowing two times and thump thud one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_657.wav", "onoffCaption": "cow mooing at 0.315-3.613, 5.99-8.959", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_719.wav", "onoffCaption": "cat meowing at 0.265-1.531, 3.646-5.378", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_975.wav", "onoffCaption": "car horn honking at 0.708-5.615", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_990.wav", "onoffCaption": "spraying at 0.108-0.84, 1.923-3.685, 4.653-5.9", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1141.wav", "onoffCaption": "sneeze at 0.18-3.388, 4.291-5.585", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1229.wav", "onoffCaption": "dog barking at 2.048-5.368, 7.558-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1299.wav", "onoffCaption": "cat meowing at 0.318-4.678, 5.599-7.154 and burping belching at 0.472-4.472", "frequencyCaption": "cat meowing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1332.wav", "onoffCaption": "dog barking at 0.323-2.323, 3.124-5.124, 7.606-9.606", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1367.wav", "onoffCaption": "whistling at 1.395-3.624 and duck quacking at 2.012-4.012 and woman laughing at 6.398-8.753", "frequencyCaption": "whistling one times and duck quacking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1382.wav", "onoffCaption": "door knocking at 1.252-3.627, 4.328-6.703, 7.558-9.933", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1517.wav", "onoffCaption": "explosion at 0.423-3.423 and spraying at 1.514-2.141, 4.439-5.008", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1731.wav", "onoffCaption": "thump thud at 0.204-4.654 and door slamming at 0.552-1.691", "frequencyCaption": "thump thud one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1764.wav", "onoffCaption": "door knocking at 2.123-5.873", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1781.wav", "onoffCaption": "dog barking at 1.246-3.246, 5.194-7.194", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1813.wav", "onoffCaption": "spraying at 0.171-2.631, 3.957-4.861, 6.029-8.421", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1908.wav", "onoffCaption": "tapping clicking clanking at 1.365-4.805", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3038.wav", "onoffCaption": "cow mooing at 1.901-4.87", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3123.wav", "onoffCaption": "spraying at 1.022-3.414 and burping belching at 5.862-9.862", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3305.wav", "onoffCaption": "cow mooing at 0.655-5.084, 5.786-8.564", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3350.wav", "onoffCaption": "sheep goat bleating at 0.712-2.712, 3.787-6.394 and duck quacking at 6.274-8.274", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3575.wav", "onoffCaption": "tapping clicking clanking at 0.812-4.252, 5.887-8.473", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3590.wav", "onoffCaption": "whistling at 0.635-6.135, 7.948-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3648.wav", "onoffCaption": "spraying at 3.426-4.076 and sheep goat bleating at 7.566-9.566", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3706.wav", "onoffCaption": "whistling at 0.945-8.6", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3753.wav", "onoffCaption": "tapping clicking clanking at 3.307-6.747", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3871.wav", "onoffCaption": "explosion at 3.077-5.83, 6.473-9.226", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3894.wav", "onoffCaption": "sheep goat bleating at 0.201-2.201, 3.431-5.431", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_54.wav", "onoffCaption": "door slamming at 3.055-4.446, 5.885-8.248", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_126.wav", "onoffCaption": "dog barking at 0.327-2.327", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_254.wav", "onoffCaption": "gunshot at 0.466-2.466, 3.589-5.589, 7.552-9.552", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_355.wav", "onoffCaption": "duck quacking at 2.57-4.57, 5.358-7.358 and sheep goat bleating at 7.642-9.642", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_570.wav", "onoffCaption": "train horn at 2.982-5.782, 6.951-9.751", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_595.wav", "onoffCaption": "sneeze at 3.065-7.594", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_602.wav", "onoffCaption": "duck quacking at 0.964-2.964, 5.143-7.143", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_618.wav", "onoffCaption": "door slamming at 2.714-3.853, 4.445-5.584, 6.529-7.668", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_703.wav", "onoffCaption": "tapping clicking clanking at 1.009-4.449, 6.831-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_874.wav", "onoffCaption": "thump thud at 0.65-5.1, 6.105-8.333", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_891.wav", "onoffCaption": "thump thud at 2.63-5.13 and cat meowing at 3.305-5.037, 5.676-7.408", "frequencyCaption": "thump thud one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1040.wav", "onoffCaption": "sheep goat bleating at 1.747-3.747", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1233.wav", "onoffCaption": "gunshot at 3.194-5.194", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1328.wav", "onoffCaption": "gunshot at 0.77-2.77, 4.869-7.375, 7.885-9.885", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1416.wav", "onoffCaption": "door slamming at 0.159-0.84 and whistling at 0.385-4.869", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1665.wav", "onoffCaption": "duck quacking at 1.743-3.743, 4.623-6.623, 7.499-9.499", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1680.wav", "onoffCaption": "train horn at 0.177-2.577, 3.945-6.558 and explosion at 0.391-3.263, 3.995-6.867, 7.672-10.0 and cat meowing at 1.708-6.068", "frequencyCaption": "train horn two times and explosion three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1809.wav", "onoffCaption": "tapping clicking clanking at 0.425-3.865, 4.963-7.504", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1912.wav", "onoffCaption": "tapping clicking clanking at 2.261-5.701", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3022.wav", "onoffCaption": "sheep goat bleating at 0.296-2.296 and train horn at 4.943-8.423", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3139.wav", "onoffCaption": "sheep goat bleating at 2.639-4.639", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3204.wav", "onoffCaption": "duck quacking at 0.444-2.444 and thump thud at 5.921-8.968", "frequencyCaption": "duck quacking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3251.wav", "onoffCaption": "tapping clicking clanking at 0.42-3.86 and spraying at 2.566-3.141", "frequencyCaption": "tapping clicking clanking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3474.wav", "onoffCaption": "tapping clicking clanking at 0.664-4.104, 6.255-8.849", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3491.wav", "onoffCaption": "cat meowing at 0.031-1.391 and door knocking at 4.522-9.224", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3607.wav", "onoffCaption": "thump thud at 0.385-4.303, 5.233-7.577 and duck quacking at 5.528-7.528", "frequencyCaption": "thump thud two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3652.wav", "onoffCaption": "gunshot at 0.653-2.653", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3749.wav", "onoffCaption": "explosion at 0.571-5.165", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3970.wav", "onoffCaption": "dog barking at 3.47-5.47, 6.969-8.969", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3995.wav", "onoffCaption": "dog barking at 0.185-2.185, 2.955-4.955", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_70.wav", "onoffCaption": "dog barking at 1.993-3.993, 4.548-6.548", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_95.wav", "onoffCaption": "thump thud at 0.47-3.517, 4.762-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_102.wav", "onoffCaption": "door knocking at 2.705-5.825, 7.509-9.76", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_173.wav", "onoffCaption": "tapping clicking clanking at 0.086-3.526 and whistling at 0.707-9.805 and explosion at 2.13-4.137, 5.322-7.329", "frequencyCaption": "tapping clicking clanking one times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_196.wav", "onoffCaption": "explosion at 0.777-2.779", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_300.wav", "onoffCaption": "thump thud at 1.121-5.571", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_371.wav", "onoffCaption": "duck quacking at 1.22-3.22, 5.626-7.626", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_394.wav", "onoffCaption": "woman laughing at 2.892-5.138", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_525.wav", "onoffCaption": "explosion at 0.454-2.715, 3.764-6.025 and whistling at 1.919-7.094", "frequencyCaption": "explosion two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_554.wav", "onoffCaption": "cat meowing at 1.653-2.928, 4.67-5.945", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_669.wav", "onoffCaption": "train horn at 2.326-6.326, 7.15-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_727.wav", "onoffCaption": "whistling at 0.48-10.0 and door knocking at 1.258-6.091", "frequencyCaption": "whistling one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_756.wav", "onoffCaption": "duck quacking at 2.343-4.343, 4.875-6.875, 7.865-9.865", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_772.wav", "onoffCaption": "explosion at 2.289-5.161, 5.982-8.854", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_797.wav", "onoffCaption": "sheep goat bleating at 1.447-3.447, 5.844-7.844 and dog barking at 2.75-4.75", "frequencyCaption": "sheep goat bleating two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_821.wav", "onoffCaption": "gunshot at 0.209-2.209, 3.627-5.627 and tapping clicking clanking at 0.683-4.123", "frequencyCaption": "gunshot two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_850.wav", "onoffCaption": "train horn at 0.525-4.295 and door knocking at 0.696-3.233 and cat meowing at 2.715-5.619, 6.598-8.169", "frequencyCaption": "train horn one times and door knocking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1015.wav", "onoffCaption": "car horn honking at 0.191-3.117 and spraying at 2.06-2.568, 3.885-4.393, 5.536-6.044 and door slamming at 4.087-5.027, 6.379-7.319", "frequencyCaption": "car horn honking one times and spraying three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1064.wav", "onoffCaption": "door slamming at 1.414-4.293", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1081.wav", "onoffCaption": "door slamming at 1.436-4.153, 4.833-5.852", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1217.wav", "onoffCaption": "explosion at 0.391-3.391 and dog barking at 2.243-4.243", "frequencyCaption": "explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1242.wav", "onoffCaption": "spraying at 0.567-1.067, 2.157-2.938, 4.063-5.32", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1266.wav", "onoffCaption": "woman laughing at 0.647-2.93, 5.093-7.293 and dog barking at 4.855-6.855, 7.951-9.951", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1283.wav", "onoffCaption": "whistling at 1.596-9.981", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1359.wav", "onoffCaption": "train horn at 0.245-3.645, 4.367-7.767 and woman laughing at 3.448-6.148", "frequencyCaption": "train horn two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1398.wav", "onoffCaption": "woman laughing at 0.118-4.17, 5.424-7.707", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1432.wav", "onoffCaption": "door slamming at 3.069-4.61", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1443.wav", "onoffCaption": "cow mooing at 2.99-7.97", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1529.wav", "onoffCaption": "spraying at 0.045-3.064, 4.628-5.136, 5.647-6.498 and whistling at 0.17-7.92", "frequencyCaption": "spraying three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1558.wav", "onoffCaption": "spraying at 0.1-0.675", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1614.wav", "onoffCaption": "train horn at 2.717-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1630.wav", "onoffCaption": "spraying at 3.453-3.975, 6.279-6.787, 7.637-8.264", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1641.wav", "onoffCaption": "tapping clicking clanking at 0.201-3.641, 4.988-8.428", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1936.wav", "onoffCaption": "woman laughing at 0.152-2.852, 4.099-6.799", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1947.wav", "onoffCaption": "door knocking at 3.182-5.912, 7.132-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3006.wav", "onoffCaption": "door knocking at 3.963-7.516", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3077.wav", "onoffCaption": "whistling at 1.496-5.98", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3092.wav", "onoffCaption": "thump thud at 1.644-6.019, 7.447-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3189.wav", "onoffCaption": "explosion at 0.54-5.54, 6.412-10.0 and door knocking at 2.556-5.286, 6.463-8.926", "frequencyCaption": "explosion two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3220.wav", "onoffCaption": "sneeze at 3.349-4.513, 5.349-7.752", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3275.wav", "onoffCaption": "duck quacking at 0.417-2.417, 3.876-5.876", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3290.wav", "onoffCaption": "tapping clicking clanking at 0.227-3.667, 4.679-6.849 and duck quacking at 3.852-5.852, 7.038-9.038 and explosion at 5.023-9.617", "frequencyCaption": "tapping clicking clanking two times and duck quacking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3421.wav", "onoffCaption": "cat meowing at 0.013-1.584, 2.383-4.331, 5.483-8.387 and door knocking at 0.75-5.283, 7.441-10.0", "frequencyCaption": "cat meowing three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3450.wav", "onoffCaption": "duck quacking at 2.7-4.7, 5.509-7.509", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3623.wav", "onoffCaption": "car horn honking at 3.793-6.579", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3676.wav", "onoffCaption": "burping belching at 1.812-3.843, 6.336-8.367", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3693.wav", "onoffCaption": "sheep goat bleating at 2.944-4.944", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3738.wav", "onoffCaption": "thump thud at 0.424-4.799", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3788.wav", "onoffCaption": "sheep goat bleating at 0.787-5.507, 6.694-8.694", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3925.wav", "onoffCaption": "woman laughing at 0.292-7.737", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3954.wav", "onoffCaption": "sneeze at 0.104-2.023, 2.539-4.484, 6.855-8.101", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_25.wav", "onoffCaption": "gunshot at 1.625-3.625, 5.899-8.172 and whistling at 3.782-8.957", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_157.wav", "onoffCaption": "train horn at 0.233-2.37, 3.224-5.601", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_225.wav", "onoffCaption": "tapping clicking clanking at 2.105-5.545, 7.367-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_324.wav", "onoffCaption": "spraying at 0.009-0.509, 2.341-3.425", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_400.wav", "onoffCaption": "train horn at 2.694-5.574", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_501.wav", "onoffCaption": "cow mooing at 0.234-3.532 and whistling at 1.104-6.737 and explosion at 2.284-7.284, 7.816-10.0", "frequencyCaption": "cow mooing one times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_673.wav", "onoffCaption": "woman laughing at 0.788-3.583 and door knocking at 2.854-5.701", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_805.wav", "onoffCaption": "spraying at 1.964-4.548", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_904.wav", "onoffCaption": "car horn honking at 0.948-4.535, 6.325-8.809", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1031.wav", "onoffCaption": "explosion at 0.294-2.468", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1130.wav", "onoffCaption": "woman laughing at 1.1-3.455, 4.433-6.788 and cat meowing at 1.581-3.575", "frequencyCaption": "woman laughing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1467.wav", "onoffCaption": "door knocking at 1.655-5.405, 7.308-9.808", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1482.wav", "onoffCaption": "tapping clicking clanking at 0.246-3.686, 4.457-7.897 and door slamming at 2.901-3.739", "frequencyCaption": "tapping clicking clanking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1498.wav", "onoffCaption": "whistling at 0.046-8.057 and sneeze at 2.099-3.644, 5.951-8.197", "frequencyCaption": "whistling one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1566.wav", "onoffCaption": "train horn at 2.16-7.904", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1583.wav", "onoffCaption": "door slamming at 0.07-2.29, 3.986-6.206", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1599.wav", "onoffCaption": "whistling at 0.128-9.793", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1862.wav", "onoffCaption": "gunshot at 0.694-2.694, 3.323-5.323 and burping belching at 2.026-6.026, 6.855-10.0", "frequencyCaption": "gunshot two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1878.wav", "onoffCaption": "whistling at 0.389-8.4", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1887.wav", "onoffCaption": "cow mooing at 0.512-4.941 and thump thud at 7.964-10.0", "frequencyCaption": "cow mooing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1963.wav", "onoffCaption": "woman laughing at 1.24-3.595 and tapping clicking clanking at 1.503-4.943", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1979.wav", "onoffCaption": "door slamming at 0.075-1.214, 3.128-5.909", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1986.wav", "onoffCaption": "door slamming at 1.025-3.806, 5.303-7.303", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3049.wav", "onoffCaption": "woman laughing at 2.442-7.481", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3053.wav", "onoffCaption": "woman laughing at 2.766-5.861, 7.24-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3148.wav", "onoffCaption": "tapping clicking clanking at 0.276-3.716", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3152.wav", "onoffCaption": "sheep goat bleating at 0.454-2.454, 3.049-5.049, 6.197-8.197", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3405.wav", "onoffCaption": "whistling at 0.099-8.11", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3504.wav", "onoffCaption": "door knocking at 0.793-5.495 and thump thud at 3.052-5.391", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3800.wav", "onoffCaption": "tapping clicking clanking at 0.979-4.419, 5.28-7.969 and explosion at 2.588-5.588 and sneeze at 7.902-9.196", "frequencyCaption": "tapping clicking clanking two times and explosion one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3901.wav", "onoffCaption": "sheep goat bleating at 0.683-2.683, 4.233-6.233", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_118.wav", "onoffCaption": "train horn at 0.018-2.818, 3.936-6.736, 7.383-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_270.wav", "onoffCaption": "duck quacking at 0.329-2.329, 3.818-5.818 and sheep goat bleating at 2.212-7.092", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_295.wav", "onoffCaption": "burping belching at 0.305-2.336, 4.218-6.249, 6.891-8.922", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_455.wav", "onoffCaption": "thump thud at 2.92-7.37 and sneeze at 4.19-5.424, 6.465-7.922", "frequencyCaption": "thump thud one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_626.wav", "onoffCaption": "dog barking at 1.493-3.493", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_696.wav", "onoffCaption": "gunshot at 1.907-3.907, 4.559-6.559, 7.45-9.45", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_768.wav", "onoffCaption": "duck quacking at 3.151-5.151, 6.78-8.78", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_951.wav", "onoffCaption": "cow mooing at 2.278-5.26, 6.421-9.403", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1165.wav", "onoffCaption": "spraying at 2.992-3.992, 4.604-5.208", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1180.wav", "onoffCaption": "duck quacking at 0.539-2.539, 3.257-5.257, 6.102-8.102", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1258.wav", "onoffCaption": "whistling at 0.723-6.223 and woman laughing at 1.377-8.111 and sheep goat bleating at 6.748-8.748", "frequencyCaption": "whistling one times and woman laughing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1316.wav", "onoffCaption": "car horn honking at 2.275-5.862", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1343.wav", "onoffCaption": "duck quacking at 1.608-3.608, 4.465-6.465, 7.576-9.576", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1428.wav", "onoffCaption": "train horn at 0.016-2.153, 3.202-5.339, 5.864-8.001", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_1533.wav", "onoffCaption": "cow mooing at 2.782-5.764, 6.836-9.177", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1715.wav", "onoffCaption": "thump thud at 1.218-3.718, 4.628-7.128", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1740.wav", "onoffCaption": "train horn at 1.767-5.127, 6.095-9.455", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1837.wav", "onoffCaption": "sheep goat bleating at 2.059-6.779 and whistling at 2.884-8.059 and woman laughing at 4.959-7.378", "frequencyCaption": "sheep goat bleating one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3107.wav", "onoffCaption": "door slamming at 0.022-0.873 and cat meowing at 1.093-2.403 and train horn at 5.959-10.0", "frequencyCaption": "door slamming one times and cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3321.wav", "onoffCaption": "cat meowing at 3.21-4.21, 5.7-6.712", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3374.wav", "onoffCaption": "dog barking at 0.078-2.078, 3.208-5.208, 5.9-7.9", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3391.wav", "onoffCaption": "explosion at 0.269-3.141 and duck quacking at 0.455-2.455 and door knocking at 5.976-8.328", "frequencyCaption": "explosion one times and duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3551.wav", "onoffCaption": "door knocking at 2.84-7.673", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3639.wav", "onoffCaption": "woman laughing at 0.177-2.377, 3.151-5.351, 6.045-8.245", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3689.wav", "onoffCaption": "sheep goat bleating at 2.212-4.212, 5.084-7.569", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3722.wav", "onoffCaption": "cat meowing at 0.333-2.327, 3.269-4.805", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3777.wav", "onoffCaption": "duck quacking at 2.801-4.801 and spraying at 3.013-5.141, 6.062-6.663", "frequencyCaption": "duck quacking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3792.wav", "onoffCaption": "explosion at 2.566-4.659, 7.128-9.212", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3855.wav", "onoffCaption": "duck quacking at 1.625-3.625 and door knocking at 7.025-9.525", "frequencyCaption": "duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_159.wav", "onoffCaption": "sneeze at 0.134-3.209, 4.258-7.138, 7.933-9.852", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_231.wav", "onoffCaption": "duck quacking at 0.318-2.318, 2.949-4.949, 6.008-8.008", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_264.wav", "onoffCaption": "dog barking at 0.128-2.128 and gunshot at 1.2-3.2, 5.28-7.28", "frequencyCaption": "dog barking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_414.wav", "onoffCaption": "door knocking at 2.845-5.469, 6.563-9.4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_632.wav", "onoffCaption": "explosion at 2.325-4.389, 4.942-7.006 and tapping clicking clanking at 3.383-6.823", "frequencyCaption": "explosion two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_667.wav", "onoffCaption": "thump thud at 2.149-6.599, 7.743-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_682.wav", "onoffCaption": "cow mooing at 0.843-5.272, 7.321-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_729.wav", "onoffCaption": "door knocking at 0.601-4.369, 5.79-8.142", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_799.wav", "onoffCaption": "sheep goat bleating at 0.047-2.047 and whistling at 1.195-5.679, 7.057-10.0", "frequencyCaption": "sheep goat bleating one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_910.wav", "onoffCaption": "thump thud at 0.422-3.193, 3.817-6.156, 7.264-9.455", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1124.wav", "onoffCaption": "woman laughing at 2.461-4.829 and explosion at 2.722-4.81, 5.629-7.693", "frequencyCaption": "woman laughing one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1302.wav", "onoffCaption": "explosion at 3.167-5.428, 6.839-9.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1357.wav", "onoffCaption": "explosion at 0.493-3.246, 4.246-6.607 and door slamming at 1.075-2.554", "frequencyCaption": "explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1469.wav", "onoffCaption": "duck quacking at 0.592-2.592, 3.215-5.215, 5.836-7.836 and train horn at 1.241-4.601, 5.638-8.998 and door slamming at 5.79-6.768", "frequencyCaption": "duck quacking three times and train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1572.wav", "onoffCaption": "sneeze at 0.074-1.987, 3.317-4.42, 5.2-7.661 and door knocking at 1.161-5.671, 7.944-10.0", "frequencyCaption": "sneeze three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1597.wav", "onoffCaption": "cat meowing at 0.941-2.496 and dog barking at 1.668-3.668 and explosion at 6.21-10.0", "frequencyCaption": "cat meowing one times and dog barking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1701.wav", "onoffCaption": "spraying at 2.038-2.642", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1754.wav", "onoffCaption": "car horn honking at 0.031-2.817, 4.405-6.905", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1876.wav", "onoffCaption": "woman laughing at 0.697-3.181, 3.691-5.891", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1893.wav", "onoffCaption": "door slamming at 0.078-1.217 and dog barking at 3.861-5.861, 6.925-8.925 and thump thud at 4.033-8.483", "frequencyCaption": "door slamming one times and dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1988.wav", "onoffCaption": "door knocking at 0.078-2.168, 2.716-4.806, 5.45-7.54 and cat meowing at 1.282-2.866, 3.698-5.282, 5.908-7.492", "frequencyCaption": "door knocking three times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3146.wav", "onoffCaption": "thump thud at 0.966-3.737, 4.659-7.43 and sneeze at 1.51-2.764, 3.704-4.95, 5.513-7.729 and cow mooing at 1.812-6.792", "frequencyCaption": "thump thud two times and sneeze three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3335.wav", "onoffCaption": "whistling at 3.008-8.508", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3385.wav", "onoffCaption": "sneeze at 3.023-7.523", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3510.wav", "onoffCaption": "dog barking at 2.689-4.689, 6.283-8.283", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3678.wav", "onoffCaption": "gunshot at 2.519-4.519 and whistling at 2.939-8.114", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3763.wav", "onoffCaption": "explosion at 2.855-5.573, 6.2-8.918", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3786.wav", "onoffCaption": "door slamming at 2.563-3.854, 5.558-6.363", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3814.wav", "onoffCaption": "whistling at 1.399-9.41", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_64.wav", "onoffCaption": "dog barking at 0.582-2.582, 4.995-6.995 and spraying at 0.846-1.596 and door knocking at 4.54-8.29", "frequencyCaption": "dog barking two times and spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_81.wav", "onoffCaption": "thump thud at 1.031-5.481", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_281.wav", "onoffCaption": "cow mooing at 0.963-3.973, 4.899-7.683 and train horn at 3.96-7.96", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_441.wav", "onoffCaption": "gunshot at 0.774-2.774, 3.707-5.707, 7.02-9.02", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_540.wav", "onoffCaption": "thump thud at 2.697-6.364, 7.409-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_844.wav", "onoffCaption": "tapping clicking clanking at 0.698-4.138, 5.254-8.184", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_945.wav", "onoffCaption": "burping belching at 2.96-7.296", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1070.wav", "onoffCaption": "gunshot at 0.081-2.081, 3.247-5.247, 6.513-8.513", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1095.wav", "onoffCaption": "spraying at 0.659-2.145 and gunshot at 5.491-7.491", "frequencyCaption": "spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1171.wav", "onoffCaption": "dog barking at 0.481-2.481, 3.695-5.695", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1194.wav", "onoffCaption": "sneeze at 1.52-4.134, 5.429-7.668 and door slamming at 2.583-3.886", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1219.wav", "onoffCaption": "sheep goat bleating at 2.316-4.316 and train horn at 7.392-10.0", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1426.wav", "onoffCaption": "sheep goat bleating at 2.896-4.896 and door slamming at 4.863-5.701, 7.543-9.067", "frequencyCaption": "sheep goat bleating one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1527.wav", "onoffCaption": "spraying at 0.031-1.115, 2.086-3.17 and car horn honking at 2.198-5.417", "frequencyCaption": "spraying two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1823.wav", "onoffCaption": "tapping clicking clanking at 0.895-4.335, 5.021-8.461 and cow mooing at 2.805-5.787", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1839.wav", "onoffCaption": "cat meowing at 2.895-4.205, 5.243-6.553, 7.151-8.461", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1922.wav", "onoffCaption": "burping belching at 0.283-3.789, 4.889-8.395", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1938.wav", "onoffCaption": "cat meowing at 2.138-3.15, 4.397-5.409 and sneeze at 7.522-8.625", "frequencyCaption": "cat meowing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3008.wav", "onoffCaption": "gunshot at 0.446-2.616 and burping belching at 4.722-6.753, 7.735-10.0", "frequencyCaption": "gunshot one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3012.wav", "onoffCaption": "train horn at 2.914-7.816", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3109.wav", "onoffCaption": "whistling at 0.123-8.134 and dog barking at 2.714-4.714, 6.881-8.881", "frequencyCaption": "whistling one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3113.wav", "onoffCaption": "thump thud at 0.383-2.611", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3360.wav", "onoffCaption": "dog barking at 0.244-2.244, 4.643-6.643 and sneeze at 1.545-3.228, 4.283-6.202 and woman laughing at 2.999-5.282, 7.507-9.943", "frequencyCaption": "dog barking two times and sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3444.wav", "onoffCaption": "door knocking at 0.208-3.958, 4.484-6.611, 7.771-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3545.wav", "onoffCaption": "door slamming at 2.58-4.58, 5.255-7.255 and car horn honking at 5.744-7.744", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3736.wav", "onoffCaption": "sheep goat bleating at 0.976-2.976 and gunshot at 1.201-3.201, 4.086-6.086, 7.889-9.889", "frequencyCaption": "sheep goat bleating one times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3841.wav", "onoffCaption": "thump thud at 3.493-7.868 and door knocking at 6.458-8.77", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3940.wav", "onoffCaption": "burping belching at 3.845-6.075 and thump thud at 6.247-8.586", "frequencyCaption": "burping belching one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_31.wav", "onoffCaption": "explosion at 0.114-5.035, 6.617-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_40.wav", "onoffCaption": "cow mooing at 0.732-5.712, 6.321-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_116.wav", "onoffCaption": "burping belching at 2.633-6.656, 7.887-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_143.wav", "onoffCaption": "burping belching at 1.128-3.651 and spraying at 1.561-2.561, 3.832-4.916, 5.585-7.851", "frequencyCaption": "burping belching one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_330.wav", "onoffCaption": "sneeze at 0.333-3.443, 4.453-7.563", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_365.wav", "onoffCaption": "duck quacking at 0.019-2.019 and sheep goat bleating at 0.592-2.592 and explosion at 5.364-10.0", "frequencyCaption": "duck quacking one times and sheep goat bleating one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_380.wav", "onoffCaption": "car horn honking at 2.009-6.916 and tapping clicking clanking at 2.997-6.437", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_515.wav", "onoffCaption": "cow mooing at 0.715-3.684 and thump thud at 1.196-3.424 and cat meowing at 8.242-9.253", "frequencyCaption": "cow mooing one times and thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_581.wav", "onoffCaption": "dog barking at 0.183-2.183", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_628.wav", "onoffCaption": "dog barking at 2.434-4.434", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_698.wav", "onoffCaption": "cat meowing at 1.384-4.414", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_733.wav", "onoffCaption": "duck quacking at 3.485-5.485, 7.687-9.687", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_766.wav", "onoffCaption": "cat meowing at 0.072-1.967, 2.942-4.936 and sneeze at 0.947-2.178, 4.106-6.423", "frequencyCaption": "cat meowing two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_783.wav", "onoffCaption": "explosion at 0.107-3.101, 5.265-8.259", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_811.wav", "onoffCaption": "thump thud at 0.206-2.668, 4.309-6.537", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_860.wav", "onoffCaption": "car horn honking at 1.847-5.434 and burping belching at 2.504-5.504, 6.182-9.182", "frequencyCaption": "car horn honking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_885.wav", "onoffCaption": "door knocking at 3.613-8.613", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1025.wav", "onoffCaption": "tapping clicking clanking at 0.035-3.475, 5.44-7.618 and whistling at 0.845-2.854 and cow mooing at 2.891-7.871", "frequencyCaption": "tapping clicking clanking two times and whistling one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1054.wav", "onoffCaption": "explosion at 2.388-7.388", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1203.wav", "onoffCaption": "sneeze at 0.363-2.359, 3.331-5.327, 6.245-8.241 and whistling at 1.162-5.646 and burping belching at 4.98-7.21", "frequencyCaption": "sneeze three times and whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1256.wav", "onoffCaption": "sheep goat bleating at 1.756-3.756, 6.06-8.06", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1318.wav", "onoffCaption": "dog barking at 2.585-4.585, 5.452-7.452", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1402.wav", "onoffCaption": "car horn honking at 0.149-2.614, 3.555-6.02, 6.806-9.271", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1473.wav", "onoffCaption": "spraying at 0.374-1.106, 2.858-3.939 and burping belching at 2.746-5.746, 7.61-10.0", "frequencyCaption": "spraying two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1496.wav", "onoffCaption": "burping belching at 1.063-5.063, 6.187-10.0 and train horn at 1.338-3.738", "frequencyCaption": "burping belching two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1519.wav", "onoffCaption": "explosion at 0.451-3.578, 4.185-7.312 and car horn honking at 4.164-7.383", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1568.wav", "onoffCaption": "sneeze at 0.072-4.572 and car horn honking at 2.743-5.208", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1600.wav", "onoffCaption": "tapping clicking clanking at 2.86-6.3, 7.003-9.229 and gunshot at 3.295-5.425", "frequencyCaption": "tapping clicking clanking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1655.wav", "onoffCaption": "woman laughing at 0.265-2.684 and train horn at 0.391-4.459", "frequencyCaption": "woman laughing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1889.wav", "onoffCaption": "door knocking at 3.623-6.086, 7.931-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1906.wav", "onoffCaption": "door slamming at 2.942-3.747, 5.972-6.777", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1977.wav", "onoffCaption": "gunshot at 3.187-5.187 and thump thud at 7.862-10.0", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1992.wav", "onoffCaption": "cat meowing at 0.094-3.335", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3036.wav", "onoffCaption": "cat meowing at 2.02-3.38, 4.785-5.812 and woman laughing at 2.276-5.557, 7.167-10.0 and dog barking at 3.171-5.171, 6.364-8.364", "frequencyCaption": "cat meowing two times and woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3047.wav", "onoffCaption": "car horn honking at 0.568-3.481, 4.098-6.975", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3234.wav", "onoffCaption": "cow mooing at 0.597-5.026, 6.061-8.703", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3261.wav", "onoffCaption": "explosion at 3.738-7.578", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3284.wav", "onoffCaption": "thump thud at 0.415-4.865, 6.186-8.495", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3411.wav", "onoffCaption": "door knocking at 0.393-2.642 and whistling at 5.46-9.944", "frequencyCaption": "door knocking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3460.wav", "onoffCaption": "explosion at 4.039-6.792", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3485.wav", "onoffCaption": "whistling at 1.733-6.217 and cow mooing at 2.774-6.072, 7.142-10.0", "frequencyCaption": "whistling one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3637.wav", "onoffCaption": "sheep goat bleating at 0.468-2.468, 3.357-5.579, 6.358-8.358", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3662.wav", "onoffCaption": "spraying at 0.628-2.114, 3.145-5.537, 6.065-8.117", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3687.wav", "onoffCaption": "cat meowing at 3.218-5.108, 5.993-6.993", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3779.wav", "onoffCaption": "cow mooing at 0.863-3.832", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3915.wav", "onoffCaption": "spraying at 0.661-1.918, 2.505-4.633", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3981.wav", "onoffCaption": "dog barking at 1.467-3.467, 5.657-7.657", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_15.wav", "onoffCaption": "duck quacking at 0.518-2.518, 3.085-5.085, 6.404-8.404 and cow mooing at 2.413-6.842, 7.53-10.0 and car horn honking at 3.791-7.632", "frequencyCaption": "duck quacking three times and cow mooing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_132.wav", "onoffCaption": "whistling at 2.824-5.053, 5.615-7.844", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_167.wav", "onoffCaption": "explosion at 2.758-5.511, 7.256-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_182.wav", "onoffCaption": "door slamming at 2.185-3.664, 5.817-7.12 and car horn honking at 2.817-7.724", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_314.wav", "onoffCaption": "tapping clicking clanking at 0.093-3.533, 4.054-7.494", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_341.wav", "onoffCaption": "car horn honking at 0.634-5.146, 6.739-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_531.wav", "onoffCaption": "explosion at 0.061-5.061, 7.036-9.994", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_564.wav", "onoffCaption": "cow mooing at 3.722-6.704", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_643.wav", "onoffCaption": "spraying at 2.746-3.83, 5.719-6.803", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_659.wav", "onoffCaption": "door slamming at 1.024-1.829, 2.415-3.22, 5.384-6.189", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_717.wav", "onoffCaption": "sheep goat bleating at 1.082-3.082, 3.853-5.853 and thump thud at 7.758-10.0", "frequencyCaption": "sheep goat bleating two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_742.wav", "onoffCaption": "burping belching at 3.38-6.639", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_758.wav", "onoffCaption": "train horn at 1.569-6.646, 7.351-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_835.wav", "onoffCaption": "sheep goat bleating at 1.813-3.813, 5.405-7.405 and burping belching at 2.387-5.931", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1001.wav", "onoffCaption": "sneeze at 0.148-4.204, 5.432-9.488", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1227.wav", "onoffCaption": "tapping clicking clanking at 0.311-3.751", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1268.wav", "onoffCaption": "spraying at 0.814-1.383 and sheep goat bleating at 0.996-2.996 and train horn at 6.867-10.0", "frequencyCaption": "spraying one times and sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1272.wav", "onoffCaption": "door slamming at 1.698-4.659, 6.438-8.801 and cat meowing at 2.838-3.85", "frequencyCaption": "door slamming two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1297.wav", "onoffCaption": "door slamming at 1.376-3.376 and burping belching at 6.395-8.625", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1369.wav", "onoffCaption": "woman laughing at 0.167-3.239, 4.685-7.757", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1373.wav", "onoffCaption": "explosion at 0.042-2.91 and sneeze at 1.567-3.68, 5.675-7.829 and dog barking at 6.99-8.99", "frequencyCaption": "explosion one times and sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1396.wav", "onoffCaption": "sheep goat bleating at 1.953-6.833, 7.799-9.799", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1457.wav", "onoffCaption": "spraying at 2.206-2.728 and sheep goat bleating at 2.967-4.967", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1624.wav", "onoffCaption": "train horn at 0.453-2.893", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1671.wav", "onoffCaption": "sneeze at 0.545-1.648", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1694.wav", "onoffCaption": "sneeze at 0.873-3.983", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1725.wav", "onoffCaption": "train horn at 0.363-4.133 and whistling at 0.652-6.152, 6.997-10.0", "frequencyCaption": "train horn one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1848.wav", "onoffCaption": "sneeze at 0.755-2.049, 2.763-4.057, 4.699-5.993 and cow mooing at 0.817-3.786", "frequencyCaption": "sneeze three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1953.wav", "onoffCaption": "whistling at 0.392-9.463 and car horn honking at 2.538-4.538", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3063.wav", "onoffCaption": "tapping clicking clanking at 1.047-4.487 and sheep goat bleating at 2.67-4.67", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3086.wav", "onoffCaption": "train horn at 1.792-4.259, 5.145-7.569", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3178.wav", "onoffCaption": "burping belching at 0.665-4.209, 6.437-9.981", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3210.wav", "onoffCaption": "tapping clicking clanking at 2.508-5.948, 7.415-9.678", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3245.wav", "onoffCaption": "tapping clicking clanking at 1.293-4.733, 5.996-9.436 and train horn at 3.296-5.951, 6.594-9.249", "frequencyCaption": "tapping clicking clanking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3311.wav", "onoffCaption": "sneeze at 1.807-3.193, 4.448-5.834, 6.424-7.81", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3435.wav", "onoffCaption": "thump thud at 2.366-4.828, 6.309-8.771", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3613.wav", "onoffCaption": "tapping clicking clanking at 0.319-3.759 and dog barking at 5.607-7.607", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3646.wav", "onoffCaption": "tapping clicking clanking at 1.605-5.045, 6.511-8.868", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3708.wav", "onoffCaption": "dog barking at 0.956-2.956 and sneeze at 1.521-3.466", "frequencyCaption": "dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3747.wav", "onoffCaption": "door slamming at 3.944-5.083, 5.988-7.127", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3931.wav", "onoffCaption": "explosion at 0.693-3.422, 4.387-7.116", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3964.wav", "onoffCaption": "sheep goat bleating at 2.099-4.099, 6.494-8.494", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_128.wav", "onoffCaption": "train horn at 0.374-5.276, 6.495-9.026", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_198.wav", "onoffCaption": "gunshot at 0.784-2.784, 3.801-5.801, 6.312-8.312", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_215.wav", "onoffCaption": "whistling at 0.949-8.699", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_240.wav", "onoffCaption": "door slamming at 1.471-2.371, 3.25-4.15, 5.324-6.224", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_430.wav", "onoffCaption": "spraying at 1.877-2.461, 3.411-3.995, 4.65-5.234", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_465.wav", "onoffCaption": "dog barking at 0.878-6.515, 7.656-10.0 and tapping clicking clanking at 3.261-6.701", "frequencyCaption": "dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_616.wav", "onoffCaption": "sheep goat bleating at 2.186-4.186, 4.941-6.941", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_934.wav", "onoffCaption": "spraying at 1.534-2.709, 5.115-5.69, 7.889-10.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_984.wav", "onoffCaption": "door knocking at 3.227-7.737", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1100.wav", "onoffCaption": "sneeze at 2.114-5.189, 6.823-9.898", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1155.wav", "onoffCaption": "explosion at 0.63-3.383, 4.501-7.194", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1326.wav", "onoffCaption": "tapping clicking clanking at 2.131-5.571, 7.638-10.0 and explosion at 2.898-7.898", "frequencyCaption": "tapping clicking clanking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1418.wav", "onoffCaption": "woman laughing at 1.728-3.928, 4.951-7.688", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1503.wav", "onoffCaption": "train horn at 1.911-4.711, 5.396-7.503", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1556.wav", "onoffCaption": "explosion at 1.122-3.875 and dog barking at 7.283-9.283", "frequencyCaption": "explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1770.wav", "onoffCaption": "sneeze at 1.912-4.987, 5.626-7.238", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1795.wav", "onoffCaption": "thump thud at 0.158-3.205, 5.267-7.767", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1852.wav", "onoffCaption": "woman laughing at 0.084-2.2, 2.942-5.058, 5.698-7.814 and duck quacking at 6.149-8.149", "frequencyCaption": "woman laughing three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1949.wav", "onoffCaption": "cat meowing at 2.368-5.272, 6.225-9.129 and sheep goat bleating at 7.594-9.594", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3079.wav", "onoffCaption": "burping belching at 2.466-4.696", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3137.wav", "onoffCaption": "sneeze at 2.975-4.069", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3162.wav", "onoffCaption": "duck quacking at 2.2-4.2 and train horn at 2.651-5.811", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3187.wav", "onoffCaption": "cow mooing at 0.219-4.648, 6.771-9.484", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3344.wav", "onoffCaption": "dog barking at 0.909-2.909", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3534.wav", "onoffCaption": "cat meowing at 3.579-6.609", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3584.wav", "onoffCaption": "explosion at 2.028-7.028, 7.591-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3609.wav", "onoffCaption": "cat meowing at 0.249-1.793, 3.33-4.874, 6.573-8.117 and thump thud at 2.851-5.079", "frequencyCaption": "cat meowing three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3712.wav", "onoffCaption": "tapping clicking clanking at 0.164-3.604 and cat meowing at 1.057-2.332 and explosion at 7.493-10.0", "frequencyCaption": "tapping clicking clanking one times and cat meowing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3830.wav", "onoffCaption": "sneeze at 2.907-4.435, 6.484-8.887", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3865.wav", "onoffCaption": "door knocking at 0.025-2.337 and dog barking at 0.545-2.545 and woman laughing at 6.061-8.166", "frequencyCaption": "door knocking one times and dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_212.wav", "onoffCaption": "woman laughing at 3.453-5.736, 6.978-9.178", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_309.wav", "onoffCaption": "sheep goat bleating at 0.718-2.718, 4.398-6.398, 7.843-9.843", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_437.wav", "onoffCaption": "car horn honking at 2.062-6.311, 6.944-9.159", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_462.wav", "onoffCaption": "duck quacking at 0.153-2.153, 2.773-4.773, 5.307-7.307", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_480.wav", "onoffCaption": "explosion at 1.567-3.631, 4.557-6.621, 7.919-9.983", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_579.wav", "onoffCaption": "cat meowing at 0.489-1.849, 2.754-3.763, 5.08-6.83 and car horn honking at 4.296-6.809 and spraying at 7.112-7.963", "frequencyCaption": "cat meowing three times and car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_644.wav", "onoffCaption": "duck quacking at 0.586-2.586", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_828.wav", "onoffCaption": "spraying at 2.197-4.781", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_933.wav", "onoffCaption": "spraying at 0.268-0.843, 1.489-2.064, 3.367-3.942 and duck quacking at 1.62-3.62", "frequencyCaption": "spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_961.wav", "onoffCaption": "cat meowing at 2.641-3.653, 4.491-5.502", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_983.wav", "onoffCaption": "dog barking at 2.408-4.408, 5.296-7.296", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1107.wav", "onoffCaption": "dog barking at 0.471-2.471, 4.869-6.869, 7.846-9.846", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1374.wav", "onoffCaption": "car horn honking at 2.04-5.215, 6.66-9.835", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1391.wav", "onoffCaption": "cat meowing at 0.847-2.579", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1551.wav", "onoffCaption": "cat meowing at 2.852-5.04, 7.233-8.508", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1639.wav", "onoffCaption": "whistling at 1.187-10.0 and thump thud at 3.478-7.928", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1722.wav", "onoffCaption": "woman laughing at 0.231-7.676 and duck quacking at 3.115-5.115, 6.127-8.127", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1807.wav", "onoffCaption": "burping belching at 0.067-2.888, 4.14-6.961", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1855.wav", "onoffCaption": "thump thud at 3.516-5.744, 6.49-8.718", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3165.wav", "onoffCaption": "thump thud at 0.594-5.044 and whistling at 7.097-10.0", "frequencyCaption": "thump thud one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3180.wav", "onoffCaption": "woman laughing at 3.115-5.72", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3316.wav", "onoffCaption": "sneeze at 0.781-2.488, 3.315-4.489 and thump thud at 7.487-10.0", "frequencyCaption": "sneeze two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3428.wav", "onoffCaption": "spraying at 3.09-6.109, 7.503-9.198", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3533.wav", "onoffCaption": "spraying at 2.095-3.176, 3.935-5.668 and door knocking at 3.221-5.951, 7.684-10.0", "frequencyCaption": "spraying two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3561.wav", "onoffCaption": "tapping clicking clanking at 2.666-6.106, 7.024-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3583.wav", "onoffCaption": "cow mooing at 1.846-6.275, 7.561-9.783", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3740.wav", "onoffCaption": "duck quacking at 0.909-2.909, 3.443-5.443, 6.475-8.475 and spraying at 3.475-4.079, 4.593-5.197, 6.294-6.898 and sneeze at 4.982-7.443", "frequencyCaption": "duck quacking three times and spraying three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3837.wav", "onoffCaption": "explosion at 0.018-2.02, 2.817-4.905, 6.975-9.847", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_3862.wav", "onoffCaption": "duck quacking at 0.105-2.105, 2.639-4.639, 5.211-7.211 and sheep goat bleating at 0.824-2.824, 3.832-5.832, 7.049-9.049 and train horn at 1.048-7.108", "frequencyCaption": "duck quacking three times and sheep goat bleating three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3880.wav", "onoffCaption": "explosion at 2.536-5.592, 6.352-9.352", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3979.wav", "onoffCaption": "thump thud at 1.813-4.584 and sheep goat bleating at 2.77-4.77 and cat meowing at 7.863-9.407", "frequencyCaption": "thump thud one times and sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_247.wav", "onoffCaption": "car horn honking at 1.058-5.458 and spraying at 8.191-8.775", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_346.wav", "onoffCaption": "tapping clicking clanking at 3.424-6.864 and gunshot at 4.765-6.765 and sneeze at 4.781-7.094", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_487.wav", "onoffCaption": "gunshot at 2.351-4.521 and door slamming at 6.019-7.252", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_611.wav", "onoffCaption": "train horn at 0.168-10.0 and dog barking at 0.823-2.823, 5.236-7.236", "frequencyCaption": "train horn one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_710.wav", "onoffCaption": "burping belching at 2.117-5.661, 6.597-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_898.wav", "onoffCaption": "duck quacking at 3.594-5.594, 7.372-9.372", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_966.wav", "onoffCaption": "train horn at 3.481-6.801", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1049.wav", "onoffCaption": "woman laughing at 2.805-5.03", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1152.wav", "onoffCaption": "dog barking at 3.6-6.521", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1220.wav", "onoffCaption": "gunshot at 0.582-2.582, 3.6-5.6, 7.272-9.778 and cat meowing at 3.225-4.372", "frequencyCaption": "gunshot three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1321.wav", "onoffCaption": "cow mooing at 0.732-3.701 and gunshot at 2.699-4.699, 5.95-7.95", "frequencyCaption": "cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1504.wav", "onoffCaption": "gunshot at 2.245-4.245, 4.846-6.846", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1676.wav", "onoffCaption": "thump thud at 0.322-4.772, 6.46-9.046 and gunshot at 1.752-3.752", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1689.wav", "onoffCaption": "tapping clicking clanking at 0.371-3.811, 5.342-8.311", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1693.wav", "onoffCaption": "sneeze at 2.165-3.872, 4.533-6.24 and door knocking at 2.828-5.068", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1777.wav", "onoffCaption": "whistling at 3.115-8.29", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1788.wav", "onoffCaption": "door knocking at 0.823-3.932 and train horn at 6.027-9.797", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1792.wav", "onoffCaption": "train horn at 1.612-6.514, 7.852-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1800.wav", "onoffCaption": "duck quacking at 0.833-2.833 and dog barking at 4.411-6.411, 7.889-9.889", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3130.wav", "onoffCaption": "gunshot at 3.509-6.015", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3242.wav", "onoffCaption": "whistling at 0.169-3.144, 4.486-7.391", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3258.wav", "onoffCaption": "gunshot at 1.913-3.913 and tapping clicking clanking at 7.042-10.0", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3343.wav", "onoffCaption": "woman laughing at 0.244-2.527", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3359.wav", "onoffCaption": "door slamming at 0.337-2.465", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3498.wav", "onoffCaption": "door slamming at 0.132-2.59, 4.435-7.409 and tapping clicking clanking at 0.201-3.641, 5.333-8.773", "frequencyCaption": "door slamming two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3566.wav", "onoffCaption": "gunshot at 0.056-2.056, 3.398-5.398 and spraying at 2.521-3.029, 4.355-4.855, 5.358-7.942", "frequencyCaption": "gunshot two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3614.wav", "onoffCaption": "spraying at 3.09-3.871, 4.998-5.779", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3715.wav", "onoffCaption": "whistling at 2.984-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3887.wav", "onoffCaption": "dog barking at 1.311-3.311, 4.547-7.468", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_47.wav", "onoffCaption": "sneeze at 2.538-3.702, 5.546-6.8 and duck quacking at 3.634-5.634, 7.978-9.978", "frequencyCaption": "sneeze two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_135.wav", "onoffCaption": "sneeze at 0.624-1.716, 2.751-5.365, 5.998-8.974", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_208.wav", "onoffCaption": "car horn honking at 0.093-4.493, 6.927-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_313.wav", "onoffCaption": "car horn honking at 0.15-2.15 and door knocking at 4.327-6.576, 7.544-10.0", "frequencyCaption": "car horn honking one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_478.wav", "onoffCaption": "burping belching at 0.132-2.362, 3.987-6.313 and woman laughing at 3.778-7.166", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_563.wav", "onoffCaption": "sneeze at 0.216-1.51, 3.715-4.809", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_586.wav", "onoffCaption": "cat meowing at 0.42-2.17 and door knocking at 2.95-6.712 and door slamming at 6.165-7.145", "frequencyCaption": "cat meowing one times and door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_745.wav", "onoffCaption": "whistling at 1.282-9.032 and sneeze at 3.16-5.156, 7.293-9.289", "frequencyCaption": "whistling one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_867.wav", "onoffCaption": "tapping clicking clanking at 0.487-3.927, 6.34-9.002", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_882.wav", "onoffCaption": "gunshot at 0.137-2.137, 4.515-6.515", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_999.wav", "onoffCaption": "thump thud at 0.977-4.024, 5.736-8.075", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1053.wav", "onoffCaption": "sheep goat bleating at 2.992-4.992, 5.618-7.618", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1148.wav", "onoffCaption": "whistling at 1.666-7.607 and cat meowing at 7.548-9.084", "frequencyCaption": "whistling one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1275.wav", "onoffCaption": "duck quacking at 1.585-3.585, 4.114-6.114", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1290.wav", "onoffCaption": "whistling at 1.248-9.259 and door knocking at 1.729-4.576, 5.851-8.698", "frequencyCaption": "whistling one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1405.wav", "onoffCaption": "explosion at 0.883-5.883 and spraying at 6.432-6.94, 8.695-9.203", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1623.wav", "onoffCaption": "door knocking at 1.398-7.458", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1738.wav", "onoffCaption": "sneeze at 2.589-3.692, 4.688-6.684, 7.629-9.174", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1901.wav", "onoffCaption": "tapping clicking clanking at 1.129-4.569", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3031.wav", "onoffCaption": "whistling at 3.449-8.949", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3217.wav", "onoffCaption": "door slamming at 1.776-3.167, 5.234-6.134", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3467.wav", "onoffCaption": "cat meowing at 3.211-4.755, 5.561-7.538", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3482.wav", "onoffCaption": "sheep goat bleating at 3.012-5.012, 6.095-8.095", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3599.wav", "onoffCaption": "burping belching at 0.315-3.605, 4.439-6.669 and dog barking at 4.742-6.742 and train horn at 5.884-8.364", "frequencyCaption": "burping belching two times and dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3641.wav", "onoffCaption": "car horn honking at 0.367-3.542, 4.043-7.218 and cow mooing at 0.835-5.815, 7.094-9.789", "frequencyCaption": "car horn honking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3878.wav", "onoffCaption": "cow mooing at 2.415-7.395 and car horn honking at 2.709-7.031", "frequencyCaption": "cow mooing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3963.wav", "onoffCaption": "tapping clicking clanking at 0.761-4.201, 6.468-9.015", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3986.wav", "onoffCaption": "tapping clicking clanking at 1.468-4.908, 6.936-9.688", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_12.wav", "onoffCaption": "car horn honking at 0.54-5.447 and duck quacking at 3.391-5.391, 7.677-9.677", "frequencyCaption": "car horn honking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_63.wav", "onoffCaption": "woman laughing at 1.87-4.436, 6.27-8.553", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_86.wav", "onoffCaption": "train horn at 3.875-7.875", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_111.wav", "onoffCaption": "car horn honking at 0.756-2.756, 4.181-6.708", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_160.wav", "onoffCaption": "cow mooing at 0.675-5.655", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_185.wav", "onoffCaption": "cow mooing at 1.172-4.182 and door knocking at 2.643-6.915", "frequencyCaption": "cow mooing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_279.wav", "onoffCaption": "cat meowing at 2.652-4.84, 5.951-8.087", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_337.wav", "onoffCaption": "spraying at 3.394-3.963", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_362.wav", "onoffCaption": "gunshot at 3.922-5.922", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_387.wav", "onoffCaption": "thump thud at 0.64-2.979, 3.98-6.751", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_536.wav", "onoffCaption": "thump thud at 1.602-6.052 and burping belching at 2.209-4.316, 5.552-7.659", "frequencyCaption": "thump thud one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_547.wav", "onoffCaption": "tapping clicking clanking at 1.737-5.177, 6.879-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_734.wav", "onoffCaption": "sneeze at 1.264-3.51, 5.768-8.014", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_832.wav", "onoffCaption": "door knocking at 0.635-2.699", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_843.wav", "onoffCaption": "dog barking at 0.476-2.476, 3.128-5.128 and duck quacking at 7.098-9.098", "frequencyCaption": "dog barking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_929.wav", "onoffCaption": "burping belching at 1.548-7.492 and cat meowing at 2.021-3.065, 4.919-5.963", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_958.wav", "onoffCaption": "burping belching at 3.236-6.236", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1006.wav", "onoffCaption": "tapping clicking clanking at 0.14-3.58, 5.135-8.575", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1077.wav", "onoffCaption": "tapping clicking clanking at 0.177-3.617, 5.908-7.999 and cat meowing at 0.395-1.581 and whistling at 0.574-8.959", "frequencyCaption": "tapping clicking clanking two times and cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1092.wav", "onoffCaption": "burping belching at 0.135-4.004, 5.916-8.783", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1189.wav", "onoffCaption": "whistling at 3.358-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1204.wav", "onoffCaption": "sheep goat bleating at 0.234-2.234, 2.756-4.756, 5.332-7.332", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1421.wav", "onoffCaption": "cat meowing at 0.461-1.563 and burping belching at 0.476-5.476, 7.964-10.0", "frequencyCaption": "cat meowing one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1450.wav", "onoffCaption": "gunshot at 0.897-2.897 and cat meowing at 5.661-7.197", "frequencyCaption": "gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1652.wav", "onoffCaption": "door slamming at 0.328-1.179 and cat meowing at 4.049-5.198, 6.096-7.245 and spraying at 5.556-6.497, 8.78-9.955", "frequencyCaption": "door slamming one times and cat meowing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1749.wav", "onoffCaption": "gunshot at 1.998-3.998 and cow mooing at 5.91-8.879", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1925.wav", "onoffCaption": "explosion at 1.732-4.604", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1954.wav", "onoffCaption": "duck quacking at 1.19-3.19, 5.027-7.027", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3015.wav", "onoffCaption": "explosion at 0.061-5.061 and door knocking at 0.093-3.855", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3064.wav", "onoffCaption": "door knocking at 0.013-2.365, 3.513-5.865", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3081.wav", "onoffCaption": "woman laughing at 1.463-4.548, 5.645-7.891", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3266.wav", "onoffCaption": "burping belching at 0.198-3.4 and gunshot at 7.023-9.524", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3283.wav", "onoffCaption": "cow mooing at 3.078-8.058", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3398.wav", "onoffCaption": "sheep goat bleating at 1.518-3.518 and sneeze at 6.834-8.16", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3432.wav", "onoffCaption": "sneeze at 2.248-3.482, 5.54-6.774", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3443.wav", "onoffCaption": "whistling at 0.159-5.334", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3529.wav", "onoffCaption": "door knocking at 1.623-4.46", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3558.wav", "onoffCaption": "spraying at 1.0-2.486 and cow mooing at 5.754-8.723", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3630.wav", "onoffCaption": "train horn at 0.886-4.286, 6.072-8.872 and tapping clicking clanking at 2.266-5.706", "frequencyCaption": "train horn two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3936.wav", "onoffCaption": "spraying at 0.586-2.281, 4.305-6.0, 6.736-8.431 and sheep goat bleating at 3.311-7.231, 7.902-9.902", "frequencyCaption": "spraying three times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3947.wav", "onoffCaption": "train horn at 0.112-2.912 and whistling at 1.057-4.032 and cow mooing at 5.466-8.476", "frequencyCaption": "train horn one times and whistling one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_36.wav", "onoffCaption": "sheep goat bleating at 2.518-5.518, 6.541-9.541", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_144.wav", "onoffCaption": "cat meowing at 0.305-5.305, 6.539-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_409.wav", "onoffCaption": "explosion at 1.848-4.848, 5.553-8.553", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_413.wav", "onoffCaption": "door slamming at 1.337-1.837, 2.978-3.478", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_508.wav", "onoffCaption": "gunshot at 2.17-4.17, 5.36-7.36", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_512.wav", "onoffCaption": "dog barking at 0.622-2.622", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_660.wav", "onoffCaption": "whistling at 2.177-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_685.wav", "onoffCaption": "cow mooing at 0.06-3.029, 4.064-6.951, 7.727-9.748", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_761.wav", "onoffCaption": "thump thud at 0.843-5.293", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_784.wav", "onoffCaption": "cow mooing at 0.475-3.444, 5.211-7.742 and tapping clicking clanking at 3.309-6.749", "frequencyCaption": "cow mooing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_816.wav", "onoffCaption": "explosion at 1.724-3.726, 5.689-7.691", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_917.wav", "onoffCaption": "explosion at 1.982-6.982 and tapping clicking clanking at 3.567-7.007", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1022.wav", "onoffCaption": "whistling at 0.122-8.133 and door slamming at 0.182-1.321, 2.275-2.775, 4.526-6.526", "frequencyCaption": "whistling one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1038.wav", "onoffCaption": "train horn at 0.264-2.938, 5.056-7.536 and tapping clicking clanking at 0.704-4.144 and door knocking at 2.107-4.944, 5.54-8.377", "frequencyCaption": "train horn two times and tapping clicking clanking one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1123.wav", "onoffCaption": "tapping clicking clanking at 0.144-3.584, 5.286-7.478", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1139.wav", "onoffCaption": "woman laughing at 0.626-2.82, 5.16-7.354", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1251.wav", "onoffCaption": "thump thud at 2.421-6.796", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1350.wav", "onoffCaption": "door knocking at 0.09-2.927, 5.354-7.487 and woman laughing at 1.506-3.7, 4.763-6.851", "frequencyCaption": "door knocking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1474.wav", "onoffCaption": "gunshot at 2.464-4.464", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1491.wav", "onoffCaption": "tapping clicking clanking at 0.394-3.834, 5.02-8.46", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1575.wav", "onoffCaption": "tapping clicking clanking at 0.154-3.594, 4.396-6.695 and burping belching at 0.413-2.507, 3.041-5.148 and door slamming at 0.419-3.38, 4.399-7.36", "frequencyCaption": "tapping clicking clanking two times and burping belching two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1590.wav", "onoffCaption": "tapping clicking clanking at 2.991-6.431 and sneeze at 4.173-6.132", "frequencyCaption": "tapping clicking clanking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1607.wav", "onoffCaption": "duck quacking at 2.35-4.35, 5.08-7.08", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1706.wav", "onoffCaption": "train horn at 2.838-7.019", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1871.wav", "onoffCaption": "woman laughing at 0.217-2.855", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1894.wav", "onoffCaption": "gunshot at 0.413-2.919, 4.301-6.807", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1970.wav", "onoffCaption": "tapping clicking clanking at 0.321-3.761, 5.461-8.901", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1995.wav", "onoffCaption": "tapping clicking clanking at 0.432-3.872, 4.555-7.09", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3040.wav", "onoffCaption": "duck quacking at 2.579-4.579, 6.775-8.775", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3141.wav", "onoffCaption": "cat meowing at 0.69-1.83, 2.623-3.763, 4.769-5.909 and car horn honking at 7.826-10.0", "frequencyCaption": "cat meowing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3229.wav", "onoffCaption": "woman laughing at 3.984-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3233.wav", "onoffCaption": "thump thud at 2.174-4.674, 6.749-9.249 and door knocking at 2.65-6.266", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3328.wav", "onoffCaption": "whistling at 0.727-7.976 and thump thud at 1.013-5.463", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3332.wav", "onoffCaption": "whistling at 0.582-7.158", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3416.wav", "onoffCaption": "cow mooing at 0.067-3.365, 3.932-6.914 and train horn at 2.227-6.667", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3517.wav", "onoffCaption": "gunshot at 1.138-3.138", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3665.wav", "onoffCaption": "car horn honking at 0.511-4.833 and whistling at 3.066-5.295, 6.173-8.456", "frequencyCaption": "car horn honking one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3680.wav", "onoffCaption": "sneeze at 0.858-2.386", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3764.wav", "onoffCaption": "tapping clicking clanking at 1.057-4.497, 6.107-8.928 and whistling at 1.373-3.382, 4.394-7.369", "frequencyCaption": "tapping clicking clanking two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3781.wav", "onoffCaption": "car horn honking at 2.047-5.634 and door slamming at 8.135-8.94", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3809.wav", "onoffCaption": "thump thud at 3.075-5.846, 6.39-9.161", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3813.wav", "onoffCaption": "train horn at 1.522-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3908.wav", "onoffCaption": "spraying at 0.562-2.295 and gunshot at 5.254-7.384", "frequencyCaption": "spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3912.wav", "onoffCaption": "gunshot at 2.345-4.345, 5.839-7.839", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_79.wav", "onoffCaption": "door knocking at 1.555-4.675, 6.777-9.897", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_189.wav", "onoffCaption": "dog barking at 2.828-4.828, 6.53-8.53", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_204.wav", "onoffCaption": "car horn honking at 2.9-5.365, 7.793-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_236.wav", "onoffCaption": "car horn honking at 2.477-4.977, 5.601-8.002", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_263.wav", "onoffCaption": "cow mooing at 0.559-3.857, 5.176-8.474", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_286.wav", "onoffCaption": "spraying at 3.75-6.186, 7.898-10.0", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_378.wav", "onoffCaption": "woman laughing at 0.099-2.894, 4.837-7.297 and explosion at 0.157-3.157", "frequencyCaption": "woman laughing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_421.wav", "onoffCaption": "tapping clicking clanking at 2.375-5.815 and sheep goat bleating at 4.545-6.545", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_446.wav", "onoffCaption": "cat meowing at 0.124-1.337, 2.502-3.715, 4.842-6.055", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_635.wav", "onoffCaption": "sheep goat bleating at 0.123-2.123, 2.76-4.76, 5.892-7.892", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_652.wav", "onoffCaption": "cow mooing at 3.561-6.571, 7.654-10.0 and duck quacking at 4.115-6.115", "frequencyCaption": "cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_859.wav", "onoffCaption": "duck quacking at 0.31-2.31, 4.52-6.52 and thump thud at 4.117-8.035", "frequencyCaption": "duck quacking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_925.wav", "onoffCaption": "whistling at 1.132-8.882", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_942.wav", "onoffCaption": "door slamming at 2.154-3.054", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1088.wav", "onoffCaption": "cow mooing at 0.067-3.365 and explosion at 6.581-10.0", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1111.wav", "onoffCaption": "whistling at 2.179-9.929", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1176.wav", "onoffCaption": "spraying at 0.051-0.801, 2.592-3.342, 5.836-6.586 and dog barking at 1.423-3.423", "frequencyCaption": "spraying three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1193.wav", "onoffCaption": "cow mooing at 0.157-3.455, 4.239-7.537 and door slamming at 1.499-2.479, 3.672-4.652", "frequencyCaption": "cow mooing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1279.wav", "onoffCaption": "tapping clicking clanking at 0.933-4.373, 4.929-7.481", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1305.wav", "onoffCaption": "woman laughing at 3.907-6.19 and duck quacking at 7.724-9.724", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1362.wav", "onoffCaption": "sheep goat bleating at 2.838-4.838", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1387.wav", "onoffCaption": "dog barking at 1.304-3.304, 3.955-5.955, 6.664-8.664", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1520.wav", "onoffCaption": "tapping clicking clanking at 2.832-6.272, 6.792-9.216", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1547.wav", "onoffCaption": "dog barking at 0.286-2.286, 4.345-6.345", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1648.wav", "onoffCaption": "gunshot at 0.398-2.398 and cat meowing at 2.098-3.11, 5.228-6.24", "frequencyCaption": "gunshot one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1734.wav", "onoffCaption": "explosion at 2.33-5.33, 6.621-8.676 and duck quacking at 3.013-5.013, 5.937-7.937", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1753.wav", "onoffCaption": "thump thud at 1.028-4.695, 5.376-9.043", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1824.wav", "onoffCaption": "dog barking at 0.275-2.275, 4.236-6.636", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1843.wav", "onoffCaption": "train horn at 2.459-5.779 and cow mooing at 2.492-5.502", "frequencyCaption": "train horn one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1958.wav", "onoffCaption": "sheep goat bleating at 2.891-4.891 and spraying at 8.283-9.347", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3068.wav", "onoffCaption": "door knocking at 2.829-6.204, 7.014-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3114.wav", "onoffCaption": "cow mooing at 2.794-5.776, 6.601-9.583", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3173.wav", "onoffCaption": "woman laughing at 3.122-6.222", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3196.wav", "onoffCaption": "duck quacking at 0.461-2.461, 2.976-4.976, 5.619-7.619", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3299.wav", "onoffCaption": "whistling at 0.056-8.441", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3300.wav", "onoffCaption": "spraying at 0.055-1.055 and woman laughing at 0.776-4.057, 4.642-7.923", "frequencyCaption": "spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3367.wav", "onoffCaption": "cow mooing at 0.244-4.673 and thump thud at 2.481-5.528, 7.722-9.95", "frequencyCaption": "cow mooing one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3382.wav", "onoffCaption": "cat meowing at 1.522-6.522, 8.431-9.443", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3459.wav", "onoffCaption": "train horn at 2.054-6.912 and door slamming at 4.29-6.486, 7.105-9.233", "frequencyCaption": "train horn one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3525.wav", "onoffCaption": "dog barking at 2.799-4.799, 6.347-8.347", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3542.wav", "onoffCaption": "car horn honking at 0.751-3.664 and duck quacking at 6.723-8.723", "frequencyCaption": "car horn honking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3731.wav", "onoffCaption": "sheep goat bleating at 0.142-2.142, 3.083-5.083 and burping belching at 7.866-9.897", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3756.wav", "onoffCaption": "cow mooing at 1.082-6.062, 7.102-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3821.wav", "onoffCaption": "sneeze at 1.225-2.611, 3.422-4.808, 5.872-7.258 and cat meowing at 2.894-4.465, 6.911-7.922", "frequencyCaption": "sneeze three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3846.wav", "onoffCaption": "car horn honking at 0.208-4.53, 6.828-9.725 and whistling at 1.886-7.888", "frequencyCaption": "car horn honking two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_51.wav", "onoffCaption": "cat meowing at 0.538-5.538, 6.28-8.719 and door knocking at 5.551-7.678", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_139.wav", "onoffCaption": "car horn honking at 0.702-3.628, 4.583-7.509", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_251.wav", "onoffCaption": "whistling at 1.799-6.974, 7.85-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_350.wav", "onoffCaption": "train horn at 0.188-3.668, 5.14-8.62 and gunshot at 2.187-4.187, 5.945-7.945 and door knocking at 2.794-5.914, 6.611-8.799", "frequencyCaption": "train horn two times and gunshot two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_474.wav", "onoffCaption": "dog barking at 0.186-3.107, 4.014-6.014, 6.644-8.644 and duck quacking at 0.882-2.882, 5.239-7.239", "frequencyCaption": "dog barking three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_491.wav", "onoffCaption": "duck quacking at 3.523-5.523", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_590.wav", "onoffCaption": "cat meowing at 2.657-7.017, 7.526-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_607.wav", "onoffCaption": "whistling at 1.666-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_706.wav", "onoffCaption": "sneeze at 0.5-2.183, 4.167-5.85", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_749.wav", "onoffCaption": "thump thud at 0.647-3.147, 4.243-6.582 and dog barking at 4.148-6.148", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_871.wav", "onoffCaption": "train horn at 1.148-3.822, 5.745-8.419 and sheep goat bleating at 2.434-4.434", "frequencyCaption": "train horn two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_894.wav", "onoffCaption": "cow mooing at 2.365-6.794, 7.82-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_970.wav", "onoffCaption": "door knocking at 2.028-4.796, 6.209-8.977 and whistling at 2.521-5.496", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_995.wav", "onoffCaption": "woman laughing at 3.336-6.131", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1045.wav", "onoffCaption": "woman laughing at 3.529-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1144.wav", "onoffCaption": "thump thud at 0.479-4.929 and tapping clicking clanking at 3.187-6.627", "frequencyCaption": "thump thud one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1236.wav", "onoffCaption": "cow mooing at 2.627-5.596, 6.917-9.886", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1337.wav", "onoffCaption": "burping belching at 2.433-4.636, 5.194-7.367", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1409.wav", "onoffCaption": "sneeze at 3.14-5.601, 7.313-9.789 and thump thud at 3.253-6.3", "frequencyCaption": "sneeze two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1413.wav", "onoffCaption": "sneeze at 0.398-2.105 and spraying at 0.939-3.067", "frequencyCaption": "sneeze one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1508.wav", "onoffCaption": "explosion at 0.103-3.29", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1512.wav", "onoffCaption": "door slamming at 0.683-2.596 and car horn honking at 5.318-9.159", "frequencyCaption": "door slamming one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1660.wav", "onoffCaption": "cow mooing at 0.418-3.428, 3.973-6.583", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1685.wav", "onoffCaption": "sheep goat bleating at 0.076-2.076, 2.698-4.698, 5.861-7.861", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1761.wav", "onoffCaption": "thump thud at 0.738-5.188, 7.319-9.819", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1784.wav", "onoffCaption": "cow mooing at 0.738-5.167, 5.903-8.899", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1816.wav", "onoffCaption": "cat meowing at 0.631-4.764, 6.378-7.527", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1917.wav", "onoffCaption": "burping belching at 0.12-3.664, 4.722-8.266", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3027.wav", "onoffCaption": "explosion at 1.301-6.301, 7.603-9.923 and gunshot at 2.288-4.762, 5.447-7.921", "frequencyCaption": "explosion two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3126.wav", "onoffCaption": "spraying at 0.002-0.906, 1.602-3.335 and gunshot at 6.789-8.789", "frequencyCaption": "spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3254.wav", "onoffCaption": "burping belching at 1.066-4.346 and sheep goat bleating at 2.429-4.429, 5.198-7.198", "frequencyCaption": "burping belching one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3355.wav", "onoffCaption": "cat meowing at 0.682-1.709, 2.888-3.915, 4.78-5.807 and burping belching at 0.736-3.359", "frequencyCaption": "cat meowing three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3471.wav", "onoffCaption": "spraying at 1.916-3.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3494.wav", "onoffCaption": "explosion at 2.713-5.441, 7.005-9.733", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3570.wav", "onoffCaption": "gunshot at 1.296-3.296, 4.812-6.812", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3595.wav", "onoffCaption": "cat meowing at 2.868-7.228, 8.142-9.291", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3602.wav", "onoffCaption": "explosion at 0.527-3.527 and train horn at 1.368-3.808, 5.59-8.03", "frequencyCaption": "explosion one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3618.wav", "onoffCaption": "sheep goat bleating at 3.257-5.257", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3703.wav", "onoffCaption": "explosion at 0.155-5.155 and sheep goat bleating at 0.686-3.686, 4.666-7.666 and duck quacking at 6.741-8.741", "frequencyCaption": "explosion one times and sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3719.wav", "onoffCaption": "sheep goat bleating at 0.275-2.275, 3.223-5.223, 5.77-7.77", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3874.wav", "onoffCaption": "duck quacking at 3.559-5.559, 6.561-8.561", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3891.wav", "onoffCaption": "whistling at 0.043-2.918 and dog barking at 0.052-2.052", "frequencyCaption": "whistling one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3990.wav", "onoffCaption": "tapping clicking clanking at 0.705-4.145, 5.601-7.818", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_90.wav", "onoffCaption": "woman laughing at 0.803-3.408, 3.992-6.787", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.202-2.202, 3.375-5.375 and sneeze at 3.476-5.472, 6.037-8.033 and cow mooing at 5.23-8.24", "frequencyCaption": "sheep goat bleating two times and sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_176.wav", "onoffCaption": "cow mooing at 3.536-6.834, 7.601-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_193.wav", "onoffCaption": "explosion at 0.339-3.339, 5.515-7.877", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_305.wav", "onoffCaption": "train horn at 3.158-5.625, 6.413-9.115 and cow mooing at 5.053-8.022", "frequencyCaption": "train horn two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_374.wav", "onoffCaption": "dog barking at 0.106-2.106, 3.22-5.22", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_391.wav", "onoffCaption": "explosion at 0.435-3.164, 4.541-6.834 and spraying at 3.273-6.292", "frequencyCaption": "explosion two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_520.wav", "onoffCaption": "thump thud at 1.037-3.376 and gunshot at 5.909-7.909", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_575.wav", "onoffCaption": "door knocking at 2.665-5.433 and spraying at 7.735-8.339, 8.877-9.481", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_639.wav", "onoffCaption": "dog barking at 0.024-2.024, 2.943-4.943, 5.802-7.802", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_648.wav", "onoffCaption": "car horn honking at 0.672-4.326 and tapping clicking clanking at 6.904-10.0", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_722.wav", "onoffCaption": "whistling at 0.563-8.574 and tapping clicking clanking at 1.14-4.58", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_753.wav", "onoffCaption": "explosion at 0.834-3.834, 5.832-8.832", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_824.wav", "onoffCaption": "burping belching at 2.968-5.198, 6.24-9.24", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_855.wav", "onoffCaption": "sneeze at 1.041-3.655, 5.037-7.651", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1010.wav", "onoffCaption": "cow mooing at 2.924-5.893", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1084.wav", "onoffCaption": "explosion at 2.02-7.02", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1212.wav", "onoffCaption": "cow mooing at 0.686-3.696 and door knocking at 7.844-10.0", "frequencyCaption": "cow mooing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1263.wav", "onoffCaption": "whistling at 2.648-5.523, 6.602-9.287", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1286.wav", "onoffCaption": "sneeze at 2.554-4.513 and thump thud at 6.691-10.0", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1309.wav", "onoffCaption": "cow mooing at 3.532-7.961 and spraying at 4.142-5.389, 6.187-7.128", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1378.wav", "onoffCaption": "door slamming at 0.234-2.234 and explosion at 5.159-7.877 and cat meowing at 5.508-7.502", "frequencyCaption": "door slamming one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1437.wav", "onoffCaption": "car horn honking at 0.038-2.964, 4.164-6.34 and dog barking at 0.425-2.425", "frequencyCaption": "car horn honking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1446.wav", "onoffCaption": "cow mooing at 3.706-6.688, 7.502-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1635.wav", "onoffCaption": "car horn honking at 3.734-6.081, 6.76-8.924", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1644.wav", "onoffCaption": "woman laughing at 4.06-6.479, 7.285-9.704", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1828.wav", "onoffCaption": "dog barking at 0.029-2.029", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1859.wav", "onoffCaption": "sneeze at 0.081-3.156 and door slamming at 1.281-1.962", "frequencyCaption": "sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1933.wav", "onoffCaption": "spraying at 0.054-0.835, 1.62-2.189, 3.452-5.185 and sheep goat bleating at 1.264-3.264, 3.814-5.814, 7.277-9.277", "frequencyCaption": "spraying three times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1942.wav", "onoffCaption": "sheep goat bleating at 1.739-3.739, 6.232-8.232", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3003.wav", "onoffCaption": "whistling at 1.472-9.857 and burping belching at 1.552-4.552", "frequencyCaption": "whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3072.wav", "onoffCaption": "gunshot at 0.892-2.892, 4.482-6.983", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3097.wav", "onoffCaption": "tapping clicking clanking at 0.349-3.789, 4.563-8.003", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3118.wav", "onoffCaption": "door knocking at 1.689-5.457", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3169.wav", "onoffCaption": "train horn at 3.286-8.188", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3201.wav", "onoffCaption": "woman laughing at 0.017-6.751 and sheep goat bleating at 4.882-6.882", "frequencyCaption": "woman laughing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3270.wav", "onoffCaption": "spraying at 0.023-1.718, 3.241-4.936 and burping belching at 2.094-5.374", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3295.wav", "onoffCaption": "gunshot at 0.894-2.894 and tapping clicking clanking at 1.696-5.136, 6.245-8.873 and cat meowing at 3.659-4.744", "frequencyCaption": "gunshot one times and tapping clicking clanking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3424.wav", "onoffCaption": "door slamming at 2.693-3.832, 5.0-6.139, 6.66-7.799", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3455.wav", "onoffCaption": "whistling at 0.61-9.665", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3626.wav", "onoffCaption": "tapping clicking clanking at 3.218-6.658", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3657.wav", "onoffCaption": "gunshot at 0.034-2.034, 3.662-5.662, 7.081-9.081", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3920.wav", "onoffCaption": "door slamming at 2.209-5.17", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3975.wav", "onoffCaption": "whistling at 1.833-7.333 and thump thud at 3.092-7.467 and cat meowing at 4.531-6.115", "frequencyCaption": "whistling one times and thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_20.wav", "onoffCaption": "thump thud at 0.02-4.47", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_75.wav", "onoffCaption": "burping belching at 0.181-2.704, 4.987-7.081 and duck quacking at 1.551-3.551, 5.137-7.137", "frequencyCaption": "burping belching two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_107.wav", "onoffCaption": "cow mooing at 2.043-7.023", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_148.wav", "onoffCaption": "door knocking at 2.621-4.809 and explosion at 6.949-9.821", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_152.wav", "onoffCaption": "burping belching at 0.519-3.778 and tapping clicking clanking at 6.986-10.0", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_321.wav", "onoffCaption": "woman laughing at 0.731-3.336, 4.07-6.675, 7.954-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_405.wav", "onoffCaption": "cow mooing at 0.652-5.632, 6.45-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_504.wav", "onoffCaption": "door knocking at 0.327-5.16, 5.971-8.528 and dog barking at 2.262-4.262 and cat meowing at 3.064-7.424", "frequencyCaption": "door knocking two times and dog barking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_551.wav", "onoffCaption": "cow mooing at 0.409-3.378, 3.897-6.651", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_689.wav", "onoffCaption": "gunshot at 1.993-3.993 and tapping clicking clanking at 6.029-9.469", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_777.wav", "onoffCaption": "burping belching at 0.758-2.961", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_792.wav", "onoffCaption": "dog barking at 0.781-2.781", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_800.wav", "onoffCaption": "spraying at 0.055-0.563, 1.558-2.558, 3.071-3.922 and woman laughing at 5.64-7.84", "frequencyCaption": "spraying three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_901.wav", "onoffCaption": "door slamming at 0.191-1.715, 3.34-4.359 and sneeze at 7.275-10.0", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1034.wav", "onoffCaption": "whistling at 2.152-9.807", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1061.wav", "onoffCaption": "train horn at 0.129-2.266 and sheep goat bleating at 1.564-4.86, 5.722-7.722 and spraying at 6.161-7.408", "frequencyCaption": "train horn one times and sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1135.wav", "onoffCaption": "cat meowing at 0.983-2.123, 2.981-5.67 and thump thud at 1.87-6.32", "frequencyCaption": "cat meowing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1247.wav", "onoffCaption": "burping belching at 3.068-6.612, 7.278-9.991", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1462.wav", "onoffCaption": "explosion at 0.227-3.095, 4.092-6.774, 7.66-9.67", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_1478.wav", "onoffCaption": "explosion at 0.04-2.912, 4.447-7.319", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1487.wav", "onoffCaption": "thump thud at 2.941-7.391", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1563.wav", "onoffCaption": "duck quacking at 2.444-4.444 and woman laughing at 7.55-10.0", "frequencyCaption": "duck quacking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1579.wav", "onoffCaption": "door knocking at 0.166-3.928, 4.859-8.621 and door slamming at 6.402-9.183", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1586.wav", "onoffCaption": "door knocking at 1.791-4.031, 5.253-7.493 and train horn at 2.702-5.902", "frequencyCaption": "door knocking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1611.wav", "onoffCaption": "thump thud at 0.696-4.363 and sheep goat bleating at 2.839-5.919, 7.879-9.879", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1867.wav", "onoffCaption": "cow mooing at 2.159-5.141, 5.718-8.591", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1882.wav", "onoffCaption": "gunshot at 0.885-3.125, 5.094-7.334 and sheep goat bleating at 2.826-4.826, 6.678-8.678", "frequencyCaption": "gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1898.wav", "onoffCaption": "duck quacking at 3.245-5.245, 6.684-8.684", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1966.wav", "onoffCaption": "explosion at 0.09-2.097, 4.572-7.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1983.wav", "onoffCaption": "dog barking at 0.369-2.369, 4.251-6.251", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1999.wav", "onoffCaption": "cat meowing at 2.229-3.24", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3056.wav", "onoffCaption": "train horn at 1.134-3.774", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3157.wav", "onoffCaption": "train horn at 0.176-4.296 and cat meowing at 7.151-8.706", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3225.wav", "onoffCaption": "car horn honking at 2.801-5.727, 6.289-8.754 and duck quacking at 4.606-6.606", "frequencyCaption": "car horn honking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3400.wav", "onoffCaption": "explosion at 1.732-4.726, 7.146-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3501.wav", "onoffCaption": "duck quacking at 0.252-2.252", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3673.wav", "onoffCaption": "spraying at 0.097-1.161, 2.725-4.853 and gunshot at 0.222-2.222, 4.038-6.038, 7.509-9.509", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3696.wav", "onoffCaption": "train horn at 1.872-5.992", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3768.wav", "onoffCaption": "burping belching at 0.085-4.085, 5.224-9.224 and train horn at 5.33-9.53", "frequencyCaption": "burping belching two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3805.wav", "onoffCaption": "explosion at 0.285-2.813, 5.27-8.27 and duck quacking at 1.387-3.387", "frequencyCaption": "explosion two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3904.wav", "onoffCaption": "tapping clicking clanking at 2.524-5.964, 7.219-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3951.wav", "onoffCaption": "burping belching at 2.919-6.121, 6.638-9.84", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_220.wav", "onoffCaption": "cow mooing at 1.917-6.346, 7.969-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_227.wav", "onoffCaption": "cat meowing at 3.979-5.006 and cow mooing at 7.653-10.0", "frequencyCaption": "cat meowing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_275.wav", "onoffCaption": "tapping clicking clanking at 0.18-3.62", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_290.wav", "onoffCaption": "train horn at 0.463-3.703, 5.072-7.818 and whistling at 5.549-7.778", "frequencyCaption": "train horn two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_402.wav", "onoffCaption": "woman laughing at 1.858-4.496, 5.165-7.649 and burping belching at 2.466-5.466", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_450.wav", "onoffCaption": "tapping clicking clanking at 3.642-7.082", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_519.wav", "onoffCaption": "tapping clicking clanking at 2.995-6.435", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_623.wav", "onoffCaption": "dog barking at 0.836-2.836, 3.847-5.847, 6.983-8.983", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_671.wav", "onoffCaption": "explosion at 2.427-4.52, 5.452-8.143", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_676.wav", "onoffCaption": "spraying at 2.443-3.443 and train horn at 2.832-6.192", "frequencyCaption": "spraying one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_693.wav", "onoffCaption": "burping belching at 1.052-6.653, 7.502-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_694.wav", "onoffCaption": "tapping clicking clanking at 0.286-3.726, 4.727-8.167", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_738.wav", "onoffCaption": "burping belching at 2.978-9.955", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_788.wav", "onoffCaption": "woman laughing at 2.872-4.964, 6.418-8.711", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_906.wav", "onoffCaption": "duck quacking at 1.199-3.199, 5.451-7.451 and spraying at 1.857-2.798", "frequencyCaption": "duck quacking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_954.wav", "onoffCaption": "woman laughing at 1.448-3.685, 5.85-8.087", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1029.wav", "onoffCaption": "woman laughing at 0.266-2.852 and spraying at 1.809-2.393, 4.645-5.229", "frequencyCaption": "woman laughing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1132.wav", "onoffCaption": "train horn at 0.148-3.508, 4.725-8.085", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1160.wav", "onoffCaption": "sheep goat bleating at 0.318-2.318, 3.094-5.094, 7.441-9.441", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1185.wav", "onoffCaption": "thump thud at 0.834-3.605 and duck quacking at 1.364-3.364, 5.81-7.81", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1208.wav", "onoffCaption": "spraying at 1.244-1.752, 4.209-4.959 and sneeze at 7.711-9.037", "frequencyCaption": "spraying two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1313.wav", "onoffCaption": "cow mooing at 0.194-3.163, 5.187-7.601 and burping belching at 0.273-2.304 and woman laughing at 3.885-6.585", "frequencyCaption": "cow mooing two times and burping belching one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1341.wav", "onoffCaption": "cow mooing at 2.719-5.701, 6.832-9.814", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1346.wav", "onoffCaption": "spraying at 0.168-1.019", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1536.wav", "onoffCaption": "car horn honking at 1.992-5.833, 6.819-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1564.wav", "onoffCaption": "cat meowing at 0.177-2.072, 4.308-6.203 and duck quacking at 0.678-2.678, 4.915-6.915", "frequencyCaption": "cat meowing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1581.wav", "onoffCaption": "train horn at 2.551-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1710.wav", "onoffCaption": "cat meowing at 0.096-1.198, 2.271-3.373, 4.359-5.461 and explosion at 0.76-2.824, 3.965-6.029 and sheep goat bleating at 4.978-6.978", "frequencyCaption": "cat meowing three times and explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1717.wav", "onoffCaption": "explosion at 0.595-3.595 and woman laughing at 6.279-8.634", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1745.wav", "onoffCaption": "burping belching at 0.264-3.264, 4.921-7.921", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1832.wav", "onoffCaption": "train horn at 2.432-5.072 and cow mooing at 2.509-5.478, 6.707-9.676", "frequencyCaption": "train horn one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1860.wav", "onoffCaption": "cat meowing at 0.869-2.009, 4.238-5.265, 6.706-8.666", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1885.wav", "onoffCaption": "thump thud at 0.058-2.286", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1929.wav", "onoffCaption": "train horn at 1.408-3.563, 4.706-6.861", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3019.wav", "onoffCaption": "tapping clicking clanking at 1.778-5.218, 6.534-9.974", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3102.wav", "onoffCaption": "dog barking at 2.444-4.444 and tapping clicking clanking at 5.38-8.82 and thump thud at 6.34-8.568", "frequencyCaption": "dog barking one times and tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3150.wav", "onoffCaption": "door knocking at 0.459-2.922 and sneeze at 7.356-9.419", "frequencyCaption": "door knocking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3238.wav", "onoffCaption": "whistling at 0.15-9.815 and sheep goat bleating at 1.787-3.787, 4.729-6.729", "frequencyCaption": "whistling one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3323.wav", "onoffCaption": "spraying at 0.067-0.817", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3324.wav", "onoffCaption": "sneeze at 2.984-7.04", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3371.wav", "onoffCaption": "duck quacking at 2.226-4.226, 5.912-7.912", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3394.wav", "onoffCaption": "explosion at 2.848-6.401, 7.091-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3506.wav", "onoffCaption": "whistling at 0.097-7.847 and door knocking at 0.498-3.266, 4.128-6.508, 7.796-10.0", "frequencyCaption": "whistling one times and door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3554.wav", "onoffCaption": "train horn at 0.045-4.113, 4.781-7.661", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3669.wav", "onoffCaption": "burping belching at 0.131-2.892, 4.13-6.891, 7.797-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3727.wav", "onoffCaption": "sneeze at 1.669-3.665", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3772.wav", "onoffCaption": "door knocking at 2.434-5.501, 6.209-9.276", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3775.wav", "onoffCaption": "car horn honking at 0.698-3.516 and sheep goat bleating at 4.98-6.98", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3790.wav", "onoffCaption": "cow mooing at 3.114-6.083, 7.337-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3797.wav", "onoffCaption": "duck quacking at 0.574-2.574, 3.148-5.148, 5.713-7.713 and dog barking at 2.459-4.459, 6.087-8.087", "frequencyCaption": "duck quacking three times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3802.wav", "onoffCaption": "whistling at 0.586-8.597 and sheep goat bleating at 1.539-3.539, 5.134-7.134", "frequencyCaption": "whistling one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3850.wav", "onoffCaption": "thump thud at 0.06-4.51, 6.619-8.847", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3919.wav", "onoffCaption": "sheep goat bleating at 0.38-2.38", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_68.wav", "onoffCaption": "spraying at 1.145-1.72, 2.499-3.074, 3.755-4.33 and tapping clicking clanking at 1.944-5.384, 6.002-8.278", "frequencyCaption": "spraying three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_72.wav", "onoffCaption": "spraying at 2.92-3.547, 4.121-5.378, 6.063-8.191", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_97.wav", "onoffCaption": "thump thud at 0.697-4.364, 5.526-7.754", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_100.wav", "onoffCaption": "door knocking at 2.269-5.389 and explosion at 7.884-10.0", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_272.wav", "onoffCaption": "burping belching at 0.523-4.029, 6.193-9.699", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_297.wav", "onoffCaption": "tapping clicking clanking at 1.345-4.785, 6.911-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_369.wav", "onoffCaption": "burping belching at 0.269-2.367", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_396.wav", "onoffCaption": "cow mooing at 0.073-3.055, 5.173-8.042", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_457.wav", "onoffCaption": "explosion at 1.716-4.903 and door knocking at 3.096-5.16, 6.335-8.399", "frequencyCaption": "explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_556.wav", "onoffCaption": "sneeze at 0.228-3.436, 4.412-5.658 and door slamming at 1.53-4.504, 6.22-9.194", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_624.wav", "onoffCaption": "sneeze at 0.49-2.449 and thump thud at 5.007-9.457", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_848.wav", "onoffCaption": "thump thud at 3.74-6.24, 7.599-9.797", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_852.wav", "onoffCaption": "tapping clicking clanking at 1.814-5.254, 6.511-9.021", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_949.wav", "onoffCaption": "spraying at 2.482-4.215, 6.71-7.794, 8.844-9.445", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_953.wav", "onoffCaption": "car horn honking at 3.959-6.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1066.wav", "onoffCaption": "explosion at 2.375-4.695, 5.778-8.098 and woman laughing at 4.233-6.349, 7.227-9.343", "frequencyCaption": "explosion two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1083.wav", "onoffCaption": "woman laughing at 2.835-5.043, 7.399-9.607", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1099.wav", "onoffCaption": "burping belching at 3.336-6.515", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1167.wav", "onoffCaption": "burping belching at 0.344-3.344, 5.134-7.499", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1182.wav", "onoffCaption": "spraying at 1.885-4.277, 5.008-5.508, 7.252-9.014", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1198.wav", "onoffCaption": "door knocking at 1.186-6.019, 7.123-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1215.wav", "onoffCaption": "tapping clicking clanking at 0.347-3.787, 5.034-8.474", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1314.wav", "onoffCaption": "thump thud at 3.539-7.989", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1430.wav", "onoffCaption": "tapping clicking clanking at 2.991-6.431, 7.373-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1531.wav", "onoffCaption": "cat meowing at 0.302-2.296, 3.126-5.12 and car horn honking at 1.986-5.481", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1643.wav", "onoffCaption": "thump thud at 2.695-7.07", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1659.wav", "onoffCaption": "explosion at 3.679-6.551, 7.448-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1742.wav", "onoffCaption": "tapping clicking clanking at 0.157-3.597, 5.964-8.192 and gunshot at 2.244-4.374, 4.978-6.978, 7.529-9.699", "frequencyCaption": "tapping clicking clanking two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1758.wav", "onoffCaption": "burping belching at 0.668-3.191, 4.416-7.208 and car horn honking at 1.278-4.191", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1835.wav", "onoffCaption": "woman laughing at 2.231-4.869, 6.638-9.276", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1934.wav", "onoffCaption": "thump thud at 0.624-4.291 and explosion at 7.861-10.0", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3004.wav", "onoffCaption": "duck quacking at 0.12-2.12, 3.932-5.932 and door slamming at 8.291-9.191", "frequencyCaption": "duck quacking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3105.wav", "onoffCaption": "thump thud at 2.398-4.898, 6.624-9.124", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3277.wav", "onoffCaption": "woman laughing at 0.86-3.143, 4.265-6.548, 7.463-9.746", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3288.wav", "onoffCaption": "sneeze at 0.323-1.415, 2.705-3.797, 5.231-6.323 and train horn at 2.628-5.068", "frequencyCaption": "sneeze three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3292.wav", "onoffCaption": "burping belching at 1.879-5.879", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3376.wav", "onoffCaption": "train horn at 0.694-4.094, 5.406-8.061", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3389.wav", "onoffCaption": "thump thud at 0.609-5.059", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3393.wav", "onoffCaption": "door slamming at 0.336-1.509", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3448.wav", "onoffCaption": "gunshot at 2.818-4.818 and door knocking at 4.83-6.99, 7.814-9.974", "frequencyCaption": "gunshot one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3452.wav", "onoffCaption": "spraying at 3.48-4.002, 5.912-6.434", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3549.wav", "onoffCaption": "cow mooing at 1.512-4.522 and sheep goat bleating at 6.085-9.381", "frequencyCaption": "cow mooing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3553.wav", "onoffCaption": "explosion at 0.046-3.102, 4.365-7.421 and cow mooing at 1.437-6.417", "frequencyCaption": "explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3621.wav", "onoffCaption": "duck quacking at 0.251-2.251, 3.603-5.603 and door slamming at 3.149-4.914, 5.503-6.503", "frequencyCaption": "duck quacking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3720.wav", "onoffCaption": "gunshot at 0.119-2.119, 3.389-5.389, 6.323-8.323", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3857.wav", "onoffCaption": "car horn honking at 1.094-4.02, 4.596-7.522", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3956.wav", "onoffCaption": "car horn honking at 3.505-8.412", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_27.wav", "onoffCaption": "sheep goat bleating at 3.479-5.479, 6.791-8.791 and cow mooing at 3.583-8.012", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_56.wav", "onoffCaption": "duck quacking at 0.787-2.787 and cat meowing at 1.168-2.18, 3.484-4.496", "frequencyCaption": "duck quacking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_124.wav", "onoffCaption": "explosion at 0.32-2.413 and dog barking at 4.449-6.449, 7.727-9.727 and tapping clicking clanking at 4.578-8.018", "frequencyCaption": "explosion one times and dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_155.wav", "onoffCaption": "sheep goat bleating at 3.874-6.954", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_268.wav", "onoffCaption": "burping belching at 2.076-7.076, 7.833-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_326.wav", "onoffCaption": "sneeze at 0.217-1.32 and duck quacking at 3.736-5.736", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_373.wav", "onoffCaption": "door knocking at 2.926-5.694, 6.774-9.493", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_418.wav", "onoffCaption": "train horn at 0.085-3.619", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_469.wav", "onoffCaption": "train horn at 2.048-5.808, 6.803-9.524", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_503.wav", "onoffCaption": "door slamming at 2.492-4.975 and explosion at 2.856-5.117, 6.1-8.361", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_597.wav", "onoffCaption": "dog barking at 2.154-4.154 and car horn honking at 6.717-9.643", "frequencyCaption": "dog barking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_725.wav", "onoffCaption": "car horn honking at 0.486-3.661, 4.479-6.592", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_770.wav", "onoffCaption": "explosion at 0.494-2.668, 3.329-5.503 and gunshot at 2.729-4.859 and spraying at 9.35-10.0", "frequencyCaption": "explosion two times and gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_795.wav", "onoffCaption": "woman laughing at 0.279-2.36, 4.524-6.605, 7.54-9.621", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_807.wav", "onoffCaption": "dog barking at 0.412-2.412 and sneeze at 3.819-5.347, 6.581-7.967", "frequencyCaption": "dog barking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_876.wav", "onoffCaption": "whistling at 2.475-4.704, 5.69-7.699 and train horn at 6.152-8.952", "frequencyCaption": "whistling two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_893.wav", "onoffCaption": "explosion at 0.254-3.254 and sneeze at 5.264-7.327", "frequencyCaption": "explosion one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_988.wav", "onoffCaption": "sheep goat bleating at 1.533-3.533 and cat meowing at 6.128-7.488", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1033.wav", "onoffCaption": "gunshot at 1.931-4.437", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1042.wav", "onoffCaption": "whistling at 1.292-9.042", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1128.wav", "onoffCaption": "sheep goat bleating at 1.761-3.761, 5.687-7.687", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1159.wav", "onoffCaption": "burping belching at 0.102-2.517 and door slamming at 0.169-3.048", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1240.wav", "onoffCaption": "car horn honking at 3.031-6.618 and spraying at 7.178-8.435", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1414.wav", "onoffCaption": "car horn honking at 2.864-5.79, 7.277-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1465.wav", "onoffCaption": "door knocking at 0.143-3.21, 4.674-7.054", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1480.wav", "onoffCaption": "train horn at 3.285-6.445", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1616.wav", "onoffCaption": "sheep goat bleating at 1.724-3.724, 4.694-6.694, 7.757-9.757", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1910.wav", "onoffCaption": "gunshot at 3.62-5.62, 7.01-9.01", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1961.wav", "onoffCaption": "explosion at 0.83-5.83 and car horn honking at 2.125-6.525, 7.843-10.0", "frequencyCaption": "explosion one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1984.wav", "onoffCaption": "explosion at 0.271-3.271 and spraying at 0.606-3.042, 4.649-7.085 and gunshot at 4.573-6.592", "frequencyCaption": "explosion one times and spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3020.wav", "onoffCaption": "thump thud at 0.274-2.613, 4.736-7.075 and door slamming at 1.616-2.421, 4.032-5.335 and cat meowing at 4.18-7.421", "frequencyCaption": "thump thud two times and door slamming two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3051.wav", "onoffCaption": "explosion at 0.583-5.583, 7.425-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3222.wav", "onoffCaption": "whistling at 2.234-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3339.wav", "onoffCaption": "sneeze at 3.7-5.659 and car horn honking at 7.905-10.0", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3407.wav", "onoffCaption": "cat meowing at 0.704-1.716, 3.124-5.064 and spraying at 1.93-4.322 and woman laughing at 7.89-10.0", "frequencyCaption": "cat meowing two times and spraying one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3476.wav", "onoffCaption": "sneeze at 0.2-1.294, 2.325-4.449", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3493.wav", "onoffCaption": "explosion at 2.754-5.015", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3588.wav", "onoffCaption": "door knocking at 3.157-5.727, 7.114-9.684", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3674.wav", "onoffCaption": "burping belching at 0.178-3.738, 4.276-7.836 and duck quacking at 7.421-9.421", "frequencyCaption": "burping belching two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3691.wav", "onoffCaption": "train horn at 2.007-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3818.wav", "onoffCaption": "train horn at 0.086-2.966, 5.129-7.598", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3869.wav", "onoffCaption": "spraying at 0.068-1.152, 3.159-3.734", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3903.wav", "onoffCaption": "gunshot at 0.325-2.325, 3.275-5.275, 6.343-8.343", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3972.wav", "onoffCaption": "explosion at 2.923-4.93 and cat meowing at 7.772-9.308", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3997.wav", "onoffCaption": "spraying at 0.219-1.086, 2.175-3.042 and gunshot at 1.073-3.073, 4.828-6.828", "frequencyCaption": "spraying two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_171.wav", "onoffCaption": "sheep goat bleating at 0.622-2.622, 3.342-5.342 and explosion at 0.656-5.577, 6.689-10.0", "frequencyCaption": "sheep goat bleating two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_194.wav", "onoffCaption": "cow mooing at 4.267-9.247", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_203.wav", "onoffCaption": "spraying at 0.497-2.192, 2.836-3.777, 4.996-5.847", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_219.wav", "onoffCaption": "burping belching at 0.188-3.188 and door knocking at 0.709-4.471", "frequencyCaption": "burping belching one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_302.wav", "onoffCaption": "spraying at 2.0-2.75", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_318.wav", "onoffCaption": "door knocking at 3.232-6.079, 7.926-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_357.wav", "onoffCaption": "car horn honking at 0.444-3.939", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_527.wav", "onoffCaption": "explosion at 0.505-3.373 and tapping clicking clanking at 6.208-9.648", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_572.wav", "onoffCaption": "whistling at 1.684-3.693, 5.499-7.606", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_701.wav", "onoffCaption": "door slamming at 0.027-0.967, 2.312-3.252 and sneeze at 6.756-8.256", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_754.wav", "onoffCaption": "dog barking at 2.725-4.725, 5.904-7.904", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_823.wav", "onoffCaption": "whistling at 0.976-6.151, 7.109-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_938.wav", "onoffCaption": "cat meowing at 0.419-3.323, 4.149-6.126, 7.37-9.364", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1017.wav", "onoffCaption": "sneeze at 1.526-3.765 and dog barking at 3.672-5.672, 6.695-8.695", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1231.wav", "onoffCaption": "door knocking at 0.655-3.03, 4.154-6.529, 7.337-9.712", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1264.wav", "onoffCaption": "sneeze at 2.384-3.912, 4.476-5.802, 6.605-9.066", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1281.wav", "onoffCaption": "burping belching at 0.058-5.058, 7.551-9.892 and door slamming at 1.441-3.637", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1365.wav", "onoffCaption": "train horn at 3.067-9.536", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1380.wav", "onoffCaption": "thump thud at 1.007-3.507, 5.962-8.301", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1441.wav", "onoffCaption": "thump thud at 1.625-4.125, 5.821-7.956", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1628.wav", "onoffCaption": "spraying at 1.654-3.387", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1632.wav", "onoffCaption": "explosion at 0.414-3.142", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1667.wav", "onoffCaption": "train horn at 0.181-5.258, 7.479-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1682.wav", "onoffCaption": "train horn at 3.047-5.202, 5.893-8.048", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1729.wav", "onoffCaption": "burping belching at 2.765-4.968, 6.119-8.322", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1733.wav", "onoffCaption": "spraying at 2.841-4.098, 6.577-8.339", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1799.wav", "onoffCaption": "cat meowing at 0.443-2.027, 4.267-5.851", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1945.wav", "onoffCaption": "dog barking at 2.352-4.352 and gunshot at 7.525-9.525", "frequencyCaption": "dog barking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3075.wav", "onoffCaption": "gunshot at 0.101-2.101", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3090.wav", "onoffCaption": "sheep goat bleating at 0.67-2.67, 3.618-5.618 and gunshot at 2.733-4.752, 6.302-8.321", "frequencyCaption": "sheep goat bleating two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3206.wav", "onoffCaption": "car horn honking at 0.644-3.109, 5.341-7.806", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3253.wav", "onoffCaption": "cat meowing at 0.657-2.017", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3307.wav", "onoffCaption": "burping belching at 2.496-6.04, 6.739-9.063 and woman laughing at 3.176-5.768", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3348.wav", "onoffCaption": "sheep goat bleating at 1.975-3.975 and burping belching at 2.183-4.944, 5.965-8.088", "frequencyCaption": "sheep goat bleating one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3423.wav", "onoffCaption": "thump thud at 2.28-4.508, 6.92-9.148", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3538.wav", "onoffCaption": "duck quacking at 2.278-4.278, 5.652-7.652", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3605.wav", "onoffCaption": "dog barking at 0.121-2.121", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3650.wav", "onoffCaption": "whistling at 0.406-2.415, 3.048-5.057, 6.24-8.249 and spraying at 4.176-4.826, 5.812-6.876", "frequencyCaption": "whistling three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3751.wav", "onoffCaption": "gunshot at 0.272-2.272, 4.238-6.331, 7.459-9.459", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3927.wav", "onoffCaption": "door slamming at 0.028-2.028", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_19.wav", "onoffCaption": "gunshot at 2.917-4.917", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_256.wav", "onoffCaption": "sheep goat bleating at 1.495-3.495, 4.567-6.567, 7.123-9.123", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_426.wav", "onoffCaption": "woman laughing at 2.649-4.932, 5.692-7.975", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_432.wav", "onoffCaption": "door knocking at 0.055-3.43 and spraying at 7.599-8.2", "frequencyCaption": "door knocking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_473.wav", "onoffCaption": "gunshot at 0.698-2.698, 3.947-5.947, 6.742-8.742 and explosion at 4.807-7.863", "frequencyCaption": "gunshot three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_496.wav", "onoffCaption": "car horn honking at 2.094-5.313, 5.949-9.168", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_529.wav", "onoffCaption": "door knocking at 3.432-5.592", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_568.wav", "onoffCaption": "dog barking at 0.403-2.403, 4.637-6.637", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_600.wav", "onoffCaption": "train horn at 2.315-8.03", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_641.wav", "onoffCaption": "cow mooing at 0.533-3.515, 4.376-7.358 and duck quacking at 3.39-5.39 and whistling at 5.257-8.232", "frequencyCaption": "cow mooing two times and duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_655.wav", "onoffCaption": "spraying at 0.001-0.852", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_839.wav", "onoffCaption": "door slamming at 2.325-3.442, 4.75-5.867, 6.927-8.044 and sneeze at 3.071-4.359, 4.959-6.247, 7.169-8.457", "frequencyCaption": "door slamming three times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_889.wav", "onoffCaption": "sheep goat bleating at 3.745-5.745, 7.04-9.04", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_922.wav", "onoffCaption": "door knocking at 2.749-5.596, 6.697-9.593", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_936.wav", "onoffCaption": "gunshot at 2.615-4.656, 5.659-7.7", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_977.wav", "onoffCaption": "car horn honking at 0.024-3.199, 4.429-7.604 and door slamming at 1.132-1.983, 2.772-3.623, 5.279-6.13 and spraying at 3.137-3.918, 5.76-6.541", "frequencyCaption": "car horn honking two times and door slamming three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_992.wav", "onoffCaption": "train horn at 0.88-6.595, 7.622-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1019.wav", "onoffCaption": "door slamming at 0.69-3.664 and duck quacking at 7.922-9.922", "frequencyCaption": "door slamming one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1058.wav", "onoffCaption": "door slamming at 0.388-2.388, 3.626-5.626 and explosion at 2.533-5.286, 6.989-9.497", "frequencyCaption": "door slamming two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1102.wav", "onoffCaption": "gunshot at 3.478-5.608, 6.53-8.77", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1116.wav", "onoffCaption": "door slamming at 2.112-4.24, 5.163-7.291", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1143.wav", "onoffCaption": "sneeze at 2.226-4.185, 5.315-7.274, 7.89-9.849", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1330.wav", "onoffCaption": "burping belching at 3.154-6.154", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1515.wav", "onoffCaption": "spraying at 0.068-0.669 and woman laughing at 2.812-4.904, 6.245-8.327", "frequencyCaption": "spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1540.wav", "onoffCaption": "sheep goat bleating at 1.643-3.643, 4.39-6.39, 7.124-9.124", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1554.wav", "onoffCaption": "cat meowing at 2.76-3.781 and door slamming at 5.94-7.079", "frequencyCaption": "cat meowing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1698.wav", "onoffCaption": "spraying at 3.088-5.216", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1766.wav", "onoffCaption": "dog barking at 2.815-4.815, 5.782-8.22", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1783.wav", "onoffCaption": "door knocking at 2.091-5.644, 6.754-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1811.wav", "onoffCaption": "duck quacking at 0.315-2.315, 3.397-5.397 and dog barking at 0.403-2.403, 3.261-5.261, 6.58-8.58", "frequencyCaption": "duck quacking two times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1844.wav", "onoffCaption": "train horn at 2.463-6.903", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1850.wav", "onoffCaption": "sheep goat bleating at 0.308-2.308, 3.201-5.201, 7.564-10.0", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3121.wav", "onoffCaption": "burping belching at 0.675-4.698, 6.615-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3160.wav", "onoffCaption": "burping belching at 0.221-2.344, 4.641-6.672, 7.976-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3174.wav", "onoffCaption": "cow mooing at 0.128-3.138, 4.765-7.734 and gunshot at 0.227-2.227, 2.757-4.85, 6.127-8.127 and tapping clicking clanking at 5.439-8.879", "frequencyCaption": "cow mooing two times and gunshot three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3185.wav", "onoffCaption": "gunshot at 1.368-3.368, 4.358-6.358", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3191.wav", "onoffCaption": "explosion at 3.122-6.122", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3249.wav", "onoffCaption": "car horn honking at 0.422-2.887, 3.961-6.426", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3352.wav", "onoffCaption": "dog barking at 0.73-4.05, 6.487-9.807", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3439.wav", "onoffCaption": "sneeze at 0.638-5.638, 6.306-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3489.wav", "onoffCaption": "burping belching at 0.191-2.952 and car horn honking at 6.918-10.0", "frequencyCaption": "burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3522.wav", "onoffCaption": "tapping clicking clanking at 2.936-6.376 and dog barking at 3.613-5.613 and spraying at 4.705-6.467, 7.598-9.293", "frequencyCaption": "tapping clicking clanking one times and dog barking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3536.wav", "onoffCaption": "gunshot at 1.616-3.616, 5.17-7.17", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3577.wav", "onoffCaption": "car horn honking at 0.724-5.631, 6.312-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3592.wav", "onoffCaption": "sheep goat bleating at 0.138-3.218, 3.865-5.865 and car horn honking at 2.443-6.765", "frequencyCaption": "sheep goat bleating two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3704.wav", "onoffCaption": "sneeze at 1.924-3.027, 4.925-6.028 and cow mooing at 7.95-10.0", "frequencyCaption": "sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3826.wav", "onoffCaption": "spraying at 0.306-2.001 and cat meowing at 6.478-10.0", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3832.wav", "onoffCaption": "burping belching at 0.032-3.032 and explosion at 5.43-10.0", "frequencyCaption": "burping belching one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3873.wav", "onoffCaption": "thump thud at 0.405-4.855, 5.664-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3896.wav", "onoffCaption": "gunshot at 0.419-2.419, 3.141-5.141, 6.145-8.145 and burping belching at 1.956-5.979", "frequencyCaption": "gunshot three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3929.wav", "onoffCaption": "burping belching at 2.342-4.377, 5.498-7.533", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3968.wav", "onoffCaption": "thump thud at 1.08-5.53, 7.974-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_58.wav", "onoffCaption": "tapping clicking clanking at 2.24-5.68, 6.942-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_217.wav", "onoffCaption": "duck quacking at 1.061-3.061, 4.757-6.757, 7.312-9.312", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_242.wav", "onoffCaption": "woman laughing at 0.008-2.233, 3.414-5.519, 7.665-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_258.wav", "onoffCaption": "door knocking at 2.89-6.506, 7.322-9.482", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_343.wav", "onoffCaption": "duck quacking at 3.426-5.426, 6.592-8.592", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_359.wav", "onoffCaption": "cow mooing at 0.104-3.086 and woman laughing at 5.605-10.0", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_467.wav", "onoffCaption": "thump thud at 1.125-3.896, 5.7-8.471", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_482.wav", "onoffCaption": "door knocking at 2.099-4.723, 6.801-9.425", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_566.wav", "onoffCaption": "dog barking at 2.388-4.388, 6.567-8.567", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_599.wav", "onoffCaption": "whistling at 0.26-5.435, 6.002-8.011", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_614.wav", "onoffCaption": "woman laughing at 0.863-2.968", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_715.wav", "onoffCaption": "woman laughing at 2.516-6.568", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_878.wav", "onoffCaption": "spraying at 0.47-0.992 and whistling at 2.956-5.831", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_963.wav", "onoffCaption": "cow mooing at 0.407-3.705, 4.406-7.704", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_986.wav", "onoffCaption": "duck quacking at 2.219-4.219, 5.442-7.442", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1157.wav", "onoffCaption": "car horn honking at 1.378-4.553", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1225.wav", "onoffCaption": "train horn at 2.208-5.978", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1324.wav", "onoffCaption": "duck quacking at 0.456-2.456, 3.38-5.38, 6.356-8.356 and thump thud at 2.341-4.68, 5.337-7.676", "frequencyCaption": "duck quacking three times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1371.wav", "onoffCaption": "gunshot at 0.281-2.411 and whistling at 4.833-9.317", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1394.wav", "onoffCaption": "sheep goat bleating at 0.453-2.453, 3.709-5.709", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1501.wav", "onoffCaption": "door slamming at 3.171-5.629", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1669.wav", "onoffCaption": "door slamming at 0.076-0.881, 1.714-2.519 and car horn honking at 5.815-8.28", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1673.wav", "onoffCaption": "train horn at 2.681-5.881, 7.622-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1696.wav", "onoffCaption": "woman laughing at 2.735-9.469", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1727.wav", "onoffCaption": "woman laughing at 2.101-4.295, 5.621-7.848 and whistling at 2.263-5.138", "frequencyCaption": "woman laughing two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1768.wav", "onoffCaption": "train horn at 0.389-2.789", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1772.wav", "onoffCaption": "cow mooing at 3.636-6.934", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1797.wav", "onoffCaption": "door slamming at 0.328-2.456, 3.815-5.943", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1805.wav", "onoffCaption": "gunshot at 3.296-5.296", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3135.wav", "onoffCaption": "car horn honking at 2.114-6.436, 7.636-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3208.wav", "onoffCaption": "door slamming at 3.143-4.121, 5.763-6.741, 7.247-8.225", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3247.wav", "onoffCaption": "spraying at 1.987-2.487, 4.039-4.539 and door slamming at 7.66-9.425", "frequencyCaption": "spraying two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3313.wav", "onoffCaption": "cow mooing at 0.136-4.565, 6.674-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3346.wav", "onoffCaption": "tapping clicking clanking at 2.996-6.436, 7.718-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3478.wav", "onoffCaption": "door slamming at 2.783-5.757, 6.792-8.988", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3563.wav", "onoffCaption": "explosion at 0.035-2.763, 3.269-5.997", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3586.wav", "onoffCaption": "dog barking at 2.001-4.001, 6.436-8.436", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3611.wav", "onoffCaption": "burping belching at 0.545-3.545", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3710.wav", "onoffCaption": "sheep goat bleating at 0.418-2.418, 3.248-5.248, 6.327-8.327", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3745.wav", "onoffCaption": "whistling at 2.291-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3867.wav", "onoffCaption": "burping belching at 1.62-4.381, 6.604-9.365", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3882.wav", "onoffCaption": "woman laughing at 0.645-3.699 and sneeze at 6.075-7.167, 7.708-8.8", "frequencyCaption": "woman laughing one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3966.wav", "onoffCaption": "thump thud at 0.699-3.746, 6.205-9.122", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3999.wav", "onoffCaption": "tapping clicking clanking at 2.904-6.344, 7.265-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_17.wav", "onoffCaption": "cow mooing at 0.467-5.447, 6.568-8.92", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_42.wav", "onoffCaption": "sheep goat bleating at 3.727-5.727", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_130.wav", "onoffCaption": "thump thud at 0.503-3.274, 5.351-8.122 and cow mooing at 2.094-6.523", "frequencyCaption": "thump thud two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_165.wav", "onoffCaption": "explosion at 0.033-3.033, 4.03-6.902", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_180.wav", "onoffCaption": "door knocking at 1.887-4.199, 5.442-7.754", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_316.wav", "onoffCaption": "cow mooing at 0.186-4.615 and gunshot at 6.965-8.965", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_498.wav", "onoffCaption": "woman laughing at 0.389-3.184 and spraying at 0.479-1.736 and sheep goat bleating at 6.814-8.814", "frequencyCaption": "woman laughing one times and spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_583.wav", "onoffCaption": "cat meowing at 2.992-4.609, 5.325-6.942, 8.008-9.625", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_740.wav", "onoffCaption": "woman laughing at 0.233-2.933, 4.001-6.701, 7.57-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_837.wav", "onoffCaption": "tapping clicking clanking at 1.154-4.594", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_862.wav", "onoffCaption": "cow mooing at 0.176-3.158", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_887.wav", "onoffCaption": "gunshot at 1.975-3.975 and explosion at 6.851-8.853", "frequencyCaption": "gunshot one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_979.wav", "onoffCaption": "cow mooing at 0.418-5.398, 6.857-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1003.wav", "onoffCaption": "door knocking at 2.873-7.575", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1056.wav", "onoffCaption": "spraying at 0.772-3.791, 4.499-5.126, 6.351-7.432 and door knocking at 2.666-6.816", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1118.wav", "onoffCaption": "sneeze at 1.98-4.219, 5.391-8.105", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1270.wav", "onoffCaption": "explosion at 0.304-3.057, 5.349-7.61", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1295.wav", "onoffCaption": "tapping clicking clanking at 0.66-4.1, 5.72-8.425", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1400.wav", "onoffCaption": "cat meowing at 0.025-1.046, 1.59-4.35", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1455.wav", "onoffCaption": "tapping clicking clanking at 0.635-4.075, 5.126-8.566 and car horn honking at 1.938-6.338", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1626.wav", "onoffCaption": "tapping clicking clanking at 2.667-6.107, 7.915-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1904.wav", "onoffCaption": "gunshot at 0.344-2.344, 3.876-5.876", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1951.wav", "onoffCaption": "thump thud at 0.001-2.501, 3.395-5.623", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3034.wav", "onoffCaption": "thump thud at 2.608-5.108, 6.01-8.51", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3061.wav", "onoffCaption": "cow mooing at 1.872-4.841", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3084.wav", "onoffCaption": "whistling at 0.882-3.757, 4.534-6.763 and cat meowing at 1.107-2.651, 3.757-5.023", "frequencyCaption": "whistling two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3212.wav", "onoffCaption": "explosion at 0.067-2.131", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3309.wav", "onoffCaption": "train horn at 0.142-4.142 and duck quacking at 1.183-3.183, 4.552-6.552 and door knocking at 2.065-4.377, 5.166-7.478", "frequencyCaption": "train horn one times and duck quacking two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3437.wav", "onoffCaption": "train horn at 3.389-7.457", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3462.wav", "onoffCaption": "cat meowing at 1.262-4.503, 6.9-10.0 and spraying at 1.386-3.846 and duck quacking at 2.352-4.352", "frequencyCaption": "cat meowing two times and spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3487.wav", "onoffCaption": "cow mooing at 0.381-3.391, 4.246-7.074 and car horn honking at 2.377-6.889", "frequencyCaption": "cow mooing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3579.wav", "onoffCaption": "door knocking at 0.011-4.544, 5.239-9.772", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3644.wav", "onoffCaption": "dog barking at 1.946-3.946, 4.723-6.723 and sneeze at 4.503-6.906, 7.706-10.0", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3898.wav", "onoffCaption": "explosion at 0.238-3.791", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3933.wav", "onoffCaption": "door slamming at 0.57-3.449, 4.57-6.085, 6.734-7.572", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3983.wav", "onoffCaption": "sneeze at 1.528-3.652, 4.23-5.775, 6.779-9.439", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_66.wav", "onoffCaption": "sheep goat bleating at 3.755-5.755, 6.54-8.54", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_83.wav", "onoffCaption": "burping belching at 1.755-4.957, 6.169-8.93", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_114.wav", "onoffCaption": "thump thud at 0.273-3.32, 4.785-7.832 and whistling at 2.289-7.464", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_299.wav", "onoffCaption": "tapping clicking clanking at 2.854-6.294", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_367.wav", "onoffCaption": "burping belching at 2.702-5.992", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_382.wav", "onoffCaption": "tapping clicking clanking at 3.22-6.66, 7.835-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_428.wav", "onoffCaption": "door knocking at 3.238-5.426, 6.557-8.745", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_459.wav", "onoffCaption": "door slamming at 0.017-1.27 and train horn at 0.31-4.31, 5.293-7.405 and duck quacking at 2.271-4.271, 5.79-7.79", "frequencyCaption": "door slamming one times and train horn two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_533.wav", "onoffCaption": "train horn at 2.06-4.527, 5.17-7.637", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_542.wav", "onoffCaption": "door knocking at 2.58-4.932, 5.435-8.165", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_731.wav", "onoffCaption": "cow mooing at 1.42-4.718", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_781.wav", "onoffCaption": "cat meowing at 2.711-3.924, 4.973-6.186, 7.019-8.232", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_846.wav", "onoffCaption": "cow mooing at 2.238-6.667, 7.302-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1072.wav", "onoffCaption": "train horn at 1.178-7.238", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1097.wav", "onoffCaption": "whistling at 0.339-4.823, 6.015-8.024", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1169.wav", "onoffCaption": "sneeze at 0.092-1.186, 1.806-2.9, 3.554-4.648 and cat meowing at 6.817-8.765", "frequencyCaption": "sneeze three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1201.wav", "onoffCaption": "train horn at 2.979-5.116, 5.771-7.908", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1254.wav", "onoffCaption": "sheep goat bleating at 0.196-2.196, 3.225-5.225 and cow mooing at 7.435-10.0", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1424.wav", "onoffCaption": "gunshot at 0.785-2.785, 3.691-5.691, 6.961-8.961", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1602.wav", "onoffCaption": "explosion at 2.752-4.926, 6.126-8.3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1657.wav", "onoffCaption": "burping belching at 1.032-4.211 and spraying at 7.591-9.077", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1719.wav", "onoffCaption": "car horn honking at 2.991-5.917, 7.165-9.757", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1920.wav", "onoffCaption": "cat meowing at 2.121-5.151, 6.69-8.226", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3010.wav", "onoffCaption": "car horn honking at 0.242-3.896, 5.279-8.933", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3236.wav", "onoffCaption": "burping belching at 1.875-5.875", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3263.wav", "onoffCaption": "dog barking at 3.569-5.569, 6.486-8.486 and thump thud at 4.01-7.928", "frequencyCaption": "dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3286.wav", "onoffCaption": "gunshot at 3.647-5.887, 7.673-9.913", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3378.wav", "onoffCaption": "door knocking at 0.404-3.172, 5.052-7.899 and tapping clicking clanking at 3.784-7.224 and car horn honking at 6.433-8.433", "frequencyCaption": "door knocking two times and tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3446.wav", "onoffCaption": "sheep goat bleating at 0.347-3.987, 5.155-8.795", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3635.wav", "onoffCaption": "cat meowing at 0.466-2.01 and duck quacking at 5.427-7.427", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3660.wav", "onoffCaption": "tapping clicking clanking at 2.607-6.047, 6.693-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3685.wav", "onoffCaption": "gunshot at 3.213-5.719", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3828.wav", "onoffCaption": "dog barking at 3.288-5.288 and burping belching at 3.369-5.572, 7.868-10.0", "frequencyCaption": "dog barking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3859.wav", "onoffCaption": "spraying at 0.065-1.065, 2.246-3.246, 5.682-6.682", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3942.wav", "onoffCaption": "door knocking at 0.673-2.922, 4.302-6.551 and cow mooing at 3.207-6.189", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_29.wav", "onoffCaption": "tapping clicking clanking at 2.876-6.316, 7.077-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_33.wav", "onoffCaption": "sheep goat bleating at 0.232-2.232, 3.068-5.068, 5.691-7.691", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_141.wav", "onoffCaption": "dog barking at 2.679-4.679, 5.988-7.988", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_229.wav", "onoffCaption": "cow mooing at 0.504-3.802 and sheep goat bleating at 6.873-10.0", "frequencyCaption": "cow mooing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_332.wav", "onoffCaption": "explosion at 2.929-5.658, 6.447-9.447", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_416.wav", "onoffCaption": "cat meowing at 1.978-3.873, 4.787-6.682", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_517.wav", "onoffCaption": "woman laughing at 0.043-7.055 and sheep goat bleating at 1.034-3.034, 4.335-6.335", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_665.wav", "onoffCaption": "thump thud at 0.333-3.104, 4.26-6.811", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_764.wav", "onoffCaption": "explosion at 2.603-5.332", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_809.wav", "onoffCaption": "tapping clicking clanking at 0.597-4.037, 5.572-8.29", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_813.wav", "onoffCaption": "sneeze at 0.338-1.664, 3.859-6.726, 7.402-8.902", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_908.wav", "onoffCaption": "duck quacking at 2.231-4.231, 5.703-7.703", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_912.wav", "onoffCaption": "whistling at 0.37-5.87", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1027.wav", "onoffCaption": "burping belching at 0.405-2.928 and spraying at 7.113-8.808", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1126.wav", "onoffCaption": "cat meowing at 2.066-3.798, 4.339-6.071, 6.707-8.439", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1471.wav", "onoffCaption": "whistling at 2.484-5.459, 7.427-9.449", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1494.wav", "onoffCaption": "whistling at 0.825-3.7, 4.527-7.523", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1570.wav", "onoffCaption": "cow mooing at 0.068-3.078 and dog barking at 5.42-7.858", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1595.wav", "onoffCaption": "door slamming at 1.444-4.444, 6.63-9.63", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1874.wav", "onoffCaption": "gunshot at 0.023-2.023", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1891.wav", "onoffCaption": "burping belching at 0.954-3.157 and tapping clicking clanking at 1.277-4.717, 6.507-9.947", "frequencyCaption": "burping belching one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1975.wav", "onoffCaption": "explosion at 1.831-4.831, 5.888-8.888 and spraying at 2.557-3.307, 4.647-5.514 and tapping clicking clanking at 4.125-7.565", "frequencyCaption": "explosion two times and spraying two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1990.wav", "onoffCaption": "door slamming at 0.013-1.186, 1.716-2.889, 3.808-4.981", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3045.wav", "onoffCaption": "explosion at 2.17-4.49 and train horn at 6.537-10.0", "frequencyCaption": "explosion one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3144.wav", "onoffCaption": "gunshot at 0.205-2.205, 4.257-6.257", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3409.wav", "onoffCaption": "door knocking at 0.161-2.382, 3.18-5.984", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3413.wav", "onoffCaption": "explosion at 0.773-3.773, 4.786-7.786 and duck quacking at 1.458-3.458, 5.083-7.083", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3508.wav", "onoffCaption": "cow mooing at 2.491-5.789, 6.803-8.835", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3512.wav", "onoffCaption": "explosion at 2.04-7.04 and thump thud at 2.667-7.042, 7.744-10.0", "frequencyCaption": "explosion one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3816.wav", "onoffCaption": "train horn at 3.343-6.143, 7.821-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3917.wav", "onoffCaption": "sneeze at 2.112-4.573", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_99.wav", "onoffCaption": "cow mooing at 0.562-3.86, 4.984-8.282", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_233.wav", "onoffCaption": "dog barking at 0.177-2.177 and woman laughing at 0.336-3.036, 4.006-6.644, 7.419-9.665", "frequencyCaption": "dog barking one times and woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_266.wav", "onoffCaption": "cow mooing at 1.671-4.681, 6.283-9.293 and woman laughing at 2.644-5.236, 5.829-7.91", "frequencyCaption": "cow mooing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_283.wav", "onoffCaption": "whistling at 0.42-3.295, 3.993-6.222, 7.259-9.502", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_328.wav", "onoffCaption": "whistling at 1.879-7.054 and tapping clicking clanking at 3.661-7.101, 7.666-9.803", "frequencyCaption": "whistling one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_398.wav", "onoffCaption": "sneeze at 2.719-4.832, 6.205-8.318", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_443.wav", "onoffCaption": "explosion at 0.593-5.593, 7.99-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_558.wav", "onoffCaption": "sheep goat bleating at 2.197-4.197, 4.74-6.74, 7.47-10.0 and door knocking at 2.662-6.215", "frequencyCaption": "sheep goat bleating three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_630.wav", "onoffCaption": "thump thud at 2.609-5.109, 6.484-8.984", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_662.wav", "onoffCaption": "gunshot at 1.884-3.884, 6.184-8.184", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_680.wav", "onoffCaption": "thump thud at 1.832-5.499, 6.258-9.925", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_779.wav", "onoffCaption": "thump thud at 2.493-4.832, 7.066-9.405", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_947.wav", "onoffCaption": "train horn at 0.341-2.821, 3.65-6.05", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1068.wav", "onoffCaption": "car horn honking at 0.181-4.43, 6.693-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1173.wav", "onoffCaption": "explosion at 0.221-2.314, 3.781-5.899", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1196.wav", "onoffCaption": "sheep goat bleating at 1.568-3.568, 4.446-6.446", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1300.wav", "onoffCaption": "sneeze at 0.212-1.458, 3.742-4.988 and dog barking at 1.234-3.234", "frequencyCaption": "sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1355.wav", "onoffCaption": "cat meowing at 2.938-4.886, 6.198-8.146", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1525.wav", "onoffCaption": "train horn at 0.469-3.109, 4.904-7.77", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1618.wav", "onoffCaption": "cat meowing at 3.514-4.535, 6.027-7.24, 8.265-9.575", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1703.wav", "onoffCaption": "burping belching at 2.136-7.737 and sheep goat bleating at 4.673-6.673, 7.794-9.794", "frequencyCaption": "burping belching one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1756.wav", "onoffCaption": "car horn honking at 1.746-4.211, 5.004-7.53", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1821.wav", "onoffCaption": "door knocking at 0.277-4.979, 6.493-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3111.wav", "onoffCaption": "thump thud at 1.355-5.805 and door slamming at 1.839-3.839, 4.949-6.862, 8.179-9.328", "frequencyCaption": "thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3279.wav", "onoffCaption": "dog barking at 0.184-3.105 and door knocking at 0.409-2.63 and cat meowing at 5.215-6.236", "frequencyCaption": "dog barking one times and door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3337.wav", "onoffCaption": "dog barking at 0.337-2.337", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3362.wav", "onoffCaption": "train horn at 0.443-2.58, 3.154-5.291, 7.352-9.489", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_3387.wav", "onoffCaption": "sheep goat bleating at 1.837-3.837, 4.859-6.859, 7.452-9.452 and car horn honking at 3.359-6.578", "frequencyCaption": "sheep goat bleating three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3547.wav", "onoffCaption": "gunshot at 3.96-6.053, 7.128-9.128", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3734.wav", "onoffCaption": "train horn at 2.954-5.434, 7.906-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3761.wav", "onoffCaption": "tapping clicking clanking at 1.41-4.85, 6.821-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3784.wav", "onoffCaption": "sneeze at 1.346-3.47", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3843.wav", "onoffCaption": "woman laughing at 0.101-3.173 and spraying at 2.896-3.747", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3958.wav", "onoffCaption": "burping belching at 0.069-7.237", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_109.wav", "onoffCaption": "spraying at 3.713-4.297", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_113.wav", "onoffCaption": "thump thud at 2.654-6.572", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_234.wav", "onoffCaption": "explosion at 3.255-6.255", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_261.wav", "onoffCaption": "car horn honking at 0.263-2.776, 3.88-6.374, 7.098-9.563 and explosion at 0.45-2.452", "frequencyCaption": "car horn honking three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_284.wav", "onoffCaption": "woman laughing at 2.194-4.894, 6.241-8.941 and explosion at 2.707-7.707 and tapping clicking clanking at 3.378-6.818", "frequencyCaption": "woman laughing two times and explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_411.wav", "onoffCaption": "door slamming at 2.769-4.002, 5.099-6.332", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_444.wav", "onoffCaption": "duck quacking at 0.65-2.65, 3.391-5.391, 6.599-8.599", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_545.wav", "onoffCaption": "whistling at 1.186-4.161, 5.356-8.315", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_637.wav", "onoffCaption": "door slamming at 2.817-4.817, 6.014-7.389", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_687.wav", "onoffCaption": "duck quacking at 0.504-2.504, 4.873-6.873 and sheep goat bleating at 5.253-7.253", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_915.wav", "onoffCaption": "cat meowing at 2.392-5.296, 6.243-9.147", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_940.wav", "onoffCaption": "cat meowing at 0.108-1.418, 3.144-4.715 and explosion at 2.253-4.427, 6.188-8.362", "frequencyCaption": "cat meowing two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1075.wav", "onoffCaption": "door slamming at 3.377-4.668, 7.131-8.27", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1121.wav", "onoffCaption": "tapping clicking clanking at 0.164-3.604, 4.325-6.915", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1174.wav", "onoffCaption": "burping belching at 2.91-9.59", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1191.wav", "onoffCaption": "dog barking at 0.073-2.073, 2.957-4.957", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1249.wav", "onoffCaption": "thump thud at 0.104-4.479, 5.297-8.068", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1307.wav", "onoffCaption": "whistling at 3.185-6.16, 7.3-9.569", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1352.wav", "onoffCaption": "sheep goat bleating at 0.42-2.42, 3.209-5.209, 5.845-7.845 and gunshot at 3.536-5.536", "frequencyCaption": "sheep goat bleating three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1423.wav", "onoffCaption": "door slamming at 2.046-4.409 and cow mooing at 7.179-10.0", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1439.wav", "onoffCaption": "thump thud at 1.856-6.306", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1489.wav", "onoffCaption": "door knocking at 0.208-3.976, 5.205-8.973", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1522.wav", "onoffCaption": "burping belching at 0.811-3.811, 5.905-8.428 and cat meowing at 0.94-4.285", "frequencyCaption": "burping belching two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1577.wav", "onoffCaption": "dog barking at 2.246-4.246", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1592.wav", "onoffCaption": "tapping clicking clanking at 1.029-4.469 and woman laughing at 6.167-9.448", "frequencyCaption": "tapping clicking clanking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1704.wav", "onoffCaption": "door slamming at 3.323-4.576, 5.335-6.588 and cow mooing at 3.682-6.651", "frequencyCaption": "door slamming two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1751.wav", "onoffCaption": "spraying at 0.112-1.196, 1.83-2.914 and duck quacking at 1.325-3.325 and train horn at 1.764-4.438", "frequencyCaption": "spraying two times and duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1826.wav", "onoffCaption": "explosion at 2.068-5.068 and cow mooing at 7.619-10.0", "frequencyCaption": "explosion one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1873.wav", "onoffCaption": "spraying at 0.0-0.867, 1.694-2.561", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1896.wav", "onoffCaption": "cat meowing at 0.271-1.826 and duck quacking at 5.576-7.576", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1968.wav", "onoffCaption": "duck quacking at 0.499-2.499, 4.923-6.923", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3058.wav", "onoffCaption": "woman laughing at 0.114-6.848, 7.984-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3116.wav", "onoffCaption": "cat meowing at 0.504-1.516, 3.439-4.451, 5.303-6.315 and train horn at 0.902-4.102 and duck quacking at 2.916-4.916", "frequencyCaption": "cat meowing three times and train horn one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3143.wav", "onoffCaption": "duck quacking at 1.039-3.039, 4.428-6.428 and sneeze at 3.454-5.767, 7.844-10.0", "frequencyCaption": "duck quacking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3330.wav", "onoffCaption": "thump thud at 1.915-6.365", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3365.wav", "onoffCaption": "burping belching at 2.845-6.351, 7.13-9.237 and cow mooing at 6.748-9.717", "frequencyCaption": "burping belching two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3380.wav", "onoffCaption": "tapping clicking clanking at 0.218-3.658, 4.449-7.889 and burping belching at 3.457-5.564, 7.966-10.0", "frequencyCaption": "tapping clicking clanking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3515.wav", "onoffCaption": "car horn honking at 0.091-2.091, 3.107-5.559", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3540.wav", "onoffCaption": "burping belching at 2.061-5.561", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3628.wav", "onoffCaption": "door slamming at 0.036-1.551, 2.263-3.402, 3.906-6.623", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3698.wav", "onoffCaption": "car horn honking at 0.33-3.243, 4.313-6.313 and door slamming at 8.776-9.276", "frequencyCaption": "car horn honking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3733.wav", "onoffCaption": "whistling at 1.749-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3766.wav", "onoffCaption": "door slamming at 3.189-4.208, 5.391-7.391, 8.414-9.929", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3783.wav", "onoffCaption": "woman laughing at 0.348-2.767, 4.374-6.793", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3811.wav", "onoffCaption": "cow mooing at 0.547-3.529 and dog barking at 6.774-8.774", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3844.wav", "onoffCaption": "duck quacking at 0.575-2.575, 3.754-5.754 and spraying at 2.908-4.083, 5.036-6.211, 7.264-8.439", "frequencyCaption": "duck quacking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3945.wav", "onoffCaption": "burping belching at 2.053-8.14 and door knocking at 3.198-6.254", "frequencyCaption": "burping belching one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_34.wav", "onoffCaption": "train horn at 1.481-5.549, 7.631-9.823", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_61.wav", "onoffCaption": "duck quacking at 0.2-2.2", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_84.wav", "onoffCaption": "car horn honking at 0.489-4.143, 5.186-8.038", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_137.wav", "onoffCaption": "door slamming at 0.068-2.849", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_146.wav", "onoffCaption": "tapping clicking clanking at 1.273-4.713, 6.496-9.936 and spraying at 1.889-3.146, 3.685-4.536, 5.216-5.8", "frequencyCaption": "tapping clicking clanking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_335.wav", "onoffCaption": "whistling at 0.294-8.679", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_360.wav", "onoffCaption": "gunshot at 2.465-4.971, 6.528-9.034", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_385.wav", "onoffCaption": "cow mooing at 3.411-6.421", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_510.wav", "onoffCaption": "whistling at 0.347-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_561.wav", "onoffCaption": "door slamming at 0.024-0.964, 2.21-3.975 and thump thud at 6.63-9.13", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_678.wav", "onoffCaption": "burping belching at 0.375-3.554, 4.177-6.336", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_736.wav", "onoffCaption": "cat meowing at 1.867-2.877, 3.724-5.295", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_763.wav", "onoffCaption": "sheep goat bleating at 2.894-4.894 and cat meowing at 3.637-4.649", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_786.wav", "onoffCaption": "door slamming at 3.21-4.734", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_814.wav", "onoffCaption": "spraying at 3.628-4.378, 5.721-6.471", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_841.wav", "onoffCaption": "train horn at 3.28-7.04", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1020.wav", "onoffCaption": "sneeze at 2.574-6.63, 7.459-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1090.wav", "onoffCaption": "door knocking at 3.538-5.602, 7.279-9.343", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1206.wav", "onoffCaption": "sheep goat bleating at 1.226-3.226", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1253.wav", "onoffCaption": "dog barking at 2.277-4.277, 6.028-8.028", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1348.wav", "onoffCaption": "car horn honking at 3.545-7.945", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1476.wav", "onoffCaption": "cow mooing at 1.587-4.569, 5.899-8.881", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1493.wav", "onoffCaption": "tapping clicking clanking at 0.435-3.875, 5.267-7.646 and spraying at 1.453-2.08", "frequencyCaption": "tapping clicking clanking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1538.wav", "onoffCaption": "door knocking at 0.508-2.91, 5.169-7.571", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1588.wav", "onoffCaption": "cow mooing at 0.188-3.17, 5.076-8.058", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1605.wav", "onoffCaption": "spraying at 3.885-5.58, 7.472-8.556", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1650.wav", "onoffCaption": "door knocking at 0.332-3.388, 4.519-7.366", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1869.wav", "onoffCaption": "door knocking at 0.021-2.148, 4.194-6.321, 7.719-9.846", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1927.wav", "onoffCaption": "duck quacking at 0.096-2.096, 3.435-5.435", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1972.wav", "onoffCaption": "train horn at 0.413-4.853", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1997.wav", "onoffCaption": "spraying at 1.763-3.02 and sheep goat bleating at 4.59-6.59", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3017.wav", "onoffCaption": "duck quacking at 0.399-2.399 and explosion at 0.665-5.665 and cat meowing at 7.386-8.386", "frequencyCaption": "duck quacking one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3042.wav", "onoffCaption": "train horn at 0.39-8.59 and spraying at 2.788-3.692, 4.267-4.836, 5.639-6.147", "frequencyCaption": "train horn one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3159.wav", "onoffCaption": "tapping clicking clanking at 0.107-3.547", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3231.wav", "onoffCaption": "burping belching at 0.186-5.787, 7.524-9.555 and door slamming at 2.231-3.996", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3264.wav", "onoffCaption": "whistling at 2.172-5.047, 6.058-8.933", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3281.wav", "onoffCaption": "burping belching at 0.199-3.639, 5.384-7.667 and cat meowing at 4.673-6.568, 7.795-9.69", "frequencyCaption": "burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3414.wav", "onoffCaption": "thump thud at 3.291-5.791, 7.021-9.521", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3441.wav", "onoffCaption": "spraying at 0.092-2.552, 4.511-5.161", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3632.wav", "onoffCaption": "thump thud at 1.952-4.452, 6.848-9.348", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3667.wav", "onoffCaption": "tapping clicking clanking at 3.039-6.479, 7.37-9.609", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3682.wav", "onoffCaption": "burping belching at 1.994-4.755, 5.399-8.16", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3729.wav", "onoffCaption": "thump thud at 2.368-4.83", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3799.wav", "onoffCaption": "woman laughing at 1.673-4.773, 5.556-8.656 and spraying at 2.654-3.738, 5.882-8.025", "frequencyCaption": "woman laughing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3910.wav", "onoffCaption": "woman laughing at 0.775-3.194, 5.597-8.563", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3961.wav", "onoffCaption": "sneeze at 0.001-1.255, 1.77-3.024 and explosion at 6.263-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_10.wav", "onoffCaption": "sneeze at 0.063-1.976, 2.801-4.714 and explosion at 7.673-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_45.wav", "onoffCaption": "cow mooing at 0.355-3.337, 4.617-7.599", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_162.wav", "onoffCaption": "woman laughing at 1.442-4.496", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_187.wav", "onoffCaption": "burping belching at 0.292-7.269 and spraying at 6.756-7.34", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_311.wav", "onoffCaption": "dog barking at 2.879-4.879, 5.486-7.486", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_344.wav", "onoffCaption": "gunshot at 3.671-5.671, 7.994-9.994", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_534.wav", "onoffCaption": "cow mooing at 0.076-5.056 and door knocking at 0.179-2.559 and woman laughing at 5.2-7.281", "frequencyCaption": "cow mooing one times and door knocking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_584.wav", "onoffCaption": "gunshot at 0.179-2.179 and burping belching at 0.417-2.647 and door slamming at 5.169-7.169", "frequencyCaption": "gunshot one times and burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_609.wav", "onoffCaption": "train horn at 0.029-2.684, 3.212-5.405 and door slamming at 2.16-3.138", "frequencyCaption": "train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_712.wav", "onoffCaption": "burping belching at 0.041-2.076, 2.801-4.836, 5.602-7.637", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_747.wav", "onoffCaption": "duck quacking at 0.484-2.484, 3.334-5.334, 5.954-7.954", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_830.wav", "onoffCaption": "woman laughing at 0.458-2.574 and spraying at 0.798-1.32 and burping belching at 5.364-8.364", "frequencyCaption": "woman laughing one times and spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_865.wav", "onoffCaption": "duck quacking at 2.366-4.366, 5.105-7.105 and cow mooing at 3.278-6.26", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_880.wav", "onoffCaption": "car horn honking at 1.336-4.249 and spraying at 2.806-4.063", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1004.wav", "onoffCaption": "whistling at 3.77-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1051.wav", "onoffCaption": "car horn honking at 2.969-5.316, 6.834-8.89", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1222.wav", "onoffCaption": "burping belching at 3.161-7.497", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1277.wav", "onoffCaption": "explosion at 0.578-5.578", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1292.wav", "onoffCaption": "dog barking at 0.479-2.479, 4.616-6.616", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1339.wav", "onoffCaption": "train horn at 1.157-7.217", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1389.wav", "onoffCaption": "cat meowing at 0.527-1.674, 2.783-3.93, 4.549-5.696 and gunshot at 2.167-4.167, 6.628-8.628", "frequencyCaption": "cat meowing three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1407.wav", "onoffCaption": "gunshot at 2.758-4.758, 6.25-8.25", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1452.wav", "onoffCaption": "thump thud at 0.107-4.557, 6.901-10.0 and gunshot at 0.577-2.596, 4.566-6.585", "frequencyCaption": "thump thud two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1549.wav", "onoffCaption": "burping belching at 0.348-3.348 and door knocking at 2.068-4.445, 5.651-8.028", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1621.wav", "onoffCaption": "door knocking at 1.772-6.211", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1674.wav", "onoffCaption": "thump thud at 3.223-5.685", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1691.wav", "onoffCaption": "woman laughing at 0.558-2.752 and duck quacking at 1.9-3.9, 5.909-7.909", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1818.wav", "onoffCaption": "woman laughing at 1.357-4.452, 5.885-8.98", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1903.wav", "onoffCaption": "spraying at 2.334-2.842, 3.905-4.413, 5.561-6.069", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1956.wav", "onoffCaption": "sheep goat bleating at 2.9-4.9 and dog barking at 2.959-4.959", "frequencyCaption": "sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3033.wav", "onoffCaption": "thump thud at 1.152-5.527", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3066.wav", "onoffCaption": "car horn honking at 0.514-3.027 and sheep goat bleating at 0.548-2.548", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3083.wav", "onoffCaption": "whistling at 3.592-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3128.wav", "onoffCaption": "burping belching at 0.775-4.335 and tapping clicking clanking at 2.973-6.413", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3198.wav", "onoffCaption": "cow mooing at 0.412-5.392, 6.283-10.0 and dog barking at 0.442-2.442, 3.318-5.318, 6.398-8.398", "frequencyCaption": "cow mooing two times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3215.wav", "onoffCaption": "dog barking at 0.091-2.091, 3.213-5.213, 6.426-8.426", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3240.wav", "onoffCaption": "tapping clicking clanking at 1.525-4.965, 5.786-9.226", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3430.wav", "onoffCaption": "whistling at 1.553-9.491", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3465.wav", "onoffCaption": "duck quacking at 1.309-3.309", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3480.wav", "onoffCaption": "spraying at 0.255-2.383, 4.595-6.723", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3616.wav", "onoffCaption": "train horn at 0.002-2.482, 3.51-5.99 and door slamming at 0.294-1.669", "frequencyCaption": "train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3643.wav", "onoffCaption": "cow mooing at 0.5-3.469", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3758.wav", "onoffCaption": "door knocking at 0.079-2.847, 4.619-6.84 and dog barking at 5.333-7.333", "frequencyCaption": "door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3934.wav", "onoffCaption": "door knocking at 0.916-4.684", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3984.wav", "onoffCaption": "train horn at 1.384-4.744 and thump thud at 2.294-4.794, 5.817-7.852 and car horn honking at 3.846-6.759", "frequencyCaption": "train horn one times and thump thud two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_178.wav", "onoffCaption": "cow mooing at 3.381-7.81", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_210.wav", "onoffCaption": "whistling at 0.071-2.08, 2.921-4.93 and explosion at 0.913-3.913, 5.782-8.782", "frequencyCaption": "whistling two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_245.wav", "onoffCaption": "explosion at 1.121-4.177, 6.294-9.35", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_435.wav", "onoffCaption": "woman laughing at 0.561-3.199, 4.189-6.827 and gunshot at 3.425-5.425", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_460.wav", "onoffCaption": "duck quacking at 3.484-5.484, 6.175-8.175", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_485.wav", "onoffCaption": "tapping clicking clanking at 0.224-3.664 and door knocking at 6.512-9.28", "frequencyCaption": "tapping clicking clanking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_613.wav", "onoffCaption": "whistling at 0.43-8.085 and dog barking at 2.187-4.625, 5.225-7.225 and cow mooing at 3.707-6.689", "frequencyCaption": "whistling one times and dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_646.wav", "onoffCaption": "train horn at 3.064-7.922", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_708.wav", "onoffCaption": "gunshot at 2.261-4.767, 7.175-9.175", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_931.wav", "onoffCaption": "thump thud at 0.167-2.667", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_964.wav", "onoffCaption": "sheep goat bleating at 1.528-3.528, 4.975-6.975", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_981.wav", "onoffCaption": "car horn honking at 0.323-4.645 and gunshot at 7.394-9.394", "frequencyCaption": "car horn honking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1105.wav", "onoffCaption": "sheep goat bleating at 2.371-4.371, 6.611-8.611", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1150.wav", "onoffCaption": "duck quacking at 2.106-4.106, 5.268-7.268", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1238.wav", "onoffCaption": "sheep goat bleating at 3.412-6.492", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1288.wav", "onoffCaption": "cat meowing at 0.746-2.363", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1323.wav", "onoffCaption": "dog barking at 0.979-2.979, 4.467-6.467, 7.269-9.269", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1376.wav", "onoffCaption": "explosion at 2.257-7.257", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1393.wav", "onoffCaption": "tapping clicking clanking at 0.14-3.58, 5.201-7.253 and duck quacking at 2.426-4.426", "frequencyCaption": "tapping clicking clanking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1448.wav", "onoffCaption": "woman laughing at 0.367-2.851, 5.089-7.789", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1506.wav", "onoffCaption": "dog barking at 0.726-2.726, 4.078-6.078", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1553.wav", "onoffCaption": "tapping clicking clanking at 0.156-3.596, 4.737-7.641", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1720.wav", "onoffCaption": "gunshot at 1.155-3.155, 5.398-7.528 and thump thud at 3.363-7.738", "frequencyCaption": "gunshot two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1775.wav", "onoffCaption": "door knocking at 2.969-5.816, 7.75-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1790.wav", "onoffCaption": "door knocking at 0.011-2.391, 4.036-6.546, 7.09-9.7 and train horn at 1.029-5.469 and thump thud at 1.711-4.211", "frequencyCaption": "door knocking three times and train horn one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1802.wav", "onoffCaption": "sneeze at 0.384-2.697, 4.475-6.788", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1857.wav", "onoffCaption": "whistling at 1.733-4.608", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1919.wav", "onoffCaption": "sheep goat bleating at 0.776-2.776, 3.517-5.517, 6.164-8.164 and car horn honking at 4.011-8.411", "frequencyCaption": "sheep goat bleating three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3029.wav", "onoffCaption": "cow mooing at 3.488-6.47, 7.941-10.0 and burping belching at 3.869-6.869", "frequencyCaption": "cow mooing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3099.wav", "onoffCaption": "gunshot at 1.876-3.876, 4.38-6.38 and woman laughing at 2.584-4.778", "frequencyCaption": "gunshot two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3132.wav", "onoffCaption": "door knocking at 0.692-5.525, 6.427-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3167.wav", "onoffCaption": "woman laughing at 3.333-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3182.wav", "onoffCaption": "tapping clicking clanking at 1.809-5.249, 6.999-9.867", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3314.wav", "onoffCaption": "sheep goat bleating at 1.93-3.93", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3341.wav", "onoffCaption": "spraying at 3.082-5.666, 7.209-9.793", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3531.wav", "onoffCaption": "sneeze at 0.185-3.26, 3.84-6.316, 7.242-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3564.wav", "onoffCaption": "sheep goat bleating at 0.809-2.809, 4.118-6.65, 7.416-9.416", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3581.wav", "onoffCaption": "explosion at 0.101-2.189", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3659.wav", "onoffCaption": "sneeze at 0.638-2.751, 3.548-5.661, 6.812-8.925", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3717.wav", "onoffCaption": "door knocking at 1.199-3.359, 4.572-7.531", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3742.wav", "onoffCaption": "cat meowing at 0.275-1.286", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3835.wav", "onoffCaption": "cat meowing at 2.698-3.707, 4.225-6.202", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3860.wav", "onoffCaption": "duck quacking at 0.859-2.859, 3.637-5.637, 6.544-8.544", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3885.wav", "onoffCaption": "cow mooing at 3.53-8.51", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2120.wav", "onoffCaption": "spraying at 0.057-2.449, 3.255-3.763", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2175.wav", "onoffCaption": "spraying at 2.071-3.071 and sheep goat bleating at 3.431-7.351", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2190.wav", "onoffCaption": "train horn at 1.771-5.305", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2248.wav", "onoffCaption": "explosion at 1.654-4.372, 5.029-7.747 and car horn honking at 2.113-4.626, 5.21-7.675 and burping belching at 5.214-7.54", "frequencyCaption": "explosion two times and car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2306.wav", "onoffCaption": "cow mooing at 0.725-3.694, 6.08-8.942", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2353.wav", "onoffCaption": "car horn honking at 1.862-3.862, 4.452-6.452, 7.24-9.24", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_2438.wav", "onoffCaption": "sheep goat bleating at 1.107-4.107 and duck quacking at 3.052-5.052", "frequencyCaption": "sheep goat bleating one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2477.wav", "onoffCaption": "whistling at 2.127-5.002, 5.545-8.42", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2488.wav", "onoffCaption": "sheep goat bleating at 0.735-2.735, 3.386-5.386 and door slamming at 1.8-3.033, 3.723-4.956 and tapping clicking clanking at 2.198-5.638, 6.329-9.769", "frequencyCaption": "sheep goat bleating two times and door slamming two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2523.wav", "onoffCaption": "woman laughing at 2.767-4.967, 5.806-8.006", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2576.wav", "onoffCaption": "explosion at 0.277-2.868 and cat meowing at 6.026-10.0", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2593.wav", "onoffCaption": "dog barking at 1.825-3.825, 4.517-6.517", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2705.wav", "onoffCaption": "woman laughing at 0.971-4.071 and explosion at 7.935-10.0", "frequencyCaption": "woman laughing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2750.wav", "onoffCaption": "gunshot at 2.845-4.845, 5.516-7.516", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2827.wav", "onoffCaption": "burping belching at 0.845-3.171", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2872.wav", "onoffCaption": "door slamming at 3.272-6.233", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2897.wav", "onoffCaption": "car horn honking at 1.909-5.75, 7.713-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2969.wav", "onoffCaption": "duck quacking at 0.892-2.892", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2996.wav", "onoffCaption": "dog barking at 2.902-4.902, 7.003-9.003", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4078.wav", "onoffCaption": "door slamming at 0.238-0.738", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4136.wav", "onoffCaption": "burping belching at 3.19-7.213 and duck quacking at 3.224-5.224, 5.953-7.953", "frequencyCaption": "burping belching one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4163.wav", "onoffCaption": "train horn at 1.452-4.332 and cow mooing at 6.448-9.746", "frequencyCaption": "train horn one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4186.wav", "onoffCaption": "car horn honking at 1.257-3.604 and dog barking at 6.705-10.0", "frequencyCaption": "car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4310.wav", "onoffCaption": "train horn at 2.161-6.342, 6.935-9.277", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4345.wav", "onoffCaption": "gunshot at 0.772-2.772", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4535.wav", "onoffCaption": "sheep goat bleating at 2.485-6.125", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4560.wav", "onoffCaption": "burping belching at 0.436-3.197 and gunshot at 1.84-4.346", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4585.wav", "onoffCaption": "thump thud at 0.62-4.538, 6.81-9.31", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4608.wav", "onoffCaption": "whistling at 3.318-7.802", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4713.wav", "onoffCaption": "train horn at 0.267-3.427, 5.079-8.239", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4746.wav", "onoffCaption": "gunshot at 1.968-3.968, 4.801-6.801", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4831.wav", "onoffCaption": "tapping clicking clanking at 0.749-4.189", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4864.wav", "onoffCaption": "sheep goat bleating at 0.308-2.308, 4.509-6.509", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4881.wav", "onoffCaption": "sneeze at 0.109-2.512, 3.965-6.397 and whistling at 2.282-7.782", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_6.wav", "onoffCaption": "thump thud at 0.14-3.807, 5.41-9.077 and spraying at 0.454-0.954, 1.799-2.65, 4.17-4.745", "frequencyCaption": "thump thud two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2021.wav", "onoffCaption": "door knocking at 0.429-5.429, 6.778-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2074.wav", "onoffCaption": "whistling at 0.727-5.211, 6.879-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2091.wav", "onoffCaption": "spraying at 1.831-4.85, 6.142-7.837, 8.47-9.12", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2207.wav", "onoffCaption": "thump thud at 2.65-5.15", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2252.wav", "onoffCaption": "explosion at 0.012-3.565, 5.39-7.843", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2349.wav", "onoffCaption": "thump thud at 0.309-2.809, 3.923-6.423, 7.405-9.905", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2422.wav", "onoffCaption": "car horn honking at 2.361-6.761, 7.274-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2453.wav", "onoffCaption": "thump thud at 3.225-5.687, 6.839-9.605", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2492.wav", "onoffCaption": "explosion at 0.455-5.455 and burping belching at 2.437-6.306, 7.467-9.59", "frequencyCaption": "explosion one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2539.wav", "onoffCaption": "sneeze at 0.488-2.612, 3.718-6.686, 7.684-9.212", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2548.wav", "onoffCaption": "door knocking at 2.153-4.616, 5.711-8.174", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2589.wav", "onoffCaption": "gunshot at 3.892-5.892", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2604.wav", "onoffCaption": "thump thud at 0.045-3.092, 3.677-5.905 and train horn at 1.656-4.856, 5.734-8.934", "frequencyCaption": "thump thud two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2651.wav", "onoffCaption": "explosion at 2.423-5.151, 6.028-8.756", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2868.wav", "onoffCaption": "thump thud at 0.855-5.23, 6.386-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2926.wav", "onoffCaption": "dog barking at 0.219-3.14, 4.211-7.132", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2973.wav", "onoffCaption": "tapping clicking clanking at 2.703-6.143, 7.187-9.19", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4037.wav", "onoffCaption": "train horn at 1.92-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4062.wav", "onoffCaption": "cat meowing at 0.012-1.596, 2.868-3.877, 5.167-7.299 and dog barking at 6.274-8.274", "frequencyCaption": "cat meowing three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4087.wav", "onoffCaption": "explosion at 2.553-6.106, 7.68-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4179.wav", "onoffCaption": "duck quacking at 2.786-4.786, 5.324-7.324", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4211.wav", "onoffCaption": "thump thud at 0.315-2.654, 4.121-6.645 and spraying at 5.969-6.75", "frequencyCaption": "thump thud two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4244.wav", "onoffCaption": "car horn honking at 0.049-3.636 and dog barking at 0.885-2.885 and spraying at 1.371-2.435", "frequencyCaption": "car horn honking one times and dog barking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4434.wav", "onoffCaption": "sneeze at 3.28-5.519, 7.824-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4461.wav", "onoffCaption": "cat meowing at 3.162-4.311, 5.935-6.946", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4484.wav", "onoffCaption": "door knocking at 3.211-5.674, 6.972-9.435", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4612.wav", "onoffCaption": "door knocking at 0.089-4.528, 5.195-9.634", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4647.wav", "onoffCaption": "sneeze at 0.569-2.156, 3.144-4.731, 5.782-7.369 and gunshot at 5.878-7.878", "frequencyCaption": "sneeze three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4709.wav", "onoffCaption": "whistling at 0.206-6.583, 7.637-10.0 and door knocking at 2.165-4.735", "frequencyCaption": "whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4930.wav", "onoffCaption": "explosion at 2.446-7.446", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4965.wav", "onoffCaption": "explosion at 3.189-6.245, 7.568-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4980.wav", "onoffCaption": "cat meowing at 2.685-3.712 and door slamming at 2.893-4.032, 6.014-7.153", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2005.wav", "onoffCaption": "thump thud at 0.343-4.01, 4.582-7.082 and car horn honking at 1.183-5.583, 7.486-10.0", "frequencyCaption": "thump thud two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2050.wav", "onoffCaption": "sneeze at 2.987-4.09 and spraying at 4.169-5.073, 5.726-6.327, 7.545-8.629", "frequencyCaption": "sneeze one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2223.wav", "onoffCaption": "duck quacking at 0.332-2.332, 2.942-4.942, 6.627-8.627", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2239.wav", "onoffCaption": "sheep goat bleating at 0.339-2.339, 3.156-5.156", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2276.wav", "onoffCaption": "thump thud at 0.322-4.697, 6.983-9.599", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2293.wav", "onoffCaption": "car horn honking at 0.996-3.461, 4.536-7.001, 7.588-10.0 and tapping clicking clanking at 2.689-6.129", "frequencyCaption": "car horn honking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2322.wav", "onoffCaption": "thump thud at 0.418-3.465, 4.569-7.161", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2338.wav", "onoffCaption": "spraying at 0.027-1.513, 3.975-5.232, 7.548-8.723 and burping belching at 2.07-4.593", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2388.wav", "onoffCaption": "woman laughing at 0.794-3.002, 4.227-6.435", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2406.wav", "onoffCaption": "car horn honking at 0.603-2.603 and train horn at 3.813-6.973", "frequencyCaption": "car horn honking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2620.wav", "onoffCaption": "cat meowing at 2.446-4.582, 5.348-7.484", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2675.wav", "onoffCaption": "dog barking at 2.962-4.962, 7.387-9.387", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2690.wav", "onoffCaption": "duck quacking at 0.393-2.393, 3.405-5.405, 5.915-7.915", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2774.wav", "onoffCaption": "sheep goat bleating at 2.619-4.619, 5.242-7.242", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2819.wav", "onoffCaption": "sheep goat bleating at 0.032-2.032, 3.521-5.521 and door knocking at 0.864-3.984, 6.406-9.526", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2902.wav", "onoffCaption": "sneeze at 0.001-2.462, 4.092-6.553", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2957.wav", "onoffCaption": "door slamming at 0.986-3.865, 4.823-7.702", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4013.wav", "onoffCaption": "tapping clicking clanking at 0.088-3.528, 5.482-8.922", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4046.wav", "onoffCaption": "cow mooing at 0.641-5.07 and door slamming at 1.475-3.24, 4.431-6.196, 6.906-8.671", "frequencyCaption": "cow mooing one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4108.wav", "onoffCaption": "spraying at 1.708-2.649, 4.881-5.508", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4112.wav", "onoffCaption": "sneeze at 3.141-4.418", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4235.wav", "onoffCaption": "car horn honking at 0.77-5.092, 6.546-9.404", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4260.wav", "onoffCaption": "woman laughing at 0.463-3.517, 5.483-8.537", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4285.wav", "onoffCaption": "cow mooing at 0.753-3.722, 4.97-7.106", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4410.wav", "onoffCaption": "woman laughing at 0.635-7.369, 7.95-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4445.wav", "onoffCaption": "whistling at 2.422-8.914", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4544.wav", "onoffCaption": "spraying at 1.576-2.326, 2.849-3.599, 4.646-5.396 and gunshot at 1.949-3.949, 5.622-7.622", "frequencyCaption": "spraying three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4636.wav", "onoffCaption": "train horn at 0.178-2.852", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4663.wav", "onoffCaption": "sneeze at 2.824-5.484, 6.738-7.832 and duck quacking at 2.914-4.914", "frequencyCaption": "sneeze two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4686.wav", "onoffCaption": "car horn honking at 1.868-5.363, 6.193-9.688", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4778.wav", "onoffCaption": "dog barking at 1.856-3.856, 4.606-6.606, 7.924-9.924 and tapping clicking clanking at 2.775-6.215", "frequencyCaption": "dog barking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4840.wav", "onoffCaption": "whistling at 0.068-4.552", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4914.wav", "onoffCaption": "dog barking at 2.546-4.546, 5.133-7.484", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4941.wav", "onoffCaption": "duck quacking at 0.364-2.364, 4.648-6.648", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2104.wav", "onoffCaption": "car horn honking at 2.211-4.558, 6.798-9.145", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2151.wav", "onoffCaption": "gunshot at 2.256-4.256, 4.891-6.891", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2156.wav", "onoffCaption": "dog barking at 0.052-2.052, 2.948-4.948, 5.728-7.728 and door slamming at 1.308-4.282, 5.326-8.3", "frequencyCaption": "dog barking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2289.wav", "onoffCaption": "spraying at 1.58-2.312, 3.778-4.51, 5.034-5.766 and gunshot at 2.716-4.716, 5.541-7.541", "frequencyCaption": "spraying three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2377.wav", "onoffCaption": "cat meowing at 3.482-7.842", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2392.wav", "onoffCaption": "sneeze at 3.873-5.869, 6.876-8.872", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2449.wav", "onoffCaption": "train horn at 0.152-2.632 and spraying at 0.716-1.583 and woman laughing at 5.347-7.702", "frequencyCaption": "train horn one times and spraying one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2500.wav", "onoffCaption": "door slamming at 1.297-2.135 and spraying at 4.361-4.869, 7.135-7.704", "frequencyCaption": "door slamming one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2507.wav", "onoffCaption": "door knocking at 0.199-2.387", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2552.wav", "onoffCaption": "sheep goat bleating at 0.008-2.008 and woman laughing at 0.341-3.441, 4.758-7.858", "frequencyCaption": "sheep goat bleating one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2668.wav", "onoffCaption": "tapping clicking clanking at 2.077-5.517, 6.443-8.613 and cow mooing at 5.63-8.612", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2721.wav", "onoffCaption": "gunshot at 3.487-5.487", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2773.wav", "onoffCaption": "door knocking at 0.433-2.896 and door slamming at 3.612-4.987, 7.366-8.741", "frequencyCaption": "door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2791.wav", "onoffCaption": "door knocking at 0.652-4.402", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2803.wav", "onoffCaption": "thump thud at 0.011-2.35 and spraying at 4.888-6.145", "frequencyCaption": "thump thud one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2804.wav", "onoffCaption": "cow mooing at 0.049-4.478, 6.606-9.378 and thump thud at 2.151-4.922, 6.328-8.828 and burping belching at 3.01-5.108", "frequencyCaption": "cow mooing two times and thump thud two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2856.wav", "onoffCaption": "duck quacking at 1.872-3.872, 6.151-8.151", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2918.wav", "onoffCaption": "gunshot at 3.13-5.13, 6.539-8.539", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4009.wav", "onoffCaption": "whistling at 1.047-6.547, 7.7-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4147.wav", "onoffCaption": "whistling at 1.949-7.124", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4298.wav", "onoffCaption": "train horn at 0.227-2.667, 4.148-6.296", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4334.wav", "onoffCaption": "explosion at 0.551-3.551 and cat meowing at 0.943-3.079", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4361.wav", "onoffCaption": "cow mooing at 2.731-5.7, 6.649-9.618", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4366.wav", "onoffCaption": "duck quacking at 0.872-2.872, 4.328-6.328 and explosion at 1.918-4.79, 5.543-8.271", "frequencyCaption": "duck quacking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_4383.wav", "onoffCaption": "car horn honking at 1.956-5.175, 5.822-8.12 and burping belching at 4.017-7.017", "frequencyCaption": "car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4384.wav", "onoffCaption": "door knocking at 0.945-4.001 and train horn at 6.737-8.892", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4511.wav", "onoffCaption": "car horn honking at 3.701-8.213", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4679.wav", "onoffCaption": "thump thud at 0.164-4.614, 5.439-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4730.wav", "onoffCaption": "train horn at 1.755-4.429, 5.275-7.949", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4737.wav", "onoffCaption": "gunshot at 1.048-3.048, 4.748-6.748 and spraying at 3.137-4.201 and sheep goat bleating at 3.936-5.936", "frequencyCaption": "gunshot two times and spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4762.wav", "onoffCaption": "door slamming at 2.032-4.49, 6.927-8.927", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4787.wav", "onoffCaption": "woman laughing at 0.311-2.511, 3.934-6.526 and sneeze at 2.213-3.8, 6.089-7.676", "frequencyCaption": "woman laughing two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4815.wav", "onoffCaption": "sneeze at 0.265-1.499 and whistling at 5.673-7.902", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4847.wav", "onoffCaption": "whistling at 0.013-7.763 and tapping clicking clanking at 0.111-3.551 and sneeze at 1.548-3.135, 4.059-5.646", "frequencyCaption": "whistling one times and tapping clicking clanking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2002.wav", "onoffCaption": "woman laughing at 3.333-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2018.wav", "onoffCaption": "spraying at 0.405-1.032, 1.643-2.27, 3.029-3.656 and explosion at 0.527-5.527, 6.694-10.0", "frequencyCaption": "spraying three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2103.wav", "onoffCaption": "sneeze at 1.671-3.984, 5.21-7.523 and burping belching at 1.76-3.99, 5.309-7.346 and door slamming at 4.12-6.837", "frequencyCaption": "sneeze two times and burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2119.wav", "onoffCaption": "dog barking at 2.203-4.203, 6.323-8.323", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2271.wav", "onoffCaption": "tapping clicking clanking at 0.592-4.032, 5.924-9.364", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2294.wav", "onoffCaption": "gunshot at 0.764-2.764 and spraying at 5.053-6.134, 8.451-9.532", "frequencyCaption": "gunshot one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2325.wav", "onoffCaption": "dog barking at 0.642-2.642, 3.414-5.414, 6.681-8.681 and cow mooing at 4.632-7.601", "frequencyCaption": "dog barking three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2370.wav", "onoffCaption": "thump thud at 0.04-2.502", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2395.wav", "onoffCaption": "spraying at 0.189-0.697, 1.468-1.976, 2.924-3.432 and burping belching at 1.098-3.328", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2454.wav", "onoffCaption": "cat meowing at 0.477-2.425, 3.605-5.553, 7.46-9.408", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2555.wav", "onoffCaption": "cat meowing at 0.068-1.484, 3.808-5.768", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2627.wav", "onoffCaption": "burping belching at 0.031-4.031, 5.198-9.198", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2697.wav", "onoffCaption": "whistling at 2.205-4.214, 4.925-6.934 and sneeze at 4.976-6.15, 8.57-9.744", "frequencyCaption": "whistling two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2726.wav", "onoffCaption": "sneeze at 1.043-3.446 and thump thud at 6.486-8.825", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2796.wav", "onoffCaption": "spraying at 0.031-1.206, 2.852-4.099, 5.641-8.225 and tapping clicking clanking at 6.006-9.446", "frequencyCaption": "spraying three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2851.wav", "onoffCaption": "burping belching at 0.624-4.13, 5.824-9.33", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2950.wav", "onoffCaption": "explosion at 3.244-5.973 and sheep goat bleating at 3.421-5.421", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4014.wav", "onoffCaption": "door knocking at 2.683-5.063, 7.022-9.402", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4041.wav", "onoffCaption": "explosion at 2.544-7.544", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4115.wav", "onoffCaption": "cow mooing at 3.295-8.275", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4140.wav", "onoffCaption": "door slamming at 1.027-3.906 and whistling at 6.755-9.73", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4228.wav", "onoffCaption": "woman laughing at 0.009-2.107, 4.329-6.427", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4232.wav", "onoffCaption": "cat meowing at 0.174-1.449, 2.265-3.54, 4.479-5.754", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4329.wav", "onoffCaption": "burping belching at 2.373-5.873", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4333.wav", "onoffCaption": "door knocking at 0.096-4.629 and cow mooing at 2.461-5.471, 6.445-9.405", "frequencyCaption": "door knocking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4417.wav", "onoffCaption": "explosion at 0.436-3.027, 4.242-6.244, 7.375-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_4458.wav", "onoffCaption": "door slamming at 2.041-3.806", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4516.wav", "onoffCaption": "thump thud at 3.061-6.108, 6.938-9.985", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4543.wav", "onoffCaption": "dog barking at 3.213-5.213 and sheep goat bleating at 7.736-9.736", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4664.wav", "onoffCaption": "car horn honking at 0.073-2.586 and cat meowing at 0.304-1.304, 2.382-3.382, 4.485-5.485", "frequencyCaption": "car horn honking one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4681.wav", "onoffCaption": "explosion at 1.201-6.201, 7.318-10.0 and train horn at 3.436-6.636", "frequencyCaption": "explosion two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4765.wav", "onoffCaption": "sheep goat bleating at 0.926-2.926, 3.878-6.477, 7.176-9.176", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4780.wav", "onoffCaption": "cat meowing at 0.028-1.303, 2.305-3.876, 5.456-6.483", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4808.wav", "onoffCaption": "duck quacking at 2.642-4.642, 6.256-8.256", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4812.wav", "onoffCaption": "burping belching at 3.173-7.509", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4909.wav", "onoffCaption": "train horn at 0.691-4.891, 5.967-8.26", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4913.wav", "onoffCaption": "sneeze at 0.276-1.51 and whistling at 5.819-10.0", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2026.wav", "onoffCaption": "sheep goat bleating at 0.154-2.154, 3.022-5.022, 6.172-8.172", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2057.wav", "onoffCaption": "burping belching at 2.683-6.227, 6.907-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2200.wav", "onoffCaption": "whistling at 2.022-9.772", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2224.wav", "onoffCaption": "burping belching at 2.351-4.382", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2255.wav", "onoffCaption": "tapping clicking clanking at 2.125-5.565 and sneeze at 2.99-5.466, 6.608-7.772", "frequencyCaption": "tapping clicking clanking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2401.wav", "onoffCaption": "cow mooing at 0.674-5.103, 6.783-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2470.wav", "onoffCaption": "door slamming at 0.396-1.771, 2.767-4.142", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2495.wav", "onoffCaption": "door slamming at 0.565-2.928, 3.911-4.762, 6.499-7.439", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2603.wav", "onoffCaption": "sheep goat bleating at 1.829-3.829 and cow mooing at 6.879-9.848", "frequencyCaption": "sheep goat bleating one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2672.wav", "onoffCaption": "cat meowing at 1.868-3.485 and door slamming at 2.29-4.516, 6.876-9.102", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2718.wav", "onoffCaption": "sneeze at 0.124-1.581, 3.955-5.483", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2769.wav", "onoffCaption": "car horn honking at 0.692-4.941", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2905.wav", "onoffCaption": "burping belching at 0.043-4.442, 6.234-10.0 and sheep goat bleating at 0.792-2.792, 3.816-5.816, 6.776-8.776", "frequencyCaption": "burping belching two times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2974.wav", "onoffCaption": "sheep goat bleating at 3.18-6.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2991.wav", "onoffCaption": "thump thud at 0.605-4.272 and tapping clicking clanking at 0.78-4.22 and sheep goat bleating at 1.8-3.8, 4.865-6.865, 7.674-9.674", "frequencyCaption": "thump thud one times and tapping clicking clanking one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4030.wav", "onoffCaption": "sheep goat bleating at 2.192-4.192, 4.895-6.895 and whistling at 3.502-9.002", "frequencyCaption": "sheep goat bleating two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4065.wav", "onoffCaption": "thump thud at 0.945-3.992, 5.083-8.13", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4080.wav", "onoffCaption": "whistling at 1.188-3.197, 5.237-7.246 and car horn honking at 3.639-7.961", "frequencyCaption": "whistling two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4216.wav", "onoffCaption": "woman laughing at 0.33-3.611, 4.272-6.404", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4267.wav", "onoffCaption": "tapping clicking clanking at 2.832-6.272 and car horn honking at 5.367-7.832", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4282.wav", "onoffCaption": "dog barking at 1.217-3.217 and train horn at 6.035-9.515", "frequencyCaption": "dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4399.wav", "onoffCaption": "door knocking at 1.174-3.637 and sheep goat bleating at 6.536-10.0", "frequencyCaption": "door knocking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4433.wav", "onoffCaption": "woman laughing at 0.612-2.98, 5.346-7.714", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4442.wav", "onoffCaption": "door knocking at 2.41-4.631, 5.512-7.733", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4466.wav", "onoffCaption": "thump thud at 0.065-2.565, 3.598-5.778, 6.412-8.912 and gunshot at 2.169-4.169, 4.865-6.865, 7.796-9.796", "frequencyCaption": "thump thud three times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4483.wav", "onoffCaption": "cow mooing at 0.202-3.171, 4.097-7.066, 7.97-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_4528.wav", "onoffCaption": "burping belching at 0.152-3.152, 5.315-8.315 and woman laughing at 3.008-5.613 and door slamming at 3.552-4.785, 6.323-7.556", "frequencyCaption": "burping belching two times and woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4559.wav", "onoffCaption": "tapping clicking clanking at 0.302-3.742, 4.486-6.714 and burping belching at 1.799-6.799", "frequencyCaption": "tapping clicking clanking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4598.wav", "onoffCaption": "duck quacking at 1.01-3.01, 3.848-5.848, 6.532-8.532", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4631.wav", "onoffCaption": "burping belching at 0.8-3.621, 5.735-8.514", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4640.wav", "onoffCaption": "cat meowing at 2.515-3.825, 4.843-6.593", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4937.wav", "onoffCaption": "spraying at 0.644-1.891, 3.986-4.986", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4946.wav", "onoffCaption": "car horn honking at 0.421-3.347, 3.927-6.853, 7.568-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4987.wav", "onoffCaption": "gunshot at 0.526-2.526 and dog barking at 5.505-7.505", "frequencyCaption": "gunshot one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1.wav", "onoffCaption": "burping belching at 1.324-8.492", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2069.wav", "onoffCaption": "sheep goat bleating at 0.568-2.568, 3.305-5.305 and door slamming at 7.71-8.561", "frequencyCaption": "sheep goat bleating two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2073.wav", "onoffCaption": "dog barking at 0.81-2.81, 4.698-6.698 and sneeze at 1.441-3.687, 5.397-7.873", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2096.wav", "onoffCaption": "sneeze at 0.54-1.643 and car horn honking at 0.905-5.305, 6.304-10.0", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2168.wav", "onoffCaption": "woman laughing at 2.695-5.976, 6.949-9.534", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2172.wav", "onoffCaption": "woman laughing at 2.653-6.229", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2197.wav", "onoffCaption": "woman laughing at 0.631-3.05, 5.474-7.715", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2301.wav", "onoffCaption": "door slamming at 0.391-2.309, 3.454-5.372 and sneeze at 2.098-5.208", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2425.wav", "onoffCaption": "car horn honking at 0.069-3.564, 5.485-8.276", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2524.wav", "onoffCaption": "cow mooing at 3.119-7.548", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2656.wav", "onoffCaption": "tapping clicking clanking at 0.232-3.672, 4.981-8.421", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2757.wav", "onoffCaption": "car horn honking at 1.995-6.395, 7.171-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2820.wav", "onoffCaption": "burping belching at 0.103-2.138, 3.472-5.507, 6.832-8.867", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2875.wav", "onoffCaption": "door knocking at 1.909-6.442", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2921.wav", "onoffCaption": "sheep goat bleating at 1.122-3.122, 4.877-6.877", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4131.wav", "onoffCaption": "train horn at 0.546-3.22 and sheep goat bleating at 1.516-3.516, 4.159-6.159, 7.47-9.47", "frequencyCaption": "train horn one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4243.wav", "onoffCaption": "woman laughing at 1.329-3.411, 5.369-7.451", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4259.wav", "onoffCaption": "tapping clicking clanking at 0.446-3.886, 5.188-7.422 and spraying at 5.404-6.404", "frequencyCaption": "tapping clicking clanking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4342.wav", "onoffCaption": "woman laughing at 1.285-4.357 and sneeze at 1.929-4.405, 5.129-7.605", "frequencyCaption": "woman laughing one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4358.wav", "onoffCaption": "dog barking at 1.161-3.161, 5.169-7.169", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4499.wav", "onoffCaption": "car horn honking at 0.823-5.73 and door slamming at 2.357-3.748, 5.822-7.213 and duck quacking at 3.094-5.094, 7.242-9.242", "frequencyCaption": "car horn honking one times and door slamming two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4567.wav", "onoffCaption": "dog barking at 2.991-4.991, 7.131-9.131", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4582.wav", "onoffCaption": "explosion at 0.407-3.125 and gunshot at 3.084-5.084, 6.253-8.253 and burping belching at 4.468-7.647", "frequencyCaption": "explosion one times and gunshot two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4615.wav", "onoffCaption": "tapping clicking clanking at 0.349-3.789, 5.332-7.939 and cow mooing at 2.213-5.511, 7.329-10.0", "frequencyCaption": "tapping clicking clanking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4714.wav", "onoffCaption": "thump thud at 1.441-4.488, 5.626-8.673", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4863.wav", "onoffCaption": "sheep goat bleating at 2.651-5.731, 7.327-9.327", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4879.wav", "onoffCaption": "sneeze at 1.717-3.005, 4.541-5.829", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4886.wav", "onoffCaption": "woman laughing at 2.456-5.048, 5.701-8.293", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4962.wav", "onoffCaption": "spraying at 0.625-2.32, 2.931-3.798", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4978.wav", "onoffCaption": "door knocking at 1.907-4.095", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2098.wav", "onoffCaption": "duck quacking at 1.532-3.532", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2127.wav", "onoffCaption": "explosion at 3.111-5.983, 6.942-9.582", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2183.wav", "onoffCaption": "thump thud at 2.161-5.208, 6.064-8.472", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2315.wav", "onoffCaption": "explosion at 1.424-6.424 and cat meowing at 2.368-4.316 and car horn honking at 2.424-4.424", "frequencyCaption": "explosion one times and cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2354.wav", "onoffCaption": "door knocking at 0.652-4.152, 5.012-7.135, 7.746-9.81", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2571.wav", "onoffCaption": "spraying at 2.456-4.584, 5.711-7.473", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2594.wav", "onoffCaption": "car horn honking at 2.033-4.533, 5.2-7.7", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2619.wav", "onoffCaption": "duck quacking at 1.522-3.522 and cat meowing at 1.603-3.791", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2658.wav", "onoffCaption": "explosion at 0.811-5.677, 7.813-10.0 and door knocking at 2.02-5.77", "frequencyCaption": "explosion two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2702.wav", "onoffCaption": "whistling at 0.515-3.39, 3.92-6.795 and explosion at 2.138-5.01, 6.901-9.773", "frequencyCaption": "whistling two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2743.wav", "onoffCaption": "tapping clicking clanking at 2.399-5.839, 6.987-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2834.wav", "onoffCaption": "cow mooing at 2.952-5.962, 7.639-10.0 and woman laughing at 3.706-6.406", "frequencyCaption": "cow mooing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2890.wav", "onoffCaption": "duck quacking at 2.845-4.845", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4125.wav", "onoffCaption": "thump thud at 1.182-3.682 and door knocking at 6.723-9.57", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4164.wav", "onoffCaption": "whistling at 3.425-5.654", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4181.wav", "onoffCaption": "car horn honking at 0.313-3.488, 5.773-8.948", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4317.wav", "onoffCaption": "sneeze at 0.088-2.201, 3.06-5.377, 6.262-7.496 and door slamming at 0.488-2.708, 3.344-4.244 and train horn at 0.872-6.587, 7.647-10.0", "frequencyCaption": "sneeze three times and door slamming two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4356.wav", "onoffCaption": "train horn at 2.962-6.722", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4429.wav", "onoffCaption": "duck quacking at 0.833-2.833, 3.531-5.531, 6.133-8.133 and woman laughing at 5.524-8.624", "frequencyCaption": "duck quacking three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4468.wav", "onoffCaption": "door slamming at 0.051-2.509", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4532.wav", "onoffCaption": "door slamming at 2.31-4.793, 5.41-6.701", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4573.wav", "onoffCaption": "burping belching at 1.245-5.114 and door slamming at 2.463-4.463, 5.179-7.179", "frequencyCaption": "burping belching one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4596.wav", "onoffCaption": "dog barking at 0.023-2.023, 2.648-4.648, 6.804-8.804", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4700.wav", "onoffCaption": "door slamming at 1.79-4.01 and door knocking at 7.65-10.0", "frequencyCaption": "door slamming one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4741.wav", "onoffCaption": "sneeze at 3.285-7.785", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4836.wav", "onoffCaption": "car horn honking at 0.067-2.993, 3.783-6.709, 7.913-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4877.wav", "onoffCaption": "explosion at 0.204-2.524, 3.236-5.41, 6.621-8.623", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_4892.wav", "onoffCaption": "thump thud at 0.249-3.916, 5.15-8.817 and cat meowing at 0.688-1.697, 2.233-3.242, 3.778-4.787", "frequencyCaption": "thump thud two times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4989.wav", "onoffCaption": "sheep goat bleating at 0.79-2.79 and woman laughing at 1.659-4.251, 4.803-7.395 and gunshot at 3.514-5.514, 6.423-8.696", "frequencyCaption": "sheep goat bleating one times and woman laughing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2028.wav", "onoffCaption": "tapping clicking clanking at 3.282-6.722", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2032.wav", "onoffCaption": "thump thud at 1.878-4.217", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2129.wav", "onoffCaption": "cat meowing at 2.578-3.663, 4.726-6.036, 7.089-8.089", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2133.wav", "onoffCaption": "train horn at 3.33-6.57, 7.334-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2166.wav", "onoffCaption": "cow mooing at 2.414-6.843, 7.633-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2241.wav", "onoffCaption": "whistling at 3.139-8.314", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2340.wav", "onoffCaption": "car horn honking at 0.508-4.349 and explosion at 7.279-9.281", "frequencyCaption": "car horn honking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2464.wav", "onoffCaption": "tapping clicking clanking at 0.896-4.336 and duck quacking at 1.153-3.153, 4.089-6.089, 6.905-8.905", "frequencyCaption": "tapping clicking clanking one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2481.wav", "onoffCaption": "spraying at 1.831-2.331 and door slamming at 4.61-6.61, 7.521-9.521", "frequencyCaption": "spraying one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2530.wav", "onoffCaption": "train horn at 3.401-6.641", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2565.wav", "onoffCaption": "sneeze at 0.005-1.505", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2580.wav", "onoffCaption": "gunshot at 3.815-5.815, 7.3-9.3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2617.wav", "onoffCaption": "duck quacking at 1.56-3.56, 4.821-6.821", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2716.wav", "onoffCaption": "sheep goat bleating at 0.826-2.826 and thump thud at 5.91-8.41", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2861.wav", "onoffCaption": "sneeze at 2.056-7.056 and tapping clicking clanking at 4.55-7.99 and thump thud at 5.027-7.798", "frequencyCaption": "sneeze one times and tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2884.wav", "onoffCaption": "door knocking at 0.648-2.951, 4.123-6.605, 7.909-9.931", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2960.wav", "onoffCaption": "woman laughing at 0.175-2.257 and door slamming at 0.823-2.823 and sneeze at 6.102-7.559", "frequencyCaption": "woman laughing one times and door slamming one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2985.wav", "onoffCaption": "thump thud at 0.261-2.6, 3.314-5.653 and sneeze at 4.087-6.172", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4071.wav", "onoffCaption": "cat meowing at 2.995-4.097, 4.741-5.843", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4094.wav", "onoffCaption": "whistling at 3.031-9.342", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4170.wav", "onoffCaption": "thump thud at 1.319-5.237 and cat meowing at 4.051-5.587", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4195.wav", "onoffCaption": "whistling at 2.822-5.051, 5.722-8.376 and explosion at 6.118-8.836", "frequencyCaption": "whistling two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4202.wav", "onoffCaption": "whistling at 2.434-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4218.wav", "onoffCaption": "cow mooing at 0.267-3.565 and woman laughing at 1.052-3.657", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4303.wav", "onoffCaption": "tapping clicking clanking at 0.66-4.1, 6.49-9.93 and cat meowing at 1.059-2.603, 4.596-6.14 and sheep goat bleating at 3.085-5.085", "frequencyCaption": "tapping clicking clanking two times and cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4319.wav", "onoffCaption": "door slamming at 3.854-5.003, 5.659-7.885", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4427.wav", "onoffCaption": "sheep goat bleating at 3.094-5.094, 6.139-8.139", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4526.wav", "onoffCaption": "spraying at 0.254-1.335, 3.355-5.05", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4654.wav", "onoffCaption": "cat meowing at 3.661-6.565, 7.709-8.895", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4755.wav", "onoffCaption": "cow mooing at 4.05-7.019 and spraying at 6.688-7.769, 9.086-9.713", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4822.wav", "onoffCaption": "train horn at 3.056-5.523 and car horn honking at 3.381-5.846 and burping belching at 4.458-7.458", "frequencyCaption": "train horn one times and car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4838.wav", "onoffCaption": "door knocking at 0.085-2.306 and spraying at 5.037-5.621 and duck quacking at 5.649-7.649", "frequencyCaption": "door knocking one times and spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4923.wav", "onoffCaption": "gunshot at 0.444-2.444, 3.118-5.248 and explosion at 2.474-7.474", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4939.wav", "onoffCaption": "gunshot at 2.558-5.032, 5.941-7.941", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2016.wav", "onoffCaption": "whistling at 0.036-5.211 and burping belching at 2.935-5.058, 6.513-9.513", "frequencyCaption": "whistling one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2067.wav", "onoffCaption": "sneeze at 0.627-2.69", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2082.wav", "onoffCaption": "spraying at 2.94-5.959", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2158.wav", "onoffCaption": "whistling at 0.357-8.012 and gunshot at 3.468-5.468, 7.306-9.306", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2199.wav", "onoffCaption": "thump thud at 3.283-6.95", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2214.wav", "onoffCaption": "cat meowing at 0.151-1.417, 3.482-4.748", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2265.wav", "onoffCaption": "whistling at 1.369-5.853 and tapping clicking clanking at 2.623-6.063, 7.495-10.0", "frequencyCaption": "whistling one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2280.wav", "onoffCaption": "burping belching at 0.359-5.359, 6.238-8.564 and tapping clicking clanking at 3.033-6.473", "frequencyCaption": "burping belching two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2415.wav", "onoffCaption": "car horn honking at 0.386-3.881, 5.676-9.171", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2431.wav", "onoffCaption": "explosion at 0.062-5.062, 7.284-9.469", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2440.wav", "onoffCaption": "door slamming at 0.123-0.623, 2.142-2.642 and sheep goat bleating at 1.348-3.348, 3.893-5.893, 6.837-8.837", "frequencyCaption": "door slamming two times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2633.wav", "onoffCaption": "whistling at 0.759-3.634, 4.275-6.577", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2642.wav", "onoffCaption": "tapping clicking clanking at 1.34-4.78, 7.05-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2728.wav", "onoffCaption": "cat meowing at 0.493-3.523 and woman laughing at 0.638-2.73", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2759.wav", "onoffCaption": "gunshot at 1.592-3.611", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2935.wav", "onoffCaption": "door knocking at 0.728-4.478 and duck quacking at 2.452-4.452, 5.784-7.784", "frequencyCaption": "door knocking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2944.wav", "onoffCaption": "door slamming at 3.33-5.33", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4024.wav", "onoffCaption": "car horn honking at 2.651-5.164, 5.914-8.427 and thump thud at 3.286-6.057, 7.195-9.966", "frequencyCaption": "car horn honking two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4055.wav", "onoffCaption": "tapping clicking clanking at 2.017-5.457, 7.274-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4226.wav", "onoffCaption": "tapping clicking clanking at 0.981-4.421, 5.748-8.421", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4257.wav", "onoffCaption": "cow mooing at 2.78-5.762", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4296.wav", "onoffCaption": "cat meowing at 0.666-2.626, 3.341-5.301", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4403.wav", "onoffCaption": "door slamming at 2.323-3.626, 4.97-6.273, 7.601-8.904", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4472.wav", "onoffCaption": "door knocking at 2.045-4.422, 5.41-7.787", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4497.wav", "onoffCaption": "whistling at 2.277-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4518.wav", "onoffCaption": "spraying at 3.256-4.503, 5.811-7.058, 8.236-9.483", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4569.wav", "onoffCaption": "train horn at 2.431-5.965, 7.28-9.954", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4601.wav", "onoffCaption": "cow mooing at 0.4-5.38, 7.824-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4670.wav", "onoffCaption": "whistling at 3.544-9.044", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4695.wav", "onoffCaption": "burping belching at 0.668-4.668, 6.056-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4888.wav", "onoffCaption": "burping belching at 3.083-6.083", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4907.wav", "onoffCaption": "whistling at 1.293-9.304", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4976.wav", "onoffCaption": "tapping clicking clanking at 0.552-3.992, 5.863-9.303", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4993.wav", "onoffCaption": "train horn at 0.195-2.869 and cat meowing at 6.247-7.291, 8.178-9.488", "frequencyCaption": "train horn one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2043.wav", "onoffCaption": "cat meowing at 3.812-5.752", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2059.wav", "onoffCaption": "burping belching at 0.09-2.197 and door knocking at 0.721-3.788, 5.023-8.09", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2142.wav", "onoffCaption": "door slamming at 0.363-1.616, 2.579-4.805, 6.932-8.223", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2230.wav", "onoffCaption": "train horn at 0.572-6.632", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2331.wav", "onoffCaption": "thump thud at 0.794-3.022, 3.604-5.832, 6.585-8.813", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2364.wav", "onoffCaption": "explosion at 2.299-5.293, 6.682-9.682", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2381.wav", "onoffCaption": "woman laughing at 0.245-2.327, 2.828-5.334, 7.526-9.642", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_2514.wav", "onoffCaption": "gunshot at 3.315-5.315, 7.012-9.012", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2629.wav", "onoffCaption": "spraying at 0.982-2.715, 4.089-5.822 and cow mooing at 1.108-6.088, 7.379-10.0", "frequencyCaption": "spraying two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2666.wav", "onoffCaption": "thump thud at 0.125-2.896, 4.014-6.785, 7.495-10.0 and woman laughing at 2.778-4.978", "frequencyCaption": "thump thud three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2683.wav", "onoffCaption": "train horn at 2.278-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2699.wav", "onoffCaption": "sheep goat bleating at 3.383-5.383, 7.401-9.401", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2732.wav", "onoffCaption": "cow mooing at 3.624-8.604", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2767.wav", "onoffCaption": "dog barking at 2.432-4.432, 5.383-7.383", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2782.wav", "onoffCaption": "door knocking at 0.89-4.01, 5.434-7.971 and burping belching at 4.746-7.072", "frequencyCaption": "door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2798.wav", "onoffCaption": "thump thud at 0.059-2.559, 3.231-5.68, 6.442-8.781 and spraying at 2.588-3.096", "frequencyCaption": "thump thud three times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2810.wav", "onoffCaption": "door slamming at 2.837-3.642", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2911.wav", "onoffCaption": "gunshot at 1.534-3.704, 6.11-8.11", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4000.wav", "onoffCaption": "burping belching at 2.612-6.118, 6.622-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4101.wav", "onoffCaption": "whistling at 0.049-7.142 and door slamming at 1.786-3.786, 5.681-7.681", "frequencyCaption": "whistling one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4269.wav", "onoffCaption": "car horn honking at 0.1-2.565", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4273.wav", "onoffCaption": "woman laughing at 3.317-5.672", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4368.wav", "onoffCaption": "sneeze at 3.702-5.615, 6.946-8.859", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4372.wav", "onoffCaption": "cat meowing at 1.592-2.592, 3.323-4.333, 4.964-7.647", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4397.wav", "onoffCaption": "dog barking at 0.044-2.444 and cow mooing at 6.758-10.0", "frequencyCaption": "dog barking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4419.wav", "onoffCaption": "car horn honking at 3.16-5.946, 6.874-9.66", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4456.wav", "onoffCaption": "cat meowing at 3.282-4.592, 6.073-8.963 and woman laughing at 3.328-5.933", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4557.wav", "onoffCaption": "car horn honking at 0.072-3.913, 6.076-9.917 and duck quacking at 0.139-2.139, 3.74-5.74", "frequencyCaption": "car horn honking two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4625.wav", "onoffCaption": "woman laughing at 0.112-3.5, 4.993-7.276 and cow mooing at 2.294-7.274", "frequencyCaption": "woman laughing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4724.wav", "onoffCaption": "whistling at 3.107-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4806.wav", "onoffCaption": "explosion at 1.533-6.399", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4849.wav", "onoffCaption": "burping belching at 2.476-5.237, 6.847-9.668", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4853.wav", "onoffCaption": "car horn honking at 2.537-4.884, 6.616-8.963", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4948.wav", "onoffCaption": "woman laughing at 0.01-3.11 and door slamming at 0.11-2.593, 4.13-7.091", "frequencyCaption": "woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4952.wav", "onoffCaption": "tapping clicking clanking at 2.225-5.665", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2110.wav", "onoffCaption": "whistling at 0.11-8.121 and gunshot at 0.66-2.66, 4.32-6.32", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2117.wav", "onoffCaption": "duck quacking at 2.303-4.303, 4.821-6.821, 7.375-9.375", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2145.wav", "onoffCaption": "car horn honking at 2.249-5.067, 6.86-9.325", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2278.wav", "onoffCaption": "car horn honking at 0.005-3.18 and cat meowing at 0.584-2.532 and sneeze at 1.507-2.784", "frequencyCaption": "car horn honking one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2336.wav", "onoffCaption": "burping belching at 2.727-6.75, 7.974-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2363.wav", "onoffCaption": "explosion at 0.209-2.937, 4.197-6.925", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2408.wav", "onoffCaption": "duck quacking at 0.919-2.919, 4.192-6.192, 6.918-8.918", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2513.wav", "onoffCaption": "tapping clicking clanking at 1.575-5.015, 5.631-7.968", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2541.wav", "onoffCaption": "door knocking at 2.431-7.431 and spraying at 3.38-3.984, 6.273-8.401", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2546.wav", "onoffCaption": "car horn honking at 0.134-3.629, 4.572-6.776", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2735.wav", "onoffCaption": "train horn at 2.719-7.049", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2760.wav", "onoffCaption": "woman laughing at 0.243-2.829 and whistling at 6.242-10.0", "frequencyCaption": "woman laughing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2785.wav", "onoffCaption": "sheep goat bleating at 0.364-3.364 and gunshot at 2.49-4.49, 5.326-7.326 and thump thud at 4.036-6.375, 7.809-10.0", "frequencyCaption": "sheep goat bleating one times and gunshot two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2817.wav", "onoffCaption": "door slamming at 0.034-2.162, 3.44-6.414", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2842.wav", "onoffCaption": "woman laughing at 3.832-6.398, 7.087-9.653 and thump thud at 3.997-7.664", "frequencyCaption": "woman laughing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2845.wav", "onoffCaption": "burping belching at 0.066-2.189 and dog barking at 4.94-6.94, 7.927-9.927", "frequencyCaption": "burping belching one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2959.wav", "onoffCaption": "thump thud at 2.464-6.131", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4106.wav", "onoffCaption": "burping belching at 2.396-5.396, 6.29-9.29", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4154.wav", "onoffCaption": "burping belching at 0.1-3.38 and duck quacking at 4.7-6.7", "frequencyCaption": "burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4320.wav", "onoffCaption": "woman laughing at 1.992-4.347 and sheep goat bleating at 2.326-4.326, 6.491-8.491", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4327.wav", "onoffCaption": "tapping clicking clanking at 0.526-3.966, 4.575-8.015", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4375.wav", "onoffCaption": "car horn honking at 1.515-5.837, 7.812-9.812", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4390.wav", "onoffCaption": "door knocking at 0.626-2.978, 3.775-6.087, 6.804-9.392", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4502.wav", "onoffCaption": "duck quacking at 2.496-4.496 and car horn honking at 2.541-5.041, 5.542-8.042", "frequencyCaption": "duck quacking one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4550.wav", "onoffCaption": "gunshot at 0.379-2.379, 3.425-5.425 and cat meowing at 6.709-9.613", "frequencyCaption": "gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4638.wav", "onoffCaption": "tapping clicking clanking at 2.149-5.589, 6.582-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4688.wav", "onoffCaption": "train horn at 1.924-4.564", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4723.wav", "onoffCaption": "train horn at 0.301-8.501 and spraying at 1.872-2.813, 3.956-5.718, 6.281-7.062", "frequencyCaption": "train horn one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4771.wav", "onoffCaption": "thump thud at 2.044-6.494, 7.6-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4776.wav", "onoffCaption": "burping belching at 1.744-5.288, 7.129-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4793.wav", "onoffCaption": "duck quacking at 0.789-2.789, 3.91-5.91", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4794.wav", "onoffCaption": "thump thud at 3.096-7.471", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4801.wav", "onoffCaption": "sheep goat bleating at 2.842-4.842, 5.542-7.542", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4854.wav", "onoffCaption": "sneeze at 0.166-1.711, 3.091-4.185", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2011.wav", "onoffCaption": "cow mooing at 3.455-7.884", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2044.wav", "onoffCaption": "cow mooing at 2.752-6.05", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2237.wav", "onoffCaption": "spraying at 2.505-3.68, 5.61-8.07 and burping belching at 4.712-7.335", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2246.wav", "onoffCaption": "train horn at 2.386-4.866, 5.514-7.994", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2262.wav", "onoffCaption": "door slamming at 0.296-1.82, 2.403-3.927, 4.953-6.477", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2287.wav", "onoffCaption": "sneeze at 2.468-4.08, 4.643-6.255, 7.115-8.727", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2379.wav", "onoffCaption": "thump thud at 1.14-3.602, 4.287-6.749", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2386.wav", "onoffCaption": "sneeze at 0.998-3.401, 4.07-5.347", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2412.wav", "onoffCaption": "whistling at 2.899-8.074", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2447.wav", "onoffCaption": "explosion at 0.429-5.429, 6.245-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2509.wav", "onoffCaption": "car horn honking at 0.405-3.9, 5.336-7.484", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2610.wav", "onoffCaption": "spraying at 0.285-2.721, 3.874-6.31, 7.014-9.45", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2634.wav", "onoffCaption": "cow mooing at 1.887-6.316 and dog barking at 2.493-4.493, 5.504-7.504", "frequencyCaption": "cow mooing one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2661.wav", "onoffCaption": "spraying at 0.128-1.823, 2.851-4.337, 5.4-6.05 and duck quacking at 1.565-3.565, 5.072-7.072 and dog barking at 1.874-3.874", "frequencyCaption": "spraying three times and duck quacking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2684.wav", "onoffCaption": "sheep goat bleating at 0.422-2.422", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2858.wav", "onoffCaption": "spraying at 1.549-2.281, 3.504-4.236, 4.962-5.694", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2916.wav", "onoffCaption": "train horn at 1.358-4.598, 5.88-9.12", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2943.wav", "onoffCaption": "door slamming at 2.913-4.913, 6.059-8.059", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4007.wav", "onoffCaption": "train horn at 0.741-4.221, 6.423-9.063 and cat meowing at 2.233-3.593, 5.797-7.692", "frequencyCaption": "train horn two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4048.wav", "onoffCaption": "woman laughing at 3.569-6.364 and sneeze at 4.859-6.855", "frequencyCaption": "woman laughing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4052.wav", "onoffCaption": "duck quacking at 2.19-4.19, 6.433-8.433", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4076.wav", "onoffCaption": "thump thud at 0.403-3.174 and sheep goat bleating at 2.8-4.8, 6.068-8.068", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4093.wav", "onoffCaption": "door knocking at 3.703-5.924, 7.166-9.387", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4149.wav", "onoffCaption": "whistling at 3.415-9.124", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4153.wav", "onoffCaption": "door knocking at 0.839-3.569, 4.834-7.564 and sneeze at 1.952-4.037", "frequencyCaption": "door knocking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4188.wav", "onoffCaption": "duck quacking at 2.14-4.14", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4221.wav", "onoffCaption": "whistling at 0.086-8.471", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4274.wav", "onoffCaption": "cat meowing at 0.042-1.144, 2.589-3.598 and woman laughing at 0.866-3.452, 4.245-6.831", "frequencyCaption": "cat meowing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4291.wav", "onoffCaption": "thump thud at 1.138-5.588 and whistling at 1.843-6.327 and woman laughing at 5.378-7.797", "frequencyCaption": "thump thud one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4404.wav", "onoffCaption": "burping belching at 2.51-5.51, 7.101-9.541 and gunshot at 2.836-4.836", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4420.wav", "onoffCaption": "duck quacking at 2.843-4.843, 5.486-7.486 and thump thud at 5.54-8.311", "frequencyCaption": "duck quacking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4451.wav", "onoffCaption": "train horn at 0.239-3.039, 5.026-7.426", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4505.wav", "onoffCaption": "door knocking at 1.084-3.654", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4622.wav", "onoffCaption": "whistling at 2.926-5.155, 7.141-9.447", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4677.wav", "onoffCaption": "cow mooing at 3.047-7.476 and woman laughing at 3.474-6.269", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4692.wav", "onoffCaption": "duck quacking at 0.013-2.013", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4739.wav", "onoffCaption": "door slamming at 1.895-3.813", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4789.wav", "onoffCaption": "sneeze at 1.002-5.058", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4900.wav", "onoffCaption": "cow mooing at 0.223-3.233, 4.638-7.648 and door knocking at 4.601-6.761", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4924.wav", "onoffCaption": "burping belching at 0.009-5.953", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4955.wav", "onoffCaption": "door knocking at 0.394-5.096, 6.053-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_8.wav", "onoffCaption": "sheep goat bleating at 0.236-2.236, 3.143-5.143, 6.147-8.147", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2035.wav", "onoffCaption": "thump thud at 3.809-6.148, 7.831-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2060.wav", "onoffCaption": "sheep goat bleating at 0.166-5.046, 6.154-8.154 and explosion at 0.486-2.66, 3.478-6.478", "frequencyCaption": "sheep goat bleating two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2085.wav", "onoffCaption": "burping belching at 3.03-5.395, 5.899-8.216", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2161.wav", "onoffCaption": "door knocking at 2.239-6.511", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2184.wav", "onoffCaption": "woman laughing at 2.236-4.473, 6.682-8.919 and sneeze at 2.47-6.97 and sheep goat bleating at 6.116-8.116", "frequencyCaption": "woman laughing two times and sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2213.wav", "onoffCaption": "duck quacking at 2.521-4.521, 6.283-8.283", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2308.wav", "onoffCaption": "cow mooing at 0.549-3.531, 5.902-8.609", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2436.wav", "onoffCaption": "thump thud at 3.198-6.865", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2463.wav", "onoffCaption": "whistling at 1.38-9.391", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2486.wav", "onoffCaption": "gunshot at 1.07-3.343", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2537.wav", "onoffCaption": "train horn at 2.728-7.586", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2578.wav", "onoffCaption": "gunshot at 2.509-4.509, 6.605-8.605", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2645.wav", "onoffCaption": "sheep goat bleating at 0.42-2.42, 4.869-6.869 and cat meowing at 2.605-3.915, 4.495-6.631", "frequencyCaption": "sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2829.wav", "onoffCaption": "whistling at 0.202-5.377", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2833.wav", "onoffCaption": "sneeze at 2.403-6.051, 7.994-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2899.wav", "onoffCaption": "door knocking at 0.036-2.66, 4.527-7.364 and dog barking at 0.341-2.341, 2.973-4.973", "frequencyCaption": "door knocking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2928.wav", "onoffCaption": "burping belching at 0.293-6.973 and woman laughing at 2.69-4.806, 5.463-7.568", "frequencyCaption": "burping belching one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2932.wav", "onoffCaption": "spraying at 0.047-0.651, 1.26-2.164, 3.361-4.608 and dog barking at 0.972-2.972, 3.608-5.608", "frequencyCaption": "spraying three times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2967.wav", "onoffCaption": "cow mooing at 2.971-5.94, 7.066-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2982.wav", "onoffCaption": "door knocking at 2.719-5.099", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4023.wav", "onoffCaption": "door knocking at 1.077-4.577, 5.819-7.908", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4039.wav", "onoffCaption": "thump thud at 2.056-4.556, 6.038-8.538", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4122.wav", "onoffCaption": "cat meowing at 3.331-4.342, 4.92-5.931, 6.61-7.621", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4138.wav", "onoffCaption": "door knocking at 2.83-5.182, 6.312-8.664", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4205.wav", "onoffCaption": "spraying at 0.092-2.484, 4.849-7.241", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4250.wav", "onoffCaption": "burping belching at 0.753-3.932", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4351.wav", "onoffCaption": "dog barking at 0.601-2.601, 5.002-7.002", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4475.wav", "onoffCaption": "burping belching at 1.004-7.684 and whistling at 1.432-5.916, 6.702-9.65 and gunshot at 2.55-4.55, 5.582-7.582", "frequencyCaption": "burping belching one times and whistling two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4490.wav", "onoffCaption": "door knocking at 2.648-4.775 and tapping clicking clanking at 7.088-10.0", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4574.wav", "onoffCaption": "sneeze at 4.1-6.346", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4606.wav", "onoffCaption": "thump thud at 1.232-4.003, 6.177-8.891", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4653.wav", "onoffCaption": "tapping clicking clanking at 2.647-6.087", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4707.wav", "onoffCaption": "explosion at 0.051-5.051", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4748.wav", "onoffCaption": "thump thud at 1.279-3.779, 5.309-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4895.wav", "onoffCaption": "gunshot at 3.74-6.214, 7.578-9.578", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4971.wav", "onoffCaption": "cat meowing at 0.262-2.256", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4994.wav", "onoffCaption": "cat meowing at 3.746-4.746, 5.536-6.536 and sheep goat bleating at 4.329-6.329", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2048.wav", "onoffCaption": "burping belching at 0.76-4.304, 5.102-8.646", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2134.wav", "onoffCaption": "spraying at 0.015-1.19, 2.932-5.368", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2153.wav", "onoffCaption": "spraying at 0.046-1.046 and explosion at 0.666-5.666, 6.451-10.0", "frequencyCaption": "spraying one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2209.wav", "onoffCaption": "gunshot at 2.013-4.013", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2312.wav", "onoffCaption": "explosion at 2.43-7.351", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2347.wav", "onoffCaption": "explosion at 0.017-3.017, 3.938-6.938 and dog barking at 1.463-3.463", "frequencyCaption": "explosion two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2479.wav", "onoffCaption": "burping belching at 1.689-7.776", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2505.wav", "onoffCaption": "thump thud at 1.734-6.109, 6.786-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2562.wav", "onoffCaption": "car horn honking at 0.787-5.694, 7.691-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2587.wav", "onoffCaption": "sheep goat bleating at 3.286-5.286", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2711.wav", "onoffCaption": "train horn at 2.669-4.824, 6.196-8.66", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2744.wav", "onoffCaption": "car horn honking at 0.047-2.973 and cat meowing at 6.678-8.219", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2801.wav", "onoffCaption": "burping belching at 1.71-5.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2866.wav", "onoffCaption": "cat meowing at 0.382-2.57, 3.17-4.181, 4.892-6.078 and whistling at 0.515-8.9 and duck quacking at 5.413-7.413", "frequencyCaption": "cat meowing three times and whistling one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2883.wav", "onoffCaption": "sheep goat bleating at 3.204-5.204", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2998.wav", "onoffCaption": "spraying at 0.088-1.345, 3.612-4.869", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4089.wav", "onoffCaption": "sneeze at 0.337-1.591 and train horn at 3.809-9.524", "frequencyCaption": "sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4177.wav", "onoffCaption": "dog barking at 0.177-2.177 and sneeze at 0.871-3.485, 5.324-7.465 and woman laughing at 3.376-6.014, 7.273-10.0", "frequencyCaption": "dog barking one times and sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4192.wav", "onoffCaption": "woman laughing at 0.638-3.433, 5.851-8.646", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4278.wav", "onoffCaption": "gunshot at 1.428-3.521, 4.261-6.354, 7.169-9.262 and door slamming at 5.994-7.369", "frequencyCaption": "gunshot three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4304.wav", "onoffCaption": "train horn at 2.458-6.639, 7.492-9.932", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4363.wav", "onoffCaption": "thump thud at 0.301-2.64 and duck quacking at 0.65-2.65, 4.552-6.552", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4386.wav", "onoffCaption": "sneeze at 0.984-4.059, 5.924-8.999", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4521.wav", "onoffCaption": "sneeze at 0.704-2.663, 3.814-5.773", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4546.wav", "onoffCaption": "dog barking at 0.715-2.715, 5.2-7.2", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4591.wav", "onoffCaption": "sheep goat bleating at 2.507-4.507, 5.973-7.973", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4649.wav", "onoffCaption": "spraying at 0.22-1.087, 2.362-3.229 and cat meowing at 6.586-8.526", "frequencyCaption": "spraying two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4735.wav", "onoffCaption": "train horn at 0.231-5.308", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4752.wav", "onoffCaption": "cow mooing at 1.023-3.992, 4.756-7.738 and cat meowing at 2.326-3.87, 4.908-6.452", "frequencyCaption": "cow mooing two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4825.wav", "onoffCaption": "burping belching at 0.564-5.564, 7.134-9.46", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4870.wav", "onoffCaption": "thump thud at 1.655-6.03, 6.852-9.508", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2106.wav", "onoffCaption": "car horn honking at 1.789-4.302, 5.357-7.87", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2274.wav", "onoffCaption": "explosion at 0.13-5.051, 6.09-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2291.wav", "onoffCaption": "cat meowing at 1.616-2.628, 3.395-6.299, 7.041-8.143", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2320.wav", "onoffCaption": "gunshot at 0.53-2.803, 3.85-5.85, 7.066-9.066", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2375.wav", "onoffCaption": "door slamming at 1.481-2.872, 3.714-5.714, 6.695-7.844 and explosion at 5.251-8.438", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2390.wav", "onoffCaption": "duck quacking at 0.077-2.077, 4.153-6.153", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2451.wav", "onoffCaption": "burping belching at 0.021-3.311 and cat meowing at 5.102-7.042", "frequencyCaption": "burping belching one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2550.wav", "onoffCaption": "train horn at 2.524-8.268", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2622.wav", "onoffCaption": "tapping clicking clanking at 0.613-4.053 and train horn at 0.753-3.393, 4.437-7.077", "frequencyCaption": "tapping clicking clanking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2638.wav", "onoffCaption": "gunshot at 0.324-2.324, 3.995-6.501 and whistling at 6.819-8.828", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2688.wav", "onoffCaption": "dog barking at 0.336-2.336, 3.264-5.264, 5.967-7.967 and tapping clicking clanking at 2.286-5.726", "frequencyCaption": "dog barking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2723.wav", "onoffCaption": "spraying at 0.799-1.974, 2.594-3.178, 4.428-6.19 and woman laughing at 2.694-4.977, 6.495-9.48", "frequencyCaption": "spraying three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2739.wav", "onoffCaption": "spraying at 1.824-2.451, 3.203-4.898, 7.269-8.516", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2776.wav", "onoffCaption": "sheep goat bleating at 2.702-4.702, 6.792-8.792", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2793.wav", "onoffCaption": "spraying at 2.441-3.068, 5.222-5.849 and explosion at 7.549-10.0", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2854.wav", "onoffCaption": "cow mooing at 2.28-7.26", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2955.wav", "onoffCaption": "sneeze at 3.434-4.82, 5.419-6.805", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4044.wav", "onoffCaption": "duck quacking at 1.493-3.493, 4.339-6.339, 7.456-9.456", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4110.wav", "onoffCaption": "tapping clicking clanking at 1.135-4.575, 5.943-8.406", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4145.wav", "onoffCaption": "sheep goat bleating at 3.411-5.411, 6.238-8.238", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4237.wav", "onoffCaption": "spraying at 0.24-1.726", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4336.wav", "onoffCaption": "burping belching at 2.645-5.824, 7.891-9.922", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4408.wav", "onoffCaption": "explosion at 1.757-4.757, 5.479-8.479 and car horn honking at 6.199-8.199", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4412.wav", "onoffCaption": "cat meowing at 0.27-1.63", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4509.wav", "onoffCaption": "woman laughing at 0.787-4.363, 6.706-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4513.wav", "onoffCaption": "gunshot at 0.752-2.752, 3.852-5.852, 6.881-8.881", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4661.wav", "onoffCaption": "thump thud at 1.443-3.782, 5.47-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4684.wav", "onoffCaption": "door slamming at 2.704-4.079, 6.287-7.811", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4760.wav", "onoffCaption": "tapping clicking clanking at 0.533-3.973, 5.147-8.587", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4785.wav", "onoffCaption": "whistling at 0.652-8.663", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4817.wav", "onoffCaption": "train horn at 0.488-4.488 and sheep goat bleating at 0.833-2.833, 3.373-5.373, 5.952-7.952", "frequencyCaption": "train horn one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4842.wav", "onoffCaption": "tapping clicking clanking at 2.652-6.092 and whistling at 2.885-8.385 and sheep goat bleating at 6.014-8.014", "frequencyCaption": "tapping clicking clanking one times and whistling one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4916.wav", "onoffCaption": "duck quacking at 2.204-4.204, 6.278-8.278", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4959.wav", "onoffCaption": "cat meowing at 0.048-1.06, 1.852-2.864, 4.697-5.709 and train horn at 0.069-3.549, 4.338-7.276", "frequencyCaption": "cat meowing three times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4.wav", "onoffCaption": "spraying at 0.375-2.503 and car horn honking at 4.549-8.136", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2007.wav", "onoffCaption": "train horn at 3.118-6.278, 7.28-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2023.wav", "onoffCaption": "car horn honking at 4.143-7.638", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2052.wav", "onoffCaption": "spraying at 2.342-2.992 and cat meowing at 3.473-4.483, 5.231-6.258, 7.243-8.553", "frequencyCaption": "spraying one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2076.wav", "onoffCaption": "spraying at 0.727-1.508 and thump thud at 5.495-9.413", "frequencyCaption": "spraying one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2089.wav", "onoffCaption": "duck quacking at 0.81-2.81 and spraying at 3.644-4.144, 5.805-6.656", "frequencyCaption": "duck quacking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2093.wav", "onoffCaption": "sneeze at 0.898-2.175, 3.979-5.305 and spraying at 3.016-3.524, 4.306-4.814", "frequencyCaption": "sneeze two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2138.wav", "onoffCaption": "burping belching at 1.702-6.038, 7.211-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2149.wav", "onoffCaption": "door knocking at 1.85-6.683, 7.825-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2188.wav", "onoffCaption": "train horn at 3.407-9.876", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2205.wav", "onoffCaption": "thump thud at 3.709-6.171", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2221.wav", "onoffCaption": "woman laughing at 2.818-5.89, 7.129-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2250.wav", "onoffCaption": "explosion at 2.937-7.937", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2404.wav", "onoffCaption": "tapping clicking clanking at 2.396-5.836, 7.121-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2420.wav", "onoffCaption": "cat meowing at 0.79-2.15 and woman laughing at 5.494-7.61", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2475.wav", "onoffCaption": "door knocking at 3.062-5.374, 6.46-8.772 and woman laughing at 3.618-6.037, 7.169-9.518", "frequencyCaption": "door knocking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2490.wav", "onoffCaption": "cow mooing at 2.202-5.171, 6.377-8.411", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2606.wav", "onoffCaption": "door knocking at 2.842-5.61", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2653.wav", "onoffCaption": "cat meowing at 3.516-4.601", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2677.wav", "onoffCaption": "whistling at 0.283-4.767, 5.551-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2692.wav", "onoffCaption": "explosion at 0.361-3.417, 5.348-8.404", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2748.wav", "onoffCaption": "thump thud at 0.302-2.802, 4.125-6.625 and dog barking at 0.683-2.683", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2789.wav", "onoffCaption": "explosion at 1.018-4.571", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2825.wav", "onoffCaption": "door knocking at 0.964-4.732", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2900.wav", "onoffCaption": "train horn at 0.446-3.12, 4.402-6.539 and thump thud at 3.737-7.404", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2924.wav", "onoffCaption": "burping belching at 0.202-6.882", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2971.wav", "onoffCaption": "burping belching at 1.991-6.39, 7.29-9.52", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2994.wav", "onoffCaption": "tapping clicking clanking at 2.3-5.74, 6.598-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4011.wav", "onoffCaption": "sneeze at 2.458-3.561, 4.16-6.297", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4035.wav", "onoffCaption": "car horn honking at 3.885-6.703, 7.76-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4060.wav", "onoffCaption": "gunshot at 2.22-4.22, 4.777-6.777, 7.486-9.486", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4085.wav", "onoffCaption": "dog barking at 0.613-2.613, 3.151-5.151, 6.447-8.447", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4213.wav", "onoffCaption": "explosion at 0.298-5.298, 5.925-8.075", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4246.wav", "onoffCaption": "duck quacking at 1.745-3.745, 4.943-6.943", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4262.wav", "onoffCaption": "car horn honking at 0.025-2.025 and spraying at 3.726-4.353, 6.476-7.103, 8.766-9.393", "frequencyCaption": "car horn honking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4287.wav", "onoffCaption": "door knocking at 2.855-5.207", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4308.wav", "onoffCaption": "explosion at 0.564-5.564, 7.629-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4379.wav", "onoffCaption": "dog barking at 0.247-2.247", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4436.wav", "onoffCaption": "gunshot at 3.193-5.433, 7.188-9.188", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4447.wav", "onoffCaption": "thump thud at 3.806-7.724", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4463.wav", "onoffCaption": "tapping clicking clanking at 0.406-3.846, 6.106-8.202", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4486.wav", "onoffCaption": "car horn honking at 3.658-6.584, 7.528-9.535", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4578.wav", "onoffCaption": "car horn honking at 1.804-5.023", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4610.wav", "onoffCaption": "sneeze at 0.396-3.604, 4.394-7.602", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4634.wav", "onoffCaption": "door slamming at 0.545-1.778 and duck quacking at 4.483-6.483", "frequencyCaption": "door slamming one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4645.wav", "onoffCaption": "door slamming at 0.033-1.324, 2.029-3.32, 4.632-5.923", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4829.wav", "onoffCaption": "car horn honking at 0.514-3.427, 4.395-7.321, 7.878-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4858.wav", "onoffCaption": "train horn at 0.62-4.688, 5.25-9.318", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4899.wav", "onoffCaption": "explosion at 2.455-7.455", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4932.wav", "onoffCaption": "burping belching at 0.383-3.927, 5.815-9.359", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4943.wav", "onoffCaption": "explosion at 0.369-3.363, 4.428-7.086 and cow mooing at 1.762-5.06", "frequencyCaption": "explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4967.wav", "onoffCaption": "door knocking at 1.44-5.59, 6.272-8.575 and cat meowing at 2.072-3.285, 3.8-5.013", "frequencyCaption": "door knocking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4982.wav", "onoffCaption": "door knocking at 0.035-2.535, 3.47-5.97, 6.806-9.306 and door slamming at 0.149-0.83, 1.334-2.185", "frequencyCaption": "door knocking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2039.wav", "onoffCaption": "explosion at 3.189-7.029", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2122.wav", "onoffCaption": "spraying at 2.788-3.639, 4.382-5.233", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2177.wav", "onoffCaption": "cat meowing at 0.579-2.715 and door knocking at 1.119-4.672, 5.375-8.928", "frequencyCaption": "cat meowing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2192.wav", "onoffCaption": "whistling at 2.771-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2304.wav", "onoffCaption": "cat meowing at 0.014-1.55, 2.249-3.785, 5.054-6.59 and door knocking at 0.12-2.341, 4.211-6.432", "frequencyCaption": "cat meowing three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2351.wav", "onoffCaption": "train horn at 0.696-5.136 and door knocking at 6.603-9.978", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2521.wav", "onoffCaption": "door knocking at 0.54-6.6, 7.245-9.422", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2574.wav", "onoffCaption": "train horn at 0.214-5.072, 7.224-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2591.wav", "onoffCaption": "car horn honking at 0.359-3.145", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2649.wav", "onoffCaption": "car horn honking at 0.111-4.511 and thump thud at 2.579-5.626 and door slamming at 2.596-3.596, 4.32-5.3, 6.247-7.622", "frequencyCaption": "car horn honking one times and thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2707.wav", "onoffCaption": "whistling at 1.983-6.467 and gunshot at 3.437-5.437, 5.946-7.946", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2752.wav", "onoffCaption": "burping belching at 0.679-6.28", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2870.wav", "onoffCaption": "door knocking at 2.203-5.259, 7.501-9.947", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2895.wav", "onoffCaption": "cow mooing at 0.668-3.65, 4.571-7.553 and dog barking at 2.804-4.804, 6.928-9.366", "frequencyCaption": "cow mooing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4134.wav", "onoffCaption": "tapping clicking clanking at 2.236-5.676", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4184.wav", "onoffCaption": "gunshot at 2.648-4.648 and woman laughing at 7.575-10.0", "frequencyCaption": "gunshot one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4209.wav", "onoffCaption": "door slamming at 2.279-3.532 and thump thud at 6.078-8.306 and door knocking at 7.164-9.352", "frequencyCaption": "door slamming one times and thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4312.wav", "onoffCaption": "tapping clicking clanking at 0.624-4.064, 4.836-8.276", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4347.wav", "onoffCaption": "cat meowing at 1.85-4.754, 6.696-7.708", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4479.wav", "onoffCaption": "duck quacking at 3.224-5.224 and whistling at 7.322-10.0", "frequencyCaption": "duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4562.wav", "onoffCaption": "sheep goat bleating at 0.675-2.675, 5.049-7.049 and spraying at 2.84-4.968", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4587.wav", "onoffCaption": "door slamming at 1.539-4.256, 5.225-7.942 and gunshot at 5.959-7.959", "frequencyCaption": "door slamming two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4711.wav", "onoffCaption": "cow mooing at 2.135-5.145, 6.111-8.674 and explosion at 3.265-8.265", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4744.wav", "onoffCaption": "explosion at 2.169-4.43, 5.09-7.351", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4833.wav", "onoffCaption": "duck quacking at 1.645-3.645, 5.743-7.743", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4866.wav", "onoffCaption": "sneeze at 0.481-5.01, 6.331-8.038", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4883.wav", "onoffCaption": "door slamming at 2.32-4.233, 5.384-7.384", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4928.wav", "onoffCaption": "thump thud at 0.16-2.622, 3.878-6.34", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4998.wav", "onoffCaption": "train horn at 0.32-2.787, 4.37-7.25 and sheep goat bleating at 3.461-6.781, 7.415-9.415", "frequencyCaption": "train horn two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2125.wav", "onoffCaption": "burping belching at 3.079-6.079, 7.222-9.897", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2170.wav", "onoffCaption": "spraying at 0.023-2.607, 3.651-4.255, 4.818-6.065 and duck quacking at 0.269-2.269", "frequencyCaption": "spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2195.wav", "onoffCaption": "burping belching at 0.087-3.346, 4.593-6.616 and explosion at 4.732-7.732", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2218.wav", "onoffCaption": "tapping clicking clanking at 2.113-5.553, 6.579-9.274", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2303.wav", "onoffCaption": "burping belching at 2.154-4.975, 5.67-8.491", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2356.wav", "onoffCaption": "cow mooing at 2.672-5.641, 7.662-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2526.wav", "onoffCaption": "cow mooing at 0.244-3.226, 4.079-6.625, 7.594-10.0 and car horn honking at 3.697-6.872", "frequencyCaption": "cow mooing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2700.wav", "onoffCaption": "dog barking at 1.524-3.524, 4.817-6.817", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2755.wav", "onoffCaption": "car horn honking at 1.663-5.317, 6.053-8.202", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2822.wav", "onoffCaption": "whistling at 2.19-5.065, 6.03-8.905", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2892.wav", "onoffCaption": "car horn honking at 0.074-3.728 and door slamming at 6.804-8.328", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2939.wav", "onoffCaption": "cow mooing at 1.111-4.093, 4.704-6.768, 7.68-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_2989.wav", "onoffCaption": "cat meowing at 2.02-3.575, 4.742-6.286 and door knocking at 2.836-6.452", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4028.wav", "onoffCaption": "whistling at 2.589-7.073", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4098.wav", "onoffCaption": "sheep goat bleating at 1.437-3.437, 3.974-6.969 and door knocking at 1.638-6.638, 7.29-10.0", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4133.wav", "onoffCaption": "explosion at 0.035-2.209, 4.651-7.242 and door slamming at 0.721-1.86, 2.689-4.454, 5.508-7.728", "frequencyCaption": "explosion two times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4161.wav", "onoffCaption": "sneeze at 2.452-4.928, 5.8-8.276", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4166.wav", "onoffCaption": "cat meowing at 1.548-3.438, 4.507-6.397, 7.079-8.969", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4183.wav", "onoffCaption": "burping belching at 1.619-5.125, 6.442-9.948", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4340.wav", "onoffCaption": "cow mooing at 0.835-3.817, 5.853-8.84", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4530.wav", "onoffCaption": "whistling at 1.37-10.0 and gunshot at 3.876-5.876", "frequencyCaption": "whistling one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4537.wav", "onoffCaption": "burping belching at 1.095-4.639, 5.725-8.07 and spraying at 1.935-2.802, 3.928-4.555", "frequencyCaption": "burping belching two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4565.wav", "onoffCaption": "burping belching at 1.951-4.712, 5.866-8.138", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4580.wav", "onoffCaption": "thump thud at 0.07-2.841, 5.122-7.893 and sheep goat bleating at 1.296-3.296", "frequencyCaption": "thump thud two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4716.wav", "onoffCaption": "burping belching at 3.69-7.25", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4834.wav", "onoffCaption": "door slamming at 2.952-5.952, 7.214-10.0", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4861.wav", "onoffCaption": "duck quacking at 0.25-2.25 and thump thud at 0.885-3.224 and door knocking at 6.572-10.0", "frequencyCaption": "duck quacking one times and thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4884.wav", "onoffCaption": "cat meowing at 0.347-1.357 and spraying at 0.846-2.021", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3.wav", "onoffCaption": "duck quacking at 0.203-2.203, 2.739-4.739 and train horn at 1.131-5.461, 6.713-9.193", "frequencyCaption": "duck quacking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2000.wav", "onoffCaption": "burping belching at 0.839-3.839, 5.914-8.914", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2024.wav", "onoffCaption": "thump thud at 0.692-5.142", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2071.wav", "onoffCaption": "woman laughing at 2.728-5.147, 6.23-8.649", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2094.wav", "onoffCaption": "burping belching at 0.22-2.843, 3.448-5.813, 7.816-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2202.wav", "onoffCaption": "tapping clicking clanking at 0.947-4.387, 5.953-9.393 and woman laughing at 1.93-4.167", "frequencyCaption": "tapping clicking clanking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2257.wav", "onoffCaption": "door knocking at 0.06-3.676 and sneeze at 0.255-3.365 and door slamming at 2.1-3.119", "frequencyCaption": "door knocking one times and sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2296.wav", "onoffCaption": "gunshot at 0.013-2.013, 3.206-5.206, 6.18-8.18", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2319.wav", "onoffCaption": "thump thud at 0.838-3.3 and woman laughing at 2.775-5.57, 6.239-9.034", "frequencyCaption": "thump thud one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2427.wav", "onoffCaption": "sneeze at 2.616-3.79, 4.47-6.177 and cat meowing at 7.473-9.017", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2456.wav", "onoffCaption": "door slamming at 0.534-3.017, 4.185-6.668, 7.849-10.0", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2468.wav", "onoffCaption": "dog barking at 0.272-2.272, 4.17-6.17, 7.135-9.135", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2472.wav", "onoffCaption": "woman laughing at 2.006-8.74", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2497.wav", "onoffCaption": "sneeze at 1.87-4.109, 4.688-6.927", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2569.wav", "onoffCaption": "car horn honking at 0.611-5.518, 7.265-10.0 and sheep goat bleating at 0.81-2.81, 4.698-6.698", "frequencyCaption": "car horn honking two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2573.wav", "onoffCaption": "dog barking at 0.815-2.815, 3.608-5.608 and explosion at 6.902-9.493", "frequencyCaption": "dog barking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2596.wav", "onoffCaption": "duck quacking at 0.302-2.302 and door slamming at 6.473-9.254", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2601.wav", "onoffCaption": "burping belching at 2.939-5.037 and spraying at 4.187-4.837, 5.94-7.949, 9.176-9.78", "frequencyCaption": "burping belching one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2654.wav", "onoffCaption": "cat meowing at 0.362-1.778, 2.393-3.809, 4.914-6.33 and car horn honking at 3.445-6.231", "frequencyCaption": "cat meowing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2838.wav", "onoffCaption": "door slamming at 1.159-3.522 and dog barking at 2.177-4.177, 4.702-6.702, 7.836-9.836", "frequencyCaption": "door slamming one times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2849.wav", "onoffCaption": "door knocking at 2.197-4.965, 5.543-8.39", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2877.wav", "onoffCaption": "woman laughing at 1.891-4.457, 5.083-7.451 and dog barking at 4.627-7.065, 7.622-10.0", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2888.wav", "onoffCaption": "gunshot at 3.402-5.402, 6.683-8.683", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2923.wav", "onoffCaption": "spraying at 2.881-3.785", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2952.wav", "onoffCaption": "duck quacking at 1.431-3.431, 5.499-7.499", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2976.wav", "onoffCaption": "dog barking at 0.106-2.106, 3.066-5.066, 7.405-9.405", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2993.wav", "onoffCaption": "spraying at 0.875-2.637, 3.389-4.389, 5.063-6.004 and door slamming at 2.355-4.268", "frequencyCaption": "spraying three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4032.wav", "onoffCaption": "car horn honking at 0.015-2.515, 3.464-5.964", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4067.wav", "onoffCaption": "cow mooing at 1.537-5.966 and burping belching at 3.187-5.81", "frequencyCaption": "cow mooing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4082.wav", "onoffCaption": "door knocking at 0.068-4.443 and duck quacking at 0.264-2.264 and dog barking at 2.234-4.234, 4.989-6.989", "frequencyCaption": "door knocking one times and duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4129.wav", "onoffCaption": "tapping clicking clanking at 1.318-4.758 and sheep goat bleating at 3.041-5.041", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4199.wav", "onoffCaption": "dog barking at 0.158-2.158, 3.19-5.19, 7.467-9.467", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4214.wav", "onoffCaption": "train horn at 1.922-5.456", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4230.wav", "onoffCaption": "train horn at 0.323-3.563, 5.508-8.471", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4241.wav", "onoffCaption": "whistling at 0.723-3.698 and tapping clicking clanking at 5.638-9.078", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4315.wav", "onoffCaption": "burping belching at 1.063-7.743 and cat meowing at 6.9-7.921, 8.476-9.52", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4415.wav", "onoffCaption": "car horn honking at 0.981-5.381, 6.324-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4431.wav", "onoffCaption": "woman laughing at 0.175-6.909, 7.845-9.926", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4464.wav", "onoffCaption": "door slamming at 2.516-3.494, 4.587-5.527, 6.595-7.97", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4481.wav", "onoffCaption": "door knocking at 2.762-5.492, 6.815-8.975", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4617.wav", "onoffCaption": "gunshot at 1.616-3.616, 4.192-6.192, 7.032-9.032", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4642.wav", "onoffCaption": "explosion at 3.239-8.105", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4658.wav", "onoffCaption": "train horn at 1.537-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4666.wav", "onoffCaption": "cow mooing at 1.708-5.006, 6.168-9.466", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4683.wav", "onoffCaption": "tapping clicking clanking at 3.129-6.569", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4743.wav", "onoffCaption": "gunshot at 2.914-4.914, 7.35-9.35", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4759.wav", "onoffCaption": "sneeze at 3.36-5.473, 6.677-8.79", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4798.wav", "onoffCaption": "tapping clicking clanking at 0.768-4.208, 5.52-8.96", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4935.wav", "onoffCaption": "cow mooing at 1.516-4.814, 7.173-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4960.wav", "onoffCaption": "spraying at 1.552-2.799, 4.582-7.042 and train horn at 2.416-5.896", "frequencyCaption": "spraying two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4985.wav", "onoffCaption": "train horn at 0.388-4.388 and sneeze at 1.181-2.458, 4.879-6.113", "frequencyCaption": "train horn one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2055.wav", "onoffCaption": "door slamming at 0.237-1.215, 1.804-3.345, 4.557-7.436 and duck quacking at 0.321-2.321", "frequencyCaption": "door slamming three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2101.wav", "onoffCaption": "gunshot at 1.697-3.827", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2154.wav", "onoffCaption": "dog barking at 0.723-2.723, 3.316-5.316, 6.483-8.483", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2226.wav", "onoffCaption": "door slamming at 0.091-2.091, 2.603-3.454, 4.045-4.883", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2269.wav", "onoffCaption": "door slamming at 1.059-2.972 and gunshot at 2.674-4.674, 5.625-7.625", "frequencyCaption": "door slamming one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2273.wav", "onoffCaption": "woman laughing at 1.117-3.354, 3.88-6.117", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2327.wav", "onoffCaption": "door slamming at 1.389-4.363, 6.253-9.227", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2368.wav", "onoffCaption": "car horn honking at 2.647-7.159", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2372.wav", "onoffCaption": "whistling at 0.111-2.986", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2397.wav", "onoffCaption": "tapping clicking clanking at 0.796-4.236", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2403.wav", "onoffCaption": "explosion at 2.342-5.469 and gunshot at 7.933-10.0", "frequencyCaption": "explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2419.wav", "onoffCaption": "gunshot at 0.477-2.477, 3.149-5.149 and door knocking at 1.493-4.613", "frequencyCaption": "gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2502.wav", "onoffCaption": "gunshot at 0.827-2.827, 4.117-6.117", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2518.wav", "onoffCaption": "cow mooing at 0.012-2.981", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2557.wav", "onoffCaption": "sneeze at 3.357-5.818, 7.048-8.576", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2625.wav", "onoffCaption": "thump thud at 1.569-6.019, 6.528-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2670.wav", "onoffCaption": "cow mooing at 1.159-4.128, 5.227-8.196 and explosion at 2.413-7.413", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2695.wav", "onoffCaption": "train horn at 0.6-4.781 and spraying at 0.721-1.453", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2724.wav", "onoffCaption": "woman laughing at 0.35-7.362 and tapping clicking clanking at 3.906-7.346", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2771.wav", "onoffCaption": "door slamming at 0.63-3.604, 4.655-7.629 and cat meowing at 2.295-4.235, 5.62-6.629 and duck quacking at 6.304-8.304", "frequencyCaption": "door slamming two times and cat meowing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2794.wav", "onoffCaption": "gunshot at 1.926-4.199, 5.42-7.42", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2806.wav", "onoffCaption": "burping belching at 0.386-2.589, 3.628-5.831, 6.503-8.706 and door knocking at 2.198-7.031", "frequencyCaption": "burping belching three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2853.wav", "onoffCaption": "thump thud at 1.981-4.443, 6.001-8.463", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2907.wav", "onoffCaption": "cow mooing at 0.749-3.759 and sheep goat bleating at 2.754-7.474 and dog barking at 3.083-5.083, 6.184-8.184", "frequencyCaption": "cow mooing one times and sheep goat bleating one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2948.wav", "onoffCaption": "cat meowing at 0.559-2.291 and car horn honking at 6.429-10.0", "frequencyCaption": "cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4016.wav", "onoffCaption": "sneeze at 2.058-3.289, 5.747-6.978 and door knocking at 4.891-7.194", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4043.wav", "onoffCaption": "door knocking at 0.075-4.225 and train horn at 6.32-10.0", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4059.wav", "onoffCaption": "gunshot at 0.745-2.745, 4.879-6.879", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4117.wav", "onoffCaption": "cat meowing at 0.091-1.112, 2.04-4.57, 5.573-7.533", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4142.wav", "onoffCaption": "dog barking at 0.414-2.414, 3.557-5.557, 6.334-8.334 and train horn at 1.087-3.242", "frequencyCaption": "dog barking three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4158.wav", "onoffCaption": "burping belching at 0.49-4.49, 5.449-9.449", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4265.wav", "onoffCaption": "dog barking at 0.175-5.812, 6.78-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4280.wav", "onoffCaption": "dog barking at 2.669-4.669 and sheep goat bleating at 7.512-9.512", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4331.wav", "onoffCaption": "cow mooing at 1.208-6.188 and duck quacking at 4.284-6.284", "frequencyCaption": "cow mooing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4364.wav", "onoffCaption": "burping belching at 0.19-3.19, 5.308-8.308 and dog barking at 0.267-2.267 and gunshot at 1.186-3.186, 5.249-7.249", "frequencyCaption": "burping belching two times and dog barking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4381.wav", "onoffCaption": "spraying at 0.23-1.925 and car horn honking at 4.134-8.646", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4440.wav", "onoffCaption": "cow mooing at 0.007-2.976, 5.371-7.46", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4514.wav", "onoffCaption": "tapping clicking clanking at 1.267-4.707, 5.587-9.027", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4541.wav", "onoffCaption": "dog barking at 2.832-4.832, 5.772-7.772", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4629.wav", "onoffCaption": "duck quacking at 0.526-2.526, 3.179-5.179", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4633.wav", "onoffCaption": "duck quacking at 0.812-2.812, 3.631-5.631, 6.45-8.45 and cat meowing at 2.522-4.077 and burping belching at 2.965-6.965", "frequencyCaption": "duck quacking three times and cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4699.wav", "onoffCaption": "door knocking at 3.099-5.121, 5.729-7.751", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4728.wav", "onoffCaption": "explosion at 0.713-5.634, 6.498-9.498", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4732.wav", "onoffCaption": "tapping clicking clanking at 2.353-5.793, 6.59-9.076", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4767.wav", "onoffCaption": "dog barking at 2.492-4.492, 6.811-8.811", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4782.wav", "onoffCaption": "whistling at 3.517-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4810.wav", "onoffCaption": "cat meowing at 2.171-3.171", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4845.wav", "onoffCaption": "door knocking at 2.751-5.321, 6.377-8.947", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4911.wav", "onoffCaption": "thump thud at 1.568-3.907, 4.45-6.789, 7.746-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4944.wav", "onoffCaption": "woman laughing at 0.34-2.54, 4.896-7.096", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2115.wav", "onoffCaption": "sheep goat bleating at 0.902-2.902, 4.596-6.596 and cow mooing at 1.699-4.709, 5.792-8.247", "frequencyCaption": "sheep goat bleating two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2140.wav", "onoffCaption": "burping belching at 0.728-3.728, 4.523-7.523 and car horn honking at 2.581-7.488", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2228.wav", "onoffCaption": "whistling at 1.17-6.67", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2298.wav", "onoffCaption": "car horn honking at 0.039-2.857, 3.365-6.183, 6.974-9.792", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_2333.wav", "onoffCaption": "gunshot at 3.701-5.701, 6.615-8.615", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2366.wav", "onoffCaption": "woman laughing at 2.103-4.328, 4.915-7.115", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2383.wav", "onoffCaption": "door slamming at 3.631-6.06, 6.913-7.764", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2458.wav", "onoffCaption": "duck quacking at 1.91-3.91, 5.691-7.691", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2516.wav", "onoffCaption": "tapping clicking clanking at 2.089-5.529 and train horn at 2.833-7.014", "frequencyCaption": "tapping clicking clanking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2543.wav", "onoffCaption": "cow mooing at 3.059-6.028, 7.252-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2730.wav", "onoffCaption": "dog barking at 0.421-2.421, 4.285-6.285", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2765.wav", "onoffCaption": "whistling at 0.918-3.793, 4.747-7.622", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2780.wav", "onoffCaption": "sneeze at 3.002-4.166, 4.977-7.343", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2812.wav", "onoffCaption": "train horn at 1.579-5.579 and sheep goat bleating at 3.344-5.344", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2847.wav", "onoffCaption": "dog barking at 0.439-2.439 and tapping clicking clanking at 5.168-8.608", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2909.wav", "onoffCaption": "duck quacking at 0.518-2.518, 4.844-6.844", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4018.wav", "onoffCaption": "spraying at 0.372-0.976, 2.675-3.302, 5.276-5.851 and car horn honking at 3.509-7.831", "frequencyCaption": "spraying three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4103.wav", "onoffCaption": "cat meowing at 2.286-5.316, 6.353-8.313 and woman laughing at 5.253-7.453 and sneeze at 5.34-6.666", "frequencyCaption": "cat meowing two times and woman laughing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4156.wav", "onoffCaption": "explosion at 0.661-3.661", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4325.wav", "onoffCaption": "gunshot at 0.211-2.211", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4370.wav", "onoffCaption": "gunshot at 1.844-3.844, 5.097-7.097, 7.762-9.762", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4395.wav", "onoffCaption": "dog barking at 0.207-2.607, 4.311-6.311", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4500.wav", "onoffCaption": "explosion at 0.364-5.23, 7.676-10.0 and burping belching at 1.243-5.579", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4555.wav", "onoffCaption": "door knocking at 0.82-5.353, 6.393-8.795", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4668.wav", "onoffCaption": "duck quacking at 1.603-3.603, 5.311-7.311", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4726.wav", "onoffCaption": "thump thud at 2.21-5.877", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4773.wav", "onoffCaption": "sheep goat bleating at 1.094-3.094, 4.244-6.244, 7.536-9.536", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4796.wav", "onoffCaption": "sneeze at 0.242-1.568, 2.708-4.034, 4.92-6.246", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4804.wav", "onoffCaption": "duck quacking at 0.366-2.366, 3.404-5.404 and woman laughing at 7.732-10.0", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4851.wav", "onoffCaption": "door slamming at 0.634-1.937, 3.148-4.451, 6.589-7.892", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4905.wav", "onoffCaption": "burping belching at 1.445-7.389 and cow mooing at 4.409-7.707", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2014.wav", "onoffCaption": "duck quacking at 0.1-2.1, 2.681-4.681, 5.3-7.3 and woman laughing at 0.878-3.103, 5.052-7.548", "frequencyCaption": "duck quacking three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2041.wav", "onoffCaption": "gunshot at 0.756-2.756, 4.93-6.93", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2232.wav", "onoffCaption": "cow mooing at 0.046-3.015 and sneeze at 0.117-1.704", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2267.wav", "onoffCaption": "thump thud at 0.246-4.621 and sheep goat bleating at 1.704-3.704, 5.086-8.086 and dog barking at 4.822-6.822", "frequencyCaption": "thump thud one times and sheep goat bleating two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2282.wav", "onoffCaption": "spraying at 0.196-0.797", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2329.wav", "onoffCaption": "sneeze at 0.061-2.378 and gunshot at 0.877-2.877 and dog barking at 1.829-3.829", "frequencyCaption": "sneeze one times and gunshot one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2399.wav", "onoffCaption": "cow mooing at 0.806-5.235", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2417.wav", "onoffCaption": "gunshot at 2.087-4.087, 4.956-6.956, 7.941-9.941", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2442.wav", "onoffCaption": "door knocking at 3.887-6.289, 7.303-9.705", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2559.wav", "onoffCaption": "car horn honking at 1.034-4.529 and dog barking at 6.57-8.57", "frequencyCaption": "car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2631.wav", "onoffCaption": "burping belching at 0.394-3.596 and door slamming at 1.021-3.021 and thump thud at 5.105-9.555", "frequencyCaption": "burping belching one times and door slamming one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2664.wav", "onoffCaption": "car horn honking at 0.045-2.831 and thump thud at 6.644-9.415", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2681.wav", "onoffCaption": "woman laughing at 0.0-3.085 and door knocking at 5.304-8.804", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2808.wav", "onoffCaption": "cat meowing at 0.026-1.026, 1.84-2.84 and woman laughing at 4.173-6.456", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2913.wav", "onoffCaption": "woman laughing at 3.473-6.268", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2946.wav", "onoffCaption": "burping belching at 2.011-7.612", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4002.wav", "onoffCaption": "gunshot at 0.13-2.13 and woman laughing at 1.245-3.445, 5.933-8.133", "frequencyCaption": "gunshot one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4057.wav", "onoffCaption": "door knocking at 1.551-3.854, 6.338-8.641", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4119.wav", "onoffCaption": "gunshot at 4.016-6.016", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4224.wav", "onoffCaption": "duck quacking at 0.733-2.733, 3.793-5.793, 6.393-8.393 and sheep goat bleating at 2.124-5.204 and cat meowing at 6.595-8.543", "frequencyCaption": "duck quacking three times and sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4271.wav", "onoffCaption": "cat meowing at 2.651-3.736, 4.47-6.63, 7.481-8.502", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4294.wav", "onoffCaption": "whistling at 1.904-4.779, 6.225-8.454", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4401.wav", "onoffCaption": "cow mooing at 0.424-3.406, 5.35-7.852 and car horn honking at 0.85-5.099, 5.722-8.254", "frequencyCaption": "cow mooing two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4454.wav", "onoffCaption": "door knocking at 2.973-5.436 and spraying at 7.871-8.621", "frequencyCaption": "door knocking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4627.wav", "onoffCaption": "cat meowing at 0.803-1.814, 2.735-4.045 and woman laughing at 3.119-6.406", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4672.wav", "onoffCaption": "car horn honking at 0.057-3.552, 4.582-7.095", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4697.wav", "onoffCaption": "gunshot at 0.129-2.129, 2.887-4.887, 5.731-8.232 and door knocking at 5.47-8.2", "frequencyCaption": "gunshot three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4769.wav", "onoffCaption": "woman laughing at 2.871-5.29 and duck quacking at 3.644-5.644, 7.393-9.393", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4921.wav", "onoffCaption": "dog barking at 0.349-2.787, 4.269-6.269", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4950.wav", "onoffCaption": "cat meowing at 0.651-2.268, 3.577-4.588, 5.723-7.267 and duck quacking at 3.073-5.073", "frequencyCaption": "cat meowing three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2030.wav", "onoffCaption": "thump thud at 3.336-7.786", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2065.wav", "onoffCaption": "sneeze at 0.087-1.375, 2.994-5.307, 6.757-8.702", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2080.wav", "onoffCaption": "dog barking at 2.86-4.86, 6.259-8.259", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2164.wav", "onoffCaption": "duck quacking at 1.029-3.029, 4.586-6.586, 7.745-9.745", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2216.wav", "onoffCaption": "car horn honking at 1.183-4.77 and cat meowing at 3.31-4.927, 6.66-8.277", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2243.wav", "onoffCaption": "sneeze at 0.224-1.327 and duck quacking at 0.322-2.322", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2358.wav", "onoffCaption": "burping belching at 2.599-5.801, 7.699-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2433.wav", "onoffCaption": "gunshot at 0.442-2.442", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2466.wav", "onoffCaption": "whistling at 1.816-6.3", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2483.wav", "onoffCaption": "dog barking at 1.83-3.83, 5.526-7.526", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2528.wav", "onoffCaption": "explosion at 1.045-3.798, 4.36-6.68", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2598.wav", "onoffCaption": "burping belching at 2.37-6.37, 7.357-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2615.wav", "onoffCaption": "duck quacking at 3.183-5.183", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2640.wav", "onoffCaption": "dog barking at 0.901-2.901, 5.102-7.102", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2879.wav", "onoffCaption": "spraying at 0.77-3.354, 5.572-6.353 and woman laughing at 3.378-6.473, 7.664-10.0 and sheep goat bleating at 4.777-6.777", "frequencyCaption": "spraying two times and woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2937.wav", "onoffCaption": "door knocking at 0.913-3.98, 5.526-7.766", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2962.wav", "onoffCaption": "burping belching at 0.503-3.503, 5.097-8.097 and door knocking at 7.1-9.502", "frequencyCaption": "burping belching two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2987.wav", "onoffCaption": "dog barking at 0.633-2.633, 4.02-6.02, 7.631-9.742", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4026.wav", "onoffCaption": "door slamming at 0.769-1.269 and dog barking at 4.994-6.994", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4073.wav", "onoffCaption": "duck quacking at 0.686-2.686, 4.676-6.676", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4096.wav", "onoffCaption": "tapping clicking clanking at 0.677-4.117", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4168.wav", "onoffCaption": "door knocking at 2.99-5.342 and tapping clicking clanking at 7.271-10.0", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4200.wav", "onoffCaption": "whistling at 2.405-5.38 and car horn honking at 3.25-7.65", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4255.wav", "onoffCaption": "door slamming at 2.703-3.82, 4.869-6.16, 6.968-8.509", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4425.wav", "onoffCaption": "whistling at 1.576-7.683", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4470.wav", "onoffCaption": "door slamming at 0.129-0.81, 1.333-3.559, 4.597-5.097 and door knocking at 7.313-10.0", "frequencyCaption": "door slamming three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4495.wav", "onoffCaption": "sneeze at 3.101-5.186, 7.527-9.234", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4603.wav", "onoffCaption": "cow mooing at 1.547-4.557, 5.816-8.826 and cat meowing at 2.421-4.415", "frequencyCaption": "cow mooing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4656.wav", "onoffCaption": "burping belching at 1.681-4.681, 5.243-8.243", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4718.wav", "onoffCaption": "train horn at 2.361-4.801 and sheep goat bleating at 7.701-9.701", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4974.wav", "onoffCaption": "cow mooing at 0.276-3.574 and burping belching at 0.774-3.774 and thump thud at 1.698-4.16", "frequencyCaption": "cow mooing one times and burping belching one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4991.wav", "onoffCaption": "explosion at 3.281-8.281", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2131.wav", "onoffCaption": "spraying at 2.591-3.113 and duck quacking at 4.191-6.191 and cow mooing at 4.489-7.499", "frequencyCaption": "spraying one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2181.wav", "onoffCaption": "door knocking at 2.812-7.345, 7.981-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2259.wav", "onoffCaption": "dog barking at 0.445-2.445, 3.999-5.999 and gunshot at 4.324-6.324", "frequencyCaption": "dog barking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2317.wav", "onoffCaption": "tapping clicking clanking at 0.145-3.585", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2342.wav", "onoffCaption": "burping belching at 1.788-6.124, 7.024-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2429.wav", "onoffCaption": "duck quacking at 2.986-4.986 and door slamming at 6.513-9.23", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2499.wav", "onoffCaption": "sheep goat bleating at 0.024-3.32 and dog barking at 1.92-3.92", "frequencyCaption": "sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2532.wav", "onoffCaption": "explosion at 2.04-4.631, 7.094-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2567.wav", "onoffCaption": "door slamming at 0.156-2.074", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2582.wav", "onoffCaption": "car horn honking at 3.286-7.535", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2714.wav", "onoffCaption": "sneeze at 2.176-4.121, 6.135-8.08", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2741.wav", "onoffCaption": "sneeze at 0.258-1.422, 3.657-4.821 and woman laughing at 7.507-10.0", "frequencyCaption": "sneeze two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2836.wav", "onoffCaption": "thump thud at 2.554-5.054", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2863.wav", "onoffCaption": "duck quacking at 0.365-2.365 and car horn honking at 4.688-8.183 and cat meowing at 5.873-7.148", "frequencyCaption": "duck quacking one times and car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2886.wav", "onoffCaption": "burping belching at 0.982-4.542", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2978.wav", "onoffCaption": "cat meowing at 0.357-1.901, 4.148-5.692, 6.877-8.421", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4069.wav", "onoffCaption": "train horn at 0.267-3.747, 4.969-7.643", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4127.wav", "onoffCaption": "explosion at 1.4-3.407, 5.53-7.544", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4172.wav", "onoffCaption": "thump thud at 3.642-6.104", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4197.wav", "onoffCaption": "car horn honking at 0.548-2.895, 4.036-6.962 and duck quacking at 2.823-4.823, 5.992-7.992", "frequencyCaption": "car horn honking two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4301.wav", "onoffCaption": "door knocking at 0.963-3.315, 4.703-7.055", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4354.wav", "onoffCaption": "car horn honking at 2.552-5.465", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4488.wav", "onoffCaption": "gunshot at 0.894-3.064, 4.684-6.684 and spraying at 1.03-1.762, 3.015-3.747", "frequencyCaption": "gunshot two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4524.wav", "onoffCaption": "explosion at 1.076-3.083, 5.017-7.024", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4571.wav", "onoffCaption": "thump thud at 0.313-2.541 and door slamming at 4.244-5.082, 5.615-6.453, 7.056-7.894", "frequencyCaption": "thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4594.wav", "onoffCaption": "woman laughing at 3.108-5.713, 6.503-8.987 and door slamming at 5.806-7.197", "frequencyCaption": "woman laughing two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4619.wav", "onoffCaption": "door knocking at 0.244-2.814, 4.673-7.012 and gunshot at 0.491-2.491 and sneeze at 3.661-4.895", "frequencyCaption": "door knocking two times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4702.wav", "onoffCaption": "thump thud at 2.28-6.73, 7.816-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4757.wav", "onoffCaption": "dog barking at 0.149-2.149 and tapping clicking clanking at 0.464-3.904 and burping belching at 5.381-8.925", "frequencyCaption": "dog barking one times and tapping clicking clanking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4820.wav", "onoffCaption": "cow mooing at 2.125-5.135 and gunshot at 3.983-5.983, 7.818-9.818", "frequencyCaption": "cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4872.wav", "onoffCaption": "thump thud at 2.268-6.718", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4875.wav", "onoffCaption": "door slamming at 0.609-2.609, 4.435-6.435 and sheep goat bleating at 2.272-4.272, 5.291-7.291", "frequencyCaption": "door slamming two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4890.wav", "onoffCaption": "sheep goat bleating at 3.192-5.192, 5.962-7.962 and spraying at 3.225-5.353, 6.646-8.774", "frequencyCaption": "sheep goat bleating two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4969.wav", "onoffCaption": "sneeze at 0.02-1.979 and door slamming at 1.078-2.453", "frequencyCaption": "sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2037.wav", "onoffCaption": "cow mooing at 1.731-4.713, 6.368-9.35 and spraying at 2.99-4.247, 5.36-6.617", "frequencyCaption": "cow mooing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2078.wav", "onoffCaption": "cat meowing at 0.134-3.038, 4.259-7.163 and explosion at 0.649-2.742, 3.601-5.694 and whistling at 1.104-5.588, 6.512-9.029", "frequencyCaption": "cat meowing two times and explosion two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_2136.wav", "onoffCaption": "spraying at 0.061-0.665, 2.012-2.793 and explosion at 6.775-8.839", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2163.wav", "onoffCaption": "train horn at 1.284-4.084, 5.069-7.95 and gunshot at 2.44-4.44", "frequencyCaption": "train horn two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2186.wav", "onoffCaption": "train horn at 2.174-7.918 and door knocking at 2.306-4.494, 6.83-9.018", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2310.wav", "onoffCaption": "cow mooing at 0.644-5.073, 7.545-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2345.wav", "onoffCaption": "cow mooing at 0.329-5.309, 6.179-9.161", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2461.wav", "onoffCaption": "explosion at 1.207-3.3, 3.876-5.969 and car horn honking at 2.517-5.43, 7.292-10.0", "frequencyCaption": "explosion two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2535.wav", "onoffCaption": "cow mooing at 0.06-3.07, 5.163-8.173 and thump thud at 0.481-4.148, 4.943-7.171", "frequencyCaption": "cow mooing two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2560.wav", "onoffCaption": "thump thud at 3.064-6.111", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2585.wav", "onoffCaption": "cow mooing at 0.75-5.73, 6.405-8.768", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2608.wav", "onoffCaption": "train horn at 1.016-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2713.wav", "onoffCaption": "explosion at 1.31-4.304, 5.298-8.298", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2746.wav", "onoffCaption": "explosion at 0.892-3.764, 4.383-6.907, 7.915-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_2831.wav", "onoffCaption": "gunshot at 2.394-4.394, 6.612-9.086", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2864.wav", "onoffCaption": "train horn at 0.519-5.377, 6.619-9.02", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2881.wav", "onoffCaption": "dog barking at 0.363-2.363, 4.329-6.329, 7.313-9.313", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2980.wav", "onoffCaption": "door slamming at 2.98-4.495, 5.851-7.366", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4120.wav", "onoffCaption": "sneeze at 0.206-2.452, 4.198-5.372, 6.923-9.537", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4175.wav", "onoffCaption": "duck quacking at 2.753-4.753, 6.321-8.321", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4190.wav", "onoffCaption": "sheep goat bleating at 0.805-2.805", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4248.wav", "onoffCaption": "duck quacking at 0.163-2.163 and tapping clicking clanking at 5.673-9.113", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4306.wav", "onoffCaption": "tapping clicking clanking at 3.279-6.719 and sheep goat bleating at 7.072-9.072", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4353.wav", "onoffCaption": "duck quacking at 0.65-2.65, 4.02-6.02", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4438.wav", "onoffCaption": "thump thud at 0.905-4.572 and cat meowing at 6.995-8.566", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4523.wav", "onoffCaption": "duck quacking at 2.504-4.504 and train horn at 6.766-10.0", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4576.wav", "onoffCaption": "gunshot at 0.73-2.73, 4.574-6.574 and train horn at 3.274-6.594", "frequencyCaption": "gunshot two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4593.wav", "onoffCaption": "car horn honking at 0.119-5.026 and sneeze at 0.407-1.638 and door slamming at 0.512-2.64", "frequencyCaption": "car horn honking one times and sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4705.wav", "onoffCaption": "sheep goat bleating at 0.586-4.226, 6.002-9.642", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4750.wav", "onoffCaption": "tapping clicking clanking at 1.972-5.412, 6.469-9.909 and woman laughing at 2.03-4.128", "frequencyCaption": "tapping clicking clanking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4827.wav", "onoffCaption": "dog barking at 0.577-2.577 and train horn at 0.636-4.756, 5.88-10.0", "frequencyCaption": "dog barking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4897.wav", "onoffCaption": "cow mooing at 2.385-5.367 and spraying at 8.29-8.917", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2013.wav", "onoffCaption": "dog barking at 1.338-3.338, 4.879-6.879", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2062.wav", "onoffCaption": "car horn honking at 1.119-3.632, 5.271-7.784 and dog barking at 4.978-6.978, 7.545-9.545", "frequencyCaption": "car horn honking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2087.wav", "onoffCaption": "door knocking at 0.82-5.259 and burping belching at 1.654-7.255 and door slamming at 6.373-9.347", "frequencyCaption": "door knocking one times and burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2108.wav", "onoffCaption": "cow mooing at 3.083-6.093", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2179.wav", "onoffCaption": "gunshot at 0.366-2.366 and door slamming at 4.91-5.761, 7.513-8.364", "frequencyCaption": "gunshot one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2211.wav", "onoffCaption": "sneeze at 0.014-4.514, 5.427-7.903", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2244.wav", "onoffCaption": "train horn at 1.183-4.583, 5.705-9.105", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2434.wav", "onoffCaption": "dog barking at 0.142-2.142, 3.472-5.472, 6.656-8.656", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2484.wav", "onoffCaption": "cat meowing at 0.11-3.351, 4.446-7.687", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2612.wav", "onoffCaption": "cat meowing at 4.007-5.562, 6.692-7.967", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2647.wav", "onoffCaption": "spraying at 0.301-0.876, 3.183-3.758", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2709.wav", "onoffCaption": "train horn at 1.594-5.594", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2930.wav", "onoffCaption": "tapping clicking clanking at 0.463-3.903, 4.444-6.621, 7.812-10.0", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_2965.wav", "onoffCaption": "door knocking at 0.177-3.552", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4021.wav", "onoffCaption": "dog barking at 2.558-4.558 and sheep goat bleating at 6.288-8.288", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4074.wav", "onoffCaption": "gunshot at 1.364-3.364, 4.698-6.698, 7.739-9.739", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4091.wav", "onoffCaption": "train horn at 3.3-6.18, 6.958-9.838", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4207.wav", "onoffCaption": "spraying at 0.091-0.995", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4252.wav", "onoffCaption": "duck quacking at 0.773-2.773, 4.659-6.659 and cat meowing at 6.245-7.257, 8.139-9.683", "frequencyCaption": "duck quacking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4349.wav", "onoffCaption": "whistling at 0.804-2.813, 4.259-7.09 and sheep goat bleating at 1.958-6.678", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4422.wav", "onoffCaption": "door knocking at 3.405-5.785", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4477.wav", "onoffCaption": "cat meowing at 0.083-1.349, 2.041-3.625 and gunshot at 2.761-4.761", "frequencyCaption": "cat meowing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4492.wav", "onoffCaption": "gunshot at 0.382-2.382, 3.831-5.831", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4539.wav", "onoffCaption": "duck quacking at 3.101-5.101 and thump thud at 7.995-10.0", "frequencyCaption": "duck quacking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4589.wav", "onoffCaption": "burping belching at 0.511-3.511, 4.303-7.246, 7.967-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_4604.wav", "onoffCaption": "thump thud at 3.427-5.927 and tapping clicking clanking at 3.66-7.1", "frequencyCaption": "thump thud one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4651.wav", "onoffCaption": "gunshot at 1.084-3.084, 3.952-5.952, 7.198-9.198 and whistling at 1.399-9.784", "frequencyCaption": "gunshot three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4690.wav", "onoffCaption": "tapping clicking clanking at 1.909-5.349 and dog barking at 7.675-9.675", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4868.wav", "onoffCaption": "tapping clicking clanking at 0.529-3.969", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4926.wav", "onoffCaption": "gunshot at 1.876-3.876, 4.842-6.842, 7.718-9.718", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4973.wav", "onoffCaption": "woman laughing at 2.777-5.546", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4996.wav", "onoffCaption": "gunshot at 2.571-4.701, 5.405-7.535", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2046.wav", "onoffCaption": "thump thud at 0.536-3.307 and gunshot at 3.348-5.348", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2235.wav", "onoffCaption": "explosion at 2.514-5.514, 7.383-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2260.wav", "onoffCaption": "train horn at 2.617-7.694", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2285.wav", "onoffCaption": "tapping clicking clanking at 0.136-3.576 and dog barking at 0.345-2.745", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2334.wav", "onoffCaption": "thump thud at 0.05-3.097, 4.756-7.218 and dog barking at 0.212-2.212", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2410.wav", "onoffCaption": "gunshot at 1.449-3.449 and cat meowing at 2.183-4.143 and sneeze at 3.218-4.392", "frequencyCaption": "gunshot one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2445.wav", "onoffCaption": "cat meowing at 0.275-1.811, 3.469-5.005, 6.461-7.997 and dog barking at 2.825-4.825 and sneeze at 5.34-8.0", "frequencyCaption": "cat meowing three times and dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2636.wav", "onoffCaption": "train horn at 2.96-6.36, 7.243-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2663.wav", "onoffCaption": "whistling at 0.052-2.281, 4.45-7.014 and train horn at 4.814-8.214", "frequencyCaption": "whistling two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2679.wav", "onoffCaption": "spraying at 2.8-3.3, 5.687-7.42", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2686.wav", "onoffCaption": "train horn at 2.685-7.762", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2762.wav", "onoffCaption": "door slamming at 0.1-2.32, 4.242-6.462 and dog barking at 3.512-5.512, 6.099-8.099", "frequencyCaption": "door slamming two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2778.wav", "onoffCaption": "car horn honking at 0.462-3.681 and sheep goat bleating at 5.35-7.35", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2787.wav", "onoffCaption": "tapping clicking clanking at 1.579-5.019 and burping belching at 2.926-5.926, 7.082-10.0", "frequencyCaption": "tapping clicking clanking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2914.wav", "onoffCaption": "cow mooing at 2.723-7.152", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2941.wav", "onoffCaption": "duck quacking at 0.171-2.171 and spraying at 3.39-3.994 and sneeze at 5.409-6.795", "frequencyCaption": "duck quacking one times and spraying one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4005.wav", "onoffCaption": "sneeze at 1.605-4.066, 6.518-8.437", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4050.wav", "onoffCaption": "cat meowing at 0.481-1.525, 3.829-4.873 and tapping clicking clanking at 1.071-4.511", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4104.wav", "onoffCaption": "explosion at 0.572-2.665, 5.145-7.238 and door slamming at 3.172-4.289, 5.792-6.909", "frequencyCaption": "explosion two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4223.wav", "onoffCaption": "duck quacking at 0.184-2.184, 3.018-5.018, 6.317-8.317 and sheep goat bleating at 6.077-8.077", "frequencyCaption": "duck quacking three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4276.wav", "onoffCaption": "gunshot at 3.244-5.718, 6.806-9.28", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4293.wav", "onoffCaption": "sneeze at 2.122-3.579, 5.94-7.397 and cow mooing at 3.03-6.012", "frequencyCaption": "sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4338.wav", "onoffCaption": "burping belching at 1.041-3.406 and sneeze at 2.307-4.392", "frequencyCaption": "burping belching one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4388.wav", "onoffCaption": "sneeze at 1.411-2.657 and explosion at 2.597-5.591", "frequencyCaption": "sneeze one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4406.wav", "onoffCaption": "door knocking at 0.49-3.61, 4.378-7.498 and train horn at 5.486-7.966", "frequencyCaption": "door knocking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4453.wav", "onoffCaption": "explosion at 0.123-3.25, 5.341-8.468", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4548.wav", "onoffCaption": "whistling at 2.538-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4620.wav", "onoffCaption": "woman laughing at 1.411-4.465, 4.987-8.041 and gunshot at 1.892-3.892", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4675.wav", "onoffCaption": "whistling at 0.21-8.595 and sneeze at 0.27-2.229", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4819.wav", "onoffCaption": "cat meowing at 1.092-4.333", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4856.wav", "onoffCaption": "thump thud at 0.115-2.577, 4.284-6.746, 7.956-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4902.wav", "onoffCaption": "spraying at 0.111-0.611, 1.41-1.979, 2.877-4.052 and gunshot at 6.942-8.942", "frequencyCaption": "spraying three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4957.wav", "onoffCaption": "whistling at 1.592-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2009.wav", "onoffCaption": "thump thud at 0.776-5.151", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2077.wav", "onoffCaption": "whistling at 1.112-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2112.wav", "onoffCaption": "explosion at 2.23-5.23 and door slamming at 8.138-9.157", "frequencyCaption": "explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2147.wav", "onoffCaption": "woman laughing at 1.159-4.547", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2189.wav", "onoffCaption": "train horn at 0.649-3.323, 4.516-7.037", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2361.wav", "onoffCaption": "train horn at 0.466-2.621, 3.841-5.996, 7.71-9.865", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_2384.wav", "onoffCaption": "door slamming at 0.36-2.488, 4.172-6.3", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2421.wav", "onoffCaption": "spraying at 0.679-1.926, 3.703-4.95 and door knocking at 3.2-5.93", "frequencyCaption": "spraying two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2511.wav", "onoffCaption": "sheep goat bleating at 0.209-2.209", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2544.wav", "onoffCaption": "whistling at 2.084-4.313 and door slamming at 4.281-5.534", "frequencyCaption": "whistling one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2737.wav", "onoffCaption": "sneeze at 0.645-2.558, 3.721-5.634, 6.689-8.602", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2815.wav", "onoffCaption": "cat meowing at 2.48-4.23", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2840.wav", "onoffCaption": "burping belching at 0.204-3.484, 4.239-7.156 and whistling at 0.478-8.863", "frequencyCaption": "burping belching two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2925.wav", "onoffCaption": "tapping clicking clanking at 0.44-3.88", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4151.wav", "onoffCaption": "gunshot at 0.002-2.002, 3.068-5.068, 6.124-8.124", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4239.wav", "onoffCaption": "cow mooing at 1.84-5.138, 6.741-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4247.wav", "onoffCaption": "gunshot at 3.188-5.318, 7.34-9.34", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4289.wav", "onoffCaption": "spraying at 2.817-3.992, 5.823-6.823", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4322.wav", "onoffCaption": "cow mooing at 0.833-3.802, 5.116-8.098", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4377.wav", "onoffCaption": "whistling at 0.24-5.74 and spraying at 0.76-2.007, 2.668-3.915 and explosion at 1.028-6.028", "frequencyCaption": "whistling one times and spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4392.wav", "onoffCaption": "spraying at 0.221-0.79, 1.312-1.881, 4.348-4.917", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4449.wav", "onoffCaption": "sheep goat bleating at 3.247-5.247", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4507.wav", "onoffCaption": "cat meowing at 0.37-2.364 and woman laughing at 2.598-5.184 and spraying at 4.817-7.836", "frequencyCaption": "cat meowing one times and woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4552.wav", "onoffCaption": "whistling at 1.266-6.441, 7.836-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4721.wav", "onoffCaption": "car horn honking at 1.973-6.295", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4774.wav", "onoffCaption": "woman laughing at 0.137-2.331, 4.071-6.657 and cat meowing at 1.489-2.799", "frequencyCaption": "woman laughing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4791.wav", "onoffCaption": "thump thud at 1.876-4.338, 5.469-7.931 and door knocking at 5.537-8.0", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4803.wav", "onoffCaption": "cat meowing at 1.28-2.546 and door slamming at 3.938-4.789, 5.958-8.184", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4918.wav", "onoffCaption": "car horn honking at 0.143-3.056, 5.143-8.069", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_5.wav", "onoffCaption": "train horn at 3.048-7.116 and cat meowing at 6.564-7.75", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2022.wav", "onoffCaption": "sneeze at 0.421-2.738", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2092.wav", "onoffCaption": "train horn at 0.127-4.457, 5.738-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2139.wav", "onoffCaption": "tapping clicking clanking at 3.088-6.528, 7.884-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2204.wav", "onoffCaption": "cat meowing at 3.308-5.058, 7.31-9.06", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2251.wav", "onoffCaption": "thump thud at 0.879-5.329", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2350.wav", "onoffCaption": "train horn at 0.102-4.302, 5.661-8.516", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2474.wav", "onoffCaption": "gunshot at 0.141-2.141, 4.521-6.521, 7.761-9.891", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2491.wav", "onoffCaption": "spraying at 0.317-0.921 and sheep goat bleating at 2.923-4.923 and burping belching at 3.97-6.97", "frequencyCaption": "spraying one times and sheep goat bleating one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2607.wav", "onoffCaption": "door knocking at 0.375-3.105 and duck quacking at 2.675-4.675", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2652.wav", "onoffCaption": "thump thud at 3.045-7.495", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2706.wav", "onoffCaption": "duck quacking at 1.028-3.028, 4.755-6.755, 7.606-9.606", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2749.wav", "onoffCaption": "spraying at 1.223-2.304", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2970.wav", "onoffCaption": "sneeze at 0.91-3.524", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2995.wav", "onoffCaption": "train horn at 3.448-9.917", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4034.wav", "onoffCaption": "dog barking at 1.213-4.533, 6.858-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4061.wav", "onoffCaption": "door slamming at 3.947-4.798 and dog barking at 7.487-9.487", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4084.wav", "onoffCaption": "sheep goat bleating at 0.916-2.916, 4.112-6.918, 7.548-9.548", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4160.wav", "onoffCaption": "gunshot at 1.982-4.483, 5.194-7.194", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4185.wav", "onoffCaption": "cow mooing at 1.666-4.964, 6.339-9.308", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4212.wav", "onoffCaption": "gunshot at 2.485-4.485, 5.365-7.365", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4309.wav", "onoffCaption": "car horn honking at 0.258-4.507, 5.729-9.978", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4437.wav", "onoffCaption": "cow mooing at 0.883-3.852, 4.639-7.608", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4462.wav", "onoffCaption": "spraying at 1.642-4.661 and train horn at 2.0-5.32", "frequencyCaption": "spraying one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4487.wav", "onoffCaption": "woman laughing at 0.162-3.257, 4.068-6.552 and dog barking at 0.198-2.198", "frequencyCaption": "woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4536.wav", "onoffCaption": "woman laughing at 2.541-5.636, 7.344-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4579.wav", "onoffCaption": "burping belching at 0.802-4.242, 5.248-8.688 and gunshot at 2.58-4.58, 6.553-8.553", "frequencyCaption": "burping belching two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4611.wav", "onoffCaption": "dog barking at 2.046-4.046, 5.252-7.252", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4644.wav", "onoffCaption": "gunshot at 1.209-3.209, 5.051-7.324 and train horn at 2.986-7.106, 7.822-10.0", "frequencyCaption": "gunshot two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4828.wav", "onoffCaption": "car horn honking at 2.865-7.114 and thump thud at 4.049-6.277", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4832.wav", "onoffCaption": "whistling at 1.547-9.202", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4898.wav", "onoffCaption": "spraying at 0.281-1.538, 2.407-3.664, 4.717-5.974", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4929.wav", "onoffCaption": "gunshot at 1.601-3.601", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4933.wav", "onoffCaption": "sheep goat bleating at 2.998-4.998, 6.444-8.444", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4966.wav", "onoffCaption": "cow mooing at 3.137-6.106", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4983.wav", "onoffCaption": "whistling at 1.786-4.015, 5.199-7.52", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2038.wav", "onoffCaption": "duck quacking at 0.312-2.312, 3.475-5.475, 6.684-8.684 and thump thud at 1.912-4.412", "frequencyCaption": "duck quacking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2088.wav", "onoffCaption": "cat meowing at 0.012-1.902 and burping belching at 5.241-8.531", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2123.wav", "onoffCaption": "tapping clicking clanking at 0.644-4.084, 5.681-8.568", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2176.wav", "onoffCaption": "cow mooing at 2.931-7.911", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2193.wav", "onoffCaption": "whistling at 0.783-9.168", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2305.wav", "onoffCaption": "sneeze at 0.788-2.747, 3.395-4.683 and gunshot at 1.228-3.228", "frequencyCaption": "sneeze two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2374.wav", "onoffCaption": "sneeze at 2.903-4.067, 4.883-6.047 and train horn at 3.325-5.965, 7.216-9.683", "frequencyCaption": "sneeze two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2520.wav", "onoffCaption": "sheep goat bleating at 0.244-4.164", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2575.wav", "onoffCaption": "door knocking at 0.167-2.519, 3.115-5.467, 7.305-9.657", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2590.wav", "onoffCaption": "door slamming at 0.593-1.533, 3.478-4.418", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2639.wav", "onoffCaption": "train horn at 3.428-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2648.wav", "onoffCaption": "gunshot at 0.06-2.06", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2722.wav", "onoffCaption": "duck quacking at 1.492-3.492, 5.29-7.29", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2753.wav", "onoffCaption": "thump thud at 0.078-2.306, 4.188-6.416 and train horn at 0.566-3.206 and dog barking at 2.021-5.341, 6.271-9.591", "frequencyCaption": "thump thud two times and train horn one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2824.wav", "onoffCaption": "woman laughing at 0.649-3.004, 5.141-7.589 and door knocking at 3.792-6.859, 7.738-10.0", "frequencyCaption": "woman laughing two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2871.wav", "onoffCaption": "cow mooing at 1.316-4.614, 5.217-8.144", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2894.wav", "onoffCaption": "gunshot at 1.267-3.267, 4.861-6.861", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4135.wav", "onoffCaption": "duck quacking at 0.516-2.516", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4144.wav", "onoffCaption": "sneeze at 0.05-2.163, 4.515-6.628, 7.634-9.747", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4208.wav", "onoffCaption": "dog barking at 3.387-5.387", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4313.wav", "onoffCaption": "burping belching at 1.703-3.826, 4.405-7.405 and dog barking at 5.109-7.109, 7.775-9.775", "frequencyCaption": "burping belching two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4346.wav", "onoffCaption": "tapping clicking clanking at 0.824-4.264 and cow mooing at 6.055-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4478.wav", "onoffCaption": "duck quacking at 0.198-2.198, 2.933-4.933 and cow mooing at 1.378-4.347", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4563.wav", "onoffCaption": "explosion at 0.394-3.394", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4586.wav", "onoffCaption": "door knocking at 3.024-6.133 and cat meowing at 3.165-4.525", "frequencyCaption": "door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4710.wav", "onoffCaption": "burping belching at 1.228-4.407 and sneeze at 2.434-3.728", "frequencyCaption": "burping belching one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4745.wav", "onoffCaption": "burping belching at 0.033-3.212, 4.738-7.917 and sheep goat bleating at 3.435-5.435, 6.487-8.487", "frequencyCaption": "burping belching two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4816.wav", "onoffCaption": "woman laughing at 3.804-6.029 and dog barking at 5.915-7.915", "frequencyCaption": "woman laughing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4867.wav", "onoffCaption": "duck quacking at 0.123-2.123, 4.217-6.217 and gunshot at 4.27-6.27", "frequencyCaption": "duck quacking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4882.wav", "onoffCaption": "spraying at 0.017-0.868, 2.396-3.247", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4999.wav", "onoffCaption": "car horn honking at 0.007-3.226, 5.675-8.894 and cat meowing at 0.879-1.981", "frequencyCaption": "car horn honking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2049.wav", "onoffCaption": "sheep goat bleating at 2.376-4.376, 5.004-7.004", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2053.wav", "onoffCaption": "cow mooing at 3.22-6.23", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2107.wav", "onoffCaption": "door knocking at 0.202-2.579 and thump thud at 7.565-10.0", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2148.wav", "onoffCaption": "spraying at 1.992-3.754 and door slamming at 2.991-4.382, 5.138-6.529, 7.82-9.211", "frequencyCaption": "spraying one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2152.wav", "onoffCaption": "tapping clicking clanking at 3.493-6.933 and cow mooing at 3.754-7.052", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2321.wav", "onoffCaption": "whistling at 2.723-7.207", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2391.wav", "onoffCaption": "cat meowing at 2.812-4.772, 6.893-8.853", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2405.wav", "onoffCaption": "whistling at 0.352-3.227", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2504.wav", "onoffCaption": "sneeze at 3.377-5.336", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2551.wav", "onoffCaption": "door slamming at 2.538-3.677, 4.833-5.972", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2689.wav", "onoffCaption": "woman laughing at 2.249-8.983", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2777.wav", "onoffCaption": "burping belching at 0.147-3.691", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2792.wav", "onoffCaption": "burping belching at 1.622-5.491", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2800.wav", "onoffCaption": "door slamming at 0.311-1.544, 3.521-4.754", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2855.wav", "onoffCaption": "car horn honking at 0.434-5.341", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2901.wav", "onoffCaption": "whistling at 0.354-5.854", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4111.wav", "onoffCaption": "woman laughing at 2.429-4.534, 5.334-7.689 and dog barking at 3.371-5.371, 6.548-8.548", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4279.wav", "onoffCaption": "car horn honking at 2.345-7.252", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4286.wav", "onoffCaption": "cow mooing at 0.317-5.297", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4337.wav", "onoffCaption": "spraying at 2.863-3.73 and cat meowing at 6.571-9.601", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4362.wav", "onoffCaption": "dog barking at 2.986-6.306", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4387.wav", "onoffCaption": "cow mooing at 1.656-4.666, 6.806-9.081", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4409.wav", "onoffCaption": "cat meowing at 0.497-2.491, 4.841-6.835 and car horn honking at 1.246-5.568", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4512.wav", "onoffCaption": "car horn honking at 1.26-5.101, 5.957-8.883", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4547.wav", "onoffCaption": "train horn at 1.784-4.944", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4734.wav", "onoffCaption": "gunshot at 3.456-5.475, 7.633-9.633", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4761.wav", "onoffCaption": "car horn honking at 1.518-5.172, 6.256-9.91", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4784.wav", "onoffCaption": "dog barking at 2.425-4.425, 5.507-7.507 and spraying at 5.914-7.171, 8.215-9.39", "frequencyCaption": "dog barking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4843.wav", "onoffCaption": "door slamming at 1.734-3.862, 5.016-7.144 and woman laughing at 4.551-6.632, 7.972-10.0", "frequencyCaption": "door slamming two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4958.wav", "onoffCaption": "cat meowing at 0.111-1.213, 2.139-3.241 and tapping clicking clanking at 1.733-5.173 and gunshot at 4.216-6.216", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2006.wav", "onoffCaption": "explosion at 0.8-3.794, 5.316-8.31", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2220.wav", "onoffCaption": "cow mooing at 0.297-3.307, 4.166-7.148", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2227.wav", "onoffCaption": "train horn at 0.365-2.765, 3.678-6.078, 6.632-9.032 and cat meowing at 3.73-5.301", "frequencyCaption": "train horn three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2275.wav", "onoffCaption": "dog barking at 0.194-2.194, 4.569-6.569 and tapping clicking clanking at 0.203-3.643", "frequencyCaption": "dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2290.wav", "onoffCaption": "door knocking at 0.345-4.617, 6.113-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2450.wav", "onoffCaption": "explosion at 3.699-8.293 and sheep goat bleating at 4.38-6.38, 7.404-9.404", "frequencyCaption": "explosion one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2623.wav", "onoffCaption": "sheep goat bleating at 1.467-6.187", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2671.wav", "onoffCaption": "sneeze at 0.735-2.012, 4.199-5.476", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2676.wav", "onoffCaption": "gunshot at 2.71-4.71", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2693.wav", "onoffCaption": "sheep goat bleating at 0.134-2.134, 2.98-4.98, 6.05-8.05", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2694.wav", "onoffCaption": "cat meowing at 2.385-4.333, 5.505-7.453", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2738.wav", "onoffCaption": "whistling at 0.685-5.169, 7.608-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2788.wav", "onoffCaption": "burping belching at 0.0-4.023, 5.919-8.245 and whistling at 0.417-2.426", "frequencyCaption": "burping belching two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2954.wav", "onoffCaption": "sneeze at 0.438-2.397, 3.235-5.194 and spraying at 3.26-3.835, 5.502-6.077", "frequencyCaption": "sneeze two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4010.wav", "onoffCaption": "door slamming at 1.845-2.962, 5.286-6.403, 7.731-8.848", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4017.wav", "onoffCaption": "explosion at 2.384-5.112 and woman laughing at 7.997-10.0", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4045.wav", "onoffCaption": "spraying at 1.773-2.273, 3.359-3.859, 5.793-6.293", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4236.wav", "onoffCaption": "door slamming at 3.015-5.241, 6.472-8.698 and spraying at 3.798-4.373, 5.108-7.544, 8.162-8.894", "frequencyCaption": "door slamming two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4263.wav", "onoffCaption": "train horn at 3.019-6.789 and sneeze at 3.37-5.483", "frequencyCaption": "train horn one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4378.wav", "onoffCaption": "tapping clicking clanking at 0.679-4.119, 5.177-7.992 and cow mooing at 3.921-6.89", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4413.wav", "onoffCaption": "whistling at 0.02-7.77 and car horn honking at 3.683-6.858", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4441.wav", "onoffCaption": "car horn honking at 0.078-4.478, 5.779-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4446.wav", "onoffCaption": "thump thud at 0.432-2.932, 4.167-6.395", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4508.wav", "onoffCaption": "door knocking at 0.277-4.027, 5.925-8.174 and dog barking at 1.887-3.887", "frequencyCaption": "door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4635.wav", "onoffCaption": "sneeze at 0.402-2.715, 5.087-7.4", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4660.wav", "onoffCaption": "thump thud at 0.05-2.278, 3.47-5.565", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4685.wav", "onoffCaption": "thump thud at 1.015-4.682, 6.109-8.296", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4859.wav", "onoffCaption": "door knocking at 0.375-2.75, 4.289-6.664 and thump thud at 4.902-7.364", "frequencyCaption": "door knocking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4917.wav", "onoffCaption": "door knocking at 0.45-3.559 and cow mooing at 1.189-4.171, 5.069-7.646", "frequencyCaption": "door knocking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4942.wav", "onoffCaption": "duck quacking at 2.841-4.841, 6.439-8.439", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4945.wav", "onoffCaption": "sneeze at 3.167-4.421, 5.355-6.609", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2001.wav", "onoffCaption": "cat meowing at 3.737-6.767, 7.269-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2054.wav", "onoffCaption": "explosion at 3.827-8.421", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2100.wav", "onoffCaption": "spraying at 2.764-3.368, 4.537-5.141", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2124.wav", "onoffCaption": "gunshot at 3.266-5.436, 7.546-9.546", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2155.wav", "onoffCaption": "cow mooing at 1.989-4.999, 7.306-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2268.wav", "onoffCaption": "whistling at 2.877-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2272.wav", "onoffCaption": "door slamming at 2.858-3.997, 4.816-5.621, 6.487-8.405", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2297.wav", "onoffCaption": "car horn honking at 3.459-5.924, 7.102-9.567", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2326.wav", "onoffCaption": "train horn at 1.47-5.67 and gunshot at 1.569-3.569, 4.393-6.393, 7.694-9.694 and whistling at 2.957-5.832", "frequencyCaption": "train horn one times and gunshot three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2369.wav", "onoffCaption": "woman laughing at 1.147-3.739, 4.978-7.57", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2373.wav", "onoffCaption": "cow mooing at 0.035-4.464 and cat meowing at 7.286-8.426", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2396.wav", "onoffCaption": "door slamming at 0.228-2.448, 3.938-6.158", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2402.wav", "onoffCaption": "door knocking at 2.793-5.849", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2418.wav", "onoffCaption": "dog barking at 0.291-2.291 and explosion at 0.294-5.294, 6.146-8.21", "frequencyCaption": "dog barking one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2457.wav", "onoffCaption": "dog barking at 3.02-5.02, 6.881-8.881", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2503.wav", "onoffCaption": "sheep goat bleating at 0.386-3.706", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2519.wav", "onoffCaption": "spraying at 3.627-4.228, 5.239-5.971, 6.818-8.976", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2556.wav", "onoffCaption": "train horn at 0.173-5.917", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2572.wav", "onoffCaption": "spraying at 0.005-0.589, 1.147-1.669, 2.456-3.703 and sheep goat bleating at 5.9-8.98", "frequencyCaption": "spraying three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2624.wav", "onoffCaption": "gunshot at 1.164-3.638", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2725.wav", "onoffCaption": "cat meowing at 0.226-1.439, 2.668-3.881 and duck quacking at 7.397-9.397", "frequencyCaption": "cat meowing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2770.wav", "onoffCaption": "train horn at 0.143-5.858, 6.746-9.611", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2795.wav", "onoffCaption": "woman laughing at 3.182-5.298, 5.997-8.113", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2807.wav", "onoffCaption": "woman laughing at 2.828-9.562 and train horn at 3.108-7.966", "frequencyCaption": "woman laughing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2848.wav", "onoffCaption": "whistling at 2.81-9.727", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2852.wav", "onoffCaption": "sneeze at 2.011-3.718 and duck quacking at 2.159-4.159", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2906.wav", "onoffCaption": "cow mooing at 2.08-5.049", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2949.wav", "onoffCaption": "woman laughing at 3.906-6.261", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2953.wav", "onoffCaption": "explosion at 0.864-3.864, 5.072-8.009", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4042.wav", "onoffCaption": "dog barking at 2.567-5.488, 7.079-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4058.wav", "onoffCaption": "sneeze at 0.092-2.495, 3.276-4.863 and duck quacking at 0.416-2.416, 4.8-6.8 and gunshot at 3.904-5.904, 6.856-8.856", "frequencyCaption": "sneeze two times and duck quacking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4116.wav", "onoffCaption": "sheep goat bleating at 0.718-2.718, 4.892-6.892 and door knocking at 1.376-5.909, 6.721-8.865 and dog barking at 5.228-7.228", "frequencyCaption": "sheep goat bleating two times and door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4143.wav", "onoffCaption": "whistling at 0.576-8.961 and dog barking at 0.748-6.385 and woman laughing at 2.533-5.138", "frequencyCaption": "whistling one times and dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4159.wav", "onoffCaption": "thump thud at 0.331-4.249 and duck quacking at 2.631-4.631, 5.476-7.476", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4231.wav", "onoffCaption": "car horn honking at 1.169-3.634 and door slamming at 7.65-9.563", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4264.wav", "onoffCaption": "burping belching at 1.205-4.711, 5.317-8.823 and woman laughing at 2.257-5.311", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4281.wav", "onoffCaption": "woman laughing at 0.493-2.718", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4330.wav", "onoffCaption": "gunshot at 1.206-3.206, 4.669-6.669 and sheep goat bleating at 2.87-4.87, 6.162-8.162", "frequencyCaption": "gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4365.wav", "onoffCaption": "tapping clicking clanking at 0.181-3.621, 4.89-7.477 and door slamming at 0.518-2.059, 2.809-4.35", "frequencyCaption": "tapping clicking clanking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4380.wav", "onoffCaption": "cow mooing at 1.129-4.427 and tapping clicking clanking at 2.306-5.746, 7.032-9.339", "frequencyCaption": "cow mooing one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4414.wav", "onoffCaption": "door slamming at 1.825-2.663, 4.903-5.741", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4515.wav", "onoffCaption": "sneeze at 0.751-2.664, 3.184-5.097", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4540.wav", "onoffCaption": "train horn at 0.83-6.545, 7.829-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4628.wav", "onoffCaption": "duck quacking at 0.235-2.235, 3.981-5.981, 7.014-9.014 and door slamming at 0.25-1.15", "frequencyCaption": "duck quacking three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4632.wav", "onoffCaption": "spraying at 0.339-1.071, 2.214-2.798, 3.436-4.5 and sneeze at 3.126-5.211, 6.084-8.169", "frequencyCaption": "spraying three times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4667.wav", "onoffCaption": "door knocking at 2.552-4.642, 6.518-8.608", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4682.wav", "onoffCaption": "door slamming at 0.297-1.776, 2.503-3.354, 4.599-5.577", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4698.wav", "onoffCaption": "cow mooing at 0.428-5.408 and whistling at 0.539-8.924", "frequencyCaption": "cow mooing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4729.wav", "onoffCaption": "car horn honking at 1.494-4.42, 5.18-7.375", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4733.wav", "onoffCaption": "woman laughing at 1.817-4.011 and door knocking at 5.746-8.593", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4766.wav", "onoffCaption": "sheep goat bleating at 0.695-2.695, 3.642-5.642, 6.254-8.254", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4783.wav", "onoffCaption": "cow mooing at 2.223-5.233, 7.477-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4799.wav", "onoffCaption": "sneeze at 0.383-4.912, 6.313-7.59", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4811.wav", "onoffCaption": "door knocking at 3.503-5.815, 7.402-9.714", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4844.wav", "onoffCaption": "car horn honking at 0.575-3.04 and whistling at 1.136-6.311, 7.842-10.0 and door knocking at 5.702-8.47", "frequencyCaption": "car horn honking one times and whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4910.wav", "onoffCaption": "dog barking at 3.332-5.332, 6.55-8.55", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2171.wav", "onoffCaption": "gunshot at 3.247-5.247", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2194.wav", "onoffCaption": "door knocking at 2.709-6.262, 7.986-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2219.wav", "onoffCaption": "cow mooing at 0.867-5.296, 6.169-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2302.wav", "onoffCaption": "burping belching at 2.824-5.824, 6.756-8.973", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2318.wav", "onoffCaption": "whistling at 0.004-2.233, 3.194-6.069, 7.167-9.818", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_2357.wav", "onoffCaption": "sneeze at 3.056-4.148, 6.17-7.262, 8.722-9.814", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2469.wav", "onoffCaption": "burping belching at 0.092-3.092, 3.702-5.932, 7.024-9.674", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2527.wav", "onoffCaption": "spraying at 1.357-3.941, 4.523-7.107", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2597.wav", "onoffCaption": "cat meowing at 3.19-5.13, 6.832-8.772", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2701.wav", "onoffCaption": "explosion at 1.956-4.685, 5.357-8.086", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2754.wav", "onoffCaption": "sneeze at 0.324-1.824, 2.589-3.843 and tapping clicking clanking at 6.289-9.729", "frequencyCaption": "sneeze two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2823.wav", "onoffCaption": "sheep goat bleating at 0.14-3.14", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2876.wav", "onoffCaption": "duck quacking at 2.428-4.428, 6.174-8.174", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2893.wav", "onoffCaption": "door slamming at 1.258-3.741", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2938.wav", "onoffCaption": "duck quacking at 0.341-2.341, 3.83-5.83", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2988.wav", "onoffCaption": "cow mooing at 1.038-4.02", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4029.wav", "onoffCaption": "tapping clicking clanking at 3.41-6.85", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4099.wav", "onoffCaption": "thump thud at 0.883-3.222, 5.031-7.37 and sneeze at 1.897-4.557, 5.208-7.167", "frequencyCaption": "thump thud two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4132.wav", "onoffCaption": "whistling at 0.88-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4167.wav", "onoffCaption": "burping belching at 0.189-7.357 and explosion at 2.581-7.581 and cat meowing at 7.678-8.689", "frequencyCaption": "burping belching one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4182.wav", "onoffCaption": "gunshot at 1.168-3.168, 5.131-7.131", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4314.wav", "onoffCaption": "burping belching at 2.924-5.924, 6.913-8.948", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4341.wav", "onoffCaption": "burping belching at 1.82-4.82", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4531.wav", "onoffCaption": "whistling at 2.216-9.966", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4564.wav", "onoffCaption": "burping belching at 1.236-4.236, 5.53-8.53 and sneeze at 5.994-7.288", "frequencyCaption": "burping belching two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4581.wav", "onoffCaption": "door knocking at 0.688-4.241 and door slamming at 7.512-8.412", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4659.wav", "onoffCaption": "whistling at 0.504-8.08 and explosion at 2.227-6.227, 6.929-9.249 and spraying at 6.965-7.473, 7.993-8.501, 9.439-9.947", "frequencyCaption": "whistling one times and explosion two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4717.wav", "onoffCaption": "sheep goat bleating at 0.537-2.537, 3.04-5.78 and duck quacking at 1.164-3.164", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4742.wav", "onoffCaption": "woman laughing at 3.527-6.165 and duck quacking at 5.318-7.318", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4835.wav", "onoffCaption": "cow mooing at 2.109-5.119 and gunshot at 7.477-9.477", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4860.wav", "onoffCaption": "thump thud at 0.222-2.561, 4.496-7.267", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4885.wav", "onoffCaption": "train horn at 0.303-3.663 and sheep goat bleating at 0.483-2.483 and cat meowing at 4.115-5.262, 7.386-8.533", "frequencyCaption": "train horn one times and sheep goat bleating one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2.wav", "onoffCaption": "gunshot at 3.979-5.979, 7.863-9.863", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2070.wav", "onoffCaption": "tapping clicking clanking at 0.312-3.752, 6.233-8.612", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2095.wav", "onoffCaption": "explosion at 0.659-3.531, 5.683-8.555", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2203.wav", "onoffCaption": "train horn at 0.751-3.991, 4.924-7.607 and burping belching at 1.081-3.842", "frequencyCaption": "train horn two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2256.wav", "onoffCaption": "whistling at 0.039-7.789", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2426.wav", "onoffCaption": "burping belching at 0.083-3.952, 4.847-8.716", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2496.wav", "onoffCaption": "burping belching at 0.734-2.841, 4.296-7.291 and duck quacking at 5.046-7.046, 7.787-9.787", "frequencyCaption": "burping belching two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2568.wav", "onoffCaption": "door knocking at 1.848-4.088, 5.701-7.941", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2600.wav", "onoffCaption": "spraying at 1.873-3.606, 5.628-7.361", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2655.wav", "onoffCaption": "door slamming at 3.365-5.728, 6.77-9.133 and woman laughing at 5.218-7.784", "frequencyCaption": "door slamming two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2839.wav", "onoffCaption": "train horn at 2.721-7.623", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2889.wav", "onoffCaption": "woman laughing at 0.784-3.389, 5.389-7.994", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2922.wav", "onoffCaption": "sheep goat bleating at 2.514-4.514, 6.364-9.354", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2977.wav", "onoffCaption": "door knocking at 3.285-6.838, 7.652-9.812", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2992.wav", "onoffCaption": "sneeze at 2.421-4.66, 5.402-7.641", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4033.wav", "onoffCaption": "door knocking at 1.626-5.922, 6.87-9.119", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4066.wav", "onoffCaption": "cat meowing at 0.594-2.73, 3.248-5.384, 7.175-9.311", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4083.wav", "onoffCaption": "door slamming at 2.678-3.911 and tapping clicking clanking at 6.349-9.789", "frequencyCaption": "door slamming one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4128.wav", "onoffCaption": "car horn honking at 0.637-5.149, 6.439-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4198.wav", "onoffCaption": "door knocking at 2.713-5.025, 6.532-9.144 and burping belching at 2.951-4.986, 6.772-8.807", "frequencyCaption": "door knocking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4215.wav", "onoffCaption": "explosion at 0.955-3.949 and cat meowing at 2.343-3.759", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4240.wav", "onoffCaption": "train horn at 3.919-8.249", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4430.wav", "onoffCaption": "burping belching at 0.108-3.367, 4.557-7.03 and explosion at 3.841-8.841", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4465.wav", "onoffCaption": "spraying at 1.705-2.213, 4.527-5.102", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4480.wav", "onoffCaption": "door slamming at 0.287-2.052, 4.508-7.482", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4616.wav", "onoffCaption": "woman laughing at 2.164-4.769 and spraying at 6.976-7.976", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4643.wav", "onoffCaption": "burping belching at 0.436-2.559 and explosion at 0.7-5.621 and spraying at 7.99-9.723", "frequencyCaption": "burping belching one times and explosion one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4758.wav", "onoffCaption": "explosion at 0.898-5.898 and train horn at 2.17-8.639 and duck quacking at 2.778-4.778, 5.802-7.802", "frequencyCaption": "explosion one times and train horn one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4934.wav", "onoffCaption": "duck quacking at 2.328-4.328, 5.244-7.244 and sneeze at 2.934-4.997, 5.606-7.669", "frequencyCaption": "duck quacking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4961.wav", "onoffCaption": "train horn at 0.419-3.219, 5.239-7.706", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4984.wav", "onoffCaption": "door knocking at 0.082-2.434, 4.087-6.099 and thump thud at 1.577-5.244 and dog barking at 5.732-7.732", "frequencyCaption": "door knocking two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2025.wav", "onoffCaption": "cat meowing at 0.213-2.349, 3.579-5.163", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2031.wav", "onoffCaption": "tapping clicking clanking at 0.15-3.59, 4.176-7.053, 7.619-9.914", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_2064.wav", "onoffCaption": "gunshot at 1.132-3.132", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2081.wav", "onoffCaption": "cow mooing at 0.092-3.074, 5.366-8.348", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2217.wav", "onoffCaption": "door knocking at 0.221-3.983, 4.815-8.577 and whistling at 0.271-8.282 and car horn honking at 1.033-5.94", "frequencyCaption": "door knocking two times and whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2242.wav", "onoffCaption": "burping belching at 3.447-6.727, 7.825-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2359.wav", "onoffCaption": "spraying at 0.454-2.149 and whistling at 5.572-10.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2432.wav", "onoffCaption": "whistling at 0.178-9.045 and sneeze at 2.702-5.91", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2467.wav", "onoffCaption": "door slamming at 2.516-4.879, 5.771-8.745", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2473.wav", "onoffCaption": "explosion at 0.536-3.536 and door knocking at 3.283-5.595, 6.181-8.493", "frequencyCaption": "explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2482.wav", "onoffCaption": "thump thud at 2.321-5.988, 6.902-9.402", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2529.wav", "onoffCaption": "cat meowing at 1.85-6.85, 7.75-10.0 and thump thud at 3.577-6.039", "frequencyCaption": "cat meowing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2599.wav", "onoffCaption": "sneeze at 0.377-1.611, 3.551-4.785", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2614.wav", "onoffCaption": "train horn at 0.234-2.701, 3.897-6.569", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2641.wav", "onoffCaption": "woman laughing at 0.106-3.206 and sheep goat bleating at 3.387-5.387, 6.381-8.381 and cat meowing at 5.944-6.971, 8.04-9.067", "frequencyCaption": "woman laughing one times and sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2878.wav", "onoffCaption": "sneeze at 2.708-4.621, 6.857-9.318", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2936.wav", "onoffCaption": "car horn honking at 2.339-6.588", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2963.wav", "onoffCaption": "duck quacking at 0.591-2.591, 3.535-5.535", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2986.wav", "onoffCaption": "thump thud at 0.431-2.931, 3.694-6.194, 6.919-9.854", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4027.wav", "onoffCaption": "burping belching at 2.781-5.304, 7.447-9.97", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4072.wav", "onoffCaption": "gunshot at 0.659-2.659, 4.787-6.787", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4097.wav", "onoffCaption": "duck quacking at 2.279-4.279", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4169.wav", "onoffCaption": "thump thud at 1.232-5.682 and sheep goat bleating at 1.31-3.31", "frequencyCaption": "thump thud one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4201.wav", "onoffCaption": "thump thud at 0.928-4.595 and duck quacking at 1.99-3.99, 4.976-6.976", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4254.wav", "onoffCaption": "burping belching at 0.617-4.117, 5.91-9.41", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4424.wav", "onoffCaption": "explosion at 4.143-9.064", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4471.wav", "onoffCaption": "door knocking at 0.0-6.06 and gunshot at 2.217-4.258", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4494.wav", "onoffCaption": "spraying at 0.609-1.673 and duck quacking at 0.711-2.711, 3.85-5.85", "frequencyCaption": "spraying one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4602.wav", "onoffCaption": "tapping clicking clanking at 2.77-6.21, 7.235-9.279 and door knocking at 3.111-7.621", "frequencyCaption": "tapping clicking clanking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4657.wav", "onoffCaption": "car horn honking at 0.411-4.66, 5.261-7.726", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4719.wav", "onoffCaption": "sneeze at 3.096-5.041", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4920.wav", "onoffCaption": "sneeze at 0.108-2.722, 4.24-5.528, 6.997-9.077", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4975.wav", "onoffCaption": "tapping clicking clanking at 2.428-5.868, 7.423-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4990.wav", "onoffCaption": "cow mooing at 0.034-3.016 and train horn at 0.166-5.881", "frequencyCaption": "cow mooing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2130.wav", "onoffCaption": "door slamming at 0.343-2.826, 4.639-7.122 and sheep goat bleating at 0.499-2.499, 4.127-6.127", "frequencyCaption": "door slamming two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2165.wav", "onoffCaption": "door slamming at 1.723-3.264, 4.189-5.73", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2180.wav", "onoffCaption": "gunshot at 0.152-2.152, 3.02-5.02, 5.995-8.496", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2258.wav", "onoffCaption": "door knocking at 0.871-5.573", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2316.wav", "onoffCaption": "dog barking at 0.47-3.79 and car horn honking at 1.013-5.413 and door knocking at 3.196-5.384", "frequencyCaption": "dog barking one times and car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2343.wav", "onoffCaption": "dog barking at 0.514-2.514 and train horn at 2.989-8.704 and woman laughing at 5.063-7.701", "frequencyCaption": "dog barking one times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2382.wav", "onoffCaption": "duck quacking at 1.839-3.839", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2428.wav", "onoffCaption": "sneeze at 2.249-4.863, 5.588-6.819, 7.797-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2498.wav", "onoffCaption": "sheep goat bleating at 3.531-5.531, 6.48-8.48", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2533.wav", "onoffCaption": "sheep goat bleating at 3.267-5.267", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2566.wav", "onoffCaption": "spraying at 0.111-0.611, 2.052-2.656, 4.625-7.015 and door knocking at 3.291-5.479, 7.437-9.625", "frequencyCaption": "spraying three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2583.wav", "onoffCaption": "woman laughing at 1.036-3.404", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2715.wav", "onoffCaption": "sheep goat bleating at 0.66-5.54, 7.267-9.267", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2740.wav", "onoffCaption": "train horn at 0.131-3.901, 6.116-9.886", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2837.wav", "onoffCaption": "whistling at 0.681-3.656, 5.848-8.823", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2862.wav", "onoffCaption": "dog barking at 0.087-2.525, 4.272-6.272", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2887.wav", "onoffCaption": "sheep goat bleating at 1.009-3.009, 4.409-6.409 and spraying at 2.342-4.778", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2979.wav", "onoffCaption": "dog barking at 1.195-3.195, 4.515-6.515, 7.789-9.789 and woman laughing at 3.19-5.776 and door slamming at 3.266-5.486", "frequencyCaption": "dog barking three times and woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4068.wav", "onoffCaption": "sneeze at 0.318-4.818, 5.939-7.265, 8.03-9.284", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4126.wav", "onoffCaption": "whistling at 2.345-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4173.wav", "onoffCaption": "sheep goat bleating at 0.475-2.475 and cat meowing at 0.576-1.661 and sneeze at 2.632-3.735", "frequencyCaption": "sheep goat bleating one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4196.wav", "onoffCaption": "gunshot at 3.737-5.737", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4300.wav", "onoffCaption": "car horn honking at 1.553-5.14", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4355.wav", "onoffCaption": "cow mooing at 0.892-4.19, 5.52-7.717", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4501.wav", "onoffCaption": "whistling at 0.143-7.798", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4525.wav", "onoffCaption": "woman laughing at 0.406-2.631, 3.797-6.022", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4570.wav", "onoffCaption": "door slamming at 0.189-1.129, 3.598-4.538, 5.299-6.239", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4595.wav", "onoffCaption": "tapping clicking clanking at 0.081-3.521", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4618.wav", "onoffCaption": "explosion at 2.497-4.504 and door knocking at 7.28-10.0", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4703.wav", "onoffCaption": "woman laughing at 1.525-3.725, 5.878-8.086", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4756.wav", "onoffCaption": "cow mooing at 4.047-9.027", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4821.wav", "onoffCaption": "cat meowing at 2.463-4.423 and burping belching at 7.505-10.0", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4874.wav", "onoffCaption": "burping belching at 1.013-4.013, 4.734-6.857", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4891.wav", "onoffCaption": "explosion at 0.084-3.924 and sneeze at 3.631-5.314, 7.57-9.529", "frequencyCaption": "explosion one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2040.wav", "onoffCaption": "sneeze at 0.191-2.136, 2.684-4.629", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2114.wav", "onoffCaption": "gunshot at 0.672-2.672, 3.76-5.76 and tapping clicking clanking at 7.833-10.0", "frequencyCaption": "gunshot two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2141.wav", "onoffCaption": "whistling at 0.057-8.442", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2229.wav", "onoffCaption": "explosion at 1.961-4.961, 6.699-9.699 and sheep goat bleating at 2.522-4.522", "frequencyCaption": "explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2299.wav", "onoffCaption": "door slamming at 0.233-1.748, 2.833-4.348", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2332.wav", "onoffCaption": "train horn at 1.612-4.932", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2367.wav", "onoffCaption": "burping belching at 2.641-5.006 and door slamming at 3.208-5.925 and dog barking at 4.614-6.614", "frequencyCaption": "burping belching one times and door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2416.wav", "onoffCaption": "whistling at 0.288-8.901", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2459.wav", "onoffCaption": "whistling at 1.055-6.636", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2517.wav", "onoffCaption": "dog barking at 2.923-4.923, 6.67-8.67", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2542.wav", "onoffCaption": "cow mooing at 0.253-3.263 and cat meowing at 0.397-1.707 and spraying at 1.526-2.393", "frequencyCaption": "cow mooing one times and cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2731.wav", "onoffCaption": "burping belching at 2.229-8.316", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2764.wav", "onoffCaption": "tapping clicking clanking at 1.128-4.568, 5.185-7.486", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2781.wav", "onoffCaption": "spraying at 2.741-3.345 and dog barking at 2.749-4.749", "frequencyCaption": "spraying one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2809.wav", "onoffCaption": "dog barking at 1.625-3.625, 4.457-6.457", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2813.wav", "onoffCaption": "whistling at 0.073-8.084 and duck quacking at 0.917-2.917, 4.215-6.215 and sheep goat bleating at 2.786-4.786", "frequencyCaption": "whistling one times and duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2846.wav", "onoffCaption": "thump thud at 3.629-6.129", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2908.wav", "onoffCaption": "sheep goat bleating at 0.053-2.053, 4.129-6.129", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2912.wav", "onoffCaption": "thump thud at 0.997-3.497, 4.826-7.597", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4003.wav", "onoffCaption": "burping belching at 2.477-8.564", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4019.wav", "onoffCaption": "explosion at 1.534-5.534, 6.985-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4102.wav", "onoffCaption": "door knocking at 2.936-7.232 and door slamming at 5.741-8.458", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4157.wav", "onoffCaption": "tapping clicking clanking at 1.987-5.427, 6.586-8.643 and car horn honking at 3.357-6.576", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4270.wav", "onoffCaption": "gunshot at 0.201-2.201, 4.088-6.088", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4295.wav", "onoffCaption": "dog barking at 2.052-4.052", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4324.wav", "onoffCaption": "sneeze at 1.859-4.934, 6.814-9.889", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4371.wav", "onoffCaption": "spraying at 0.137-0.869, 2.847-3.911, 5.493-8.077", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4394.wav", "onoffCaption": "cat meowing at 0.311-1.586, 2.6-3.875 and tapping clicking clanking at 2.157-5.597", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4455.wav", "onoffCaption": "duck quacking at 0.261-2.261, 3.583-5.583, 6.193-8.193", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4554.wav", "onoffCaption": "car horn honking at 2.582-6.423, 7.928-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4626.wav", "onoffCaption": "burping belching at 2.098-5.642, 6.868-9.323", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4669.wav", "onoffCaption": "thump thud at 2.425-4.887, 5.471-8.074", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4727.wav", "onoffCaption": "tapping clicking clanking at 2.643-6.083, 7.748-9.849", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4772.wav", "onoffCaption": "sneeze at 0.269-1.5, 3.185-4.416", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4797.wav", "onoffCaption": "thump thud at 1.096-5.014", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4805.wav", "onoffCaption": "cat meowing at 0.861-3.765", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4850.wav", "onoffCaption": "whistling at 3.246-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2015.wav", "onoffCaption": "tapping clicking clanking at 0.308-3.748, 4.664-8.104 and spraying at 1.951-2.732, 4.676-5.457", "frequencyCaption": "tapping clicking clanking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2047.wav", "onoffCaption": "tapping clicking clanking at 3.433-6.873", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2233.wav", "onoffCaption": "train horn at 3.961-6.428", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2234.wav", "onoffCaption": "duck quacking at 0.296-2.296, 2.961-4.961", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2266.wav", "onoffCaption": "car horn honking at 0.862-4.449, 5.271-8.858", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2283.wav", "onoffCaption": "explosion at 0.487-2.58, 3.857-6.725", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2328.wav", "onoffCaption": "whistling at 0.999-6.047 and sheep goat bleating at 1.176-3.176, 3.966-5.966, 6.751-8.751", "frequencyCaption": "whistling one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2398.wav", "onoffCaption": "door knocking at 1.309-4.418, 6.562-9.671 and woman laughing at 1.446-4.146, 4.759-6.771 and door slamming at 2.147-3.662, 5.397-6.397", "frequencyCaption": "door knocking two times and woman laughing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2411.wav", "onoffCaption": "tapping clicking clanking at 0.115-3.555, 5.71-8.203", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2443.wav", "onoffCaption": "cat meowing at 2.254-4.004, 4.915-6.19 and train horn at 2.628-5.988, 6.935-9.02", "frequencyCaption": "cat meowing two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2558.wav", "onoffCaption": "whistling at 1.279-3.288", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2630.wav", "onoffCaption": "whistling at 0.199-9.864 and thump thud at 0.926-4.593", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2662.wav", "onoffCaption": "car horn honking at 2.538-6.938", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2665.wav", "onoffCaption": "door knocking at 0.383-5.085 and door slamming at 7.078-9.561", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2680.wav", "onoffCaption": "cow mooing at 0.205-5.185, 7.154-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2687.wav", "onoffCaption": "woman laughing at 2.787-6.074, 6.661-8.766", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2779.wav", "onoffCaption": "explosion at 2.806-5.067, 6.494-8.755", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2915.wav", "onoffCaption": "sheep goat bleating at 0.649-2.649, 3.203-5.203, 5.869-7.869 and dog barking at 3.465-5.465", "frequencyCaption": "sheep goat bleating three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2947.wav", "onoffCaption": "dog barking at 0.971-2.971 and thump thud at 5.698-10.0", "frequencyCaption": "dog barking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4004.wav", "onoffCaption": "tapping clicking clanking at 1.792-5.232, 7.002-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4056.wav", "onoffCaption": "duck quacking at 3.429-5.429, 6.15-8.15", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4118.wav", "onoffCaption": "sheep goat bleating at 3.47-5.47, 6.715-8.715", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4225.wav", "onoffCaption": "gunshot at 0.572-2.572 and cat meowing at 7.208-8.22", "frequencyCaption": "gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4277.wav", "onoffCaption": "thump thud at 2.987-6.034", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4292.wav", "onoffCaption": "explosion at 0.097-2.185, 4.513-6.926", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4389.wav", "onoffCaption": "burping belching at 0.173-6.117, 6.715-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4400.wav", "onoffCaption": "sheep goat bleating at 2.724-7.604 and thump thud at 3.747-6.247", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4452.wav", "onoffCaption": "cat meowing at 3.312-4.856, 6.196-8.904", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4549.wav", "onoffCaption": "sneeze at 1.038-4.113, 5.958-7.665 and door slamming at 5.131-8.092", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4621.wav", "onoffCaption": "whistling at 0.065-5.24", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4673.wav", "onoffCaption": "tapping clicking clanking at 0.128-3.568", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4696.wav", "onoffCaption": "dog barking at 3.747-9.384", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4768.wav", "onoffCaption": "tapping clicking clanking at 1.367-4.807", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4904.wav", "onoffCaption": "train horn at 3.17-5.637, 7.827-9.933", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4951.wav", "onoffCaption": "sneeze at 2.27-3.434", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4956.wav", "onoffCaption": "cat meowing at 2.574-4.145, 4.744-6.315, 7.461-9.032", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2008.wav", "onoffCaption": "gunshot at 0.098-2.098 and burping belching at 0.614-3.816, 5.977-8.798", "frequencyCaption": "gunshot one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2012.wav", "onoffCaption": "thump thud at 0.055-2.394, 3.578-6.349", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2109.wav", "onoffCaption": "sneeze at 1.144-4.792, 6.589-7.835", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2113.wav", "onoffCaption": "thump thud at 0.539-3.039, 4.952-7.18", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2146.wav", "onoffCaption": "car horn honking at 0.552-3.338, 5.746-8.532", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2261.wav", "onoffCaption": "car horn honking at 0.166-3.341 and dog barking at 2.611-4.611 and sheep goat bleating at 5.761-7.761", "frequencyCaption": "car horn honking one times and dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2284.wav", "onoffCaption": "whistling at 1.837-4.812, 5.973-8.948 and sheep goat bleating at 5.456-8.536", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2360.wav", "onoffCaption": "tapping clicking clanking at 2.241-5.681, 7.788-9.791", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2385.wav", "onoffCaption": "cat meowing at 2.894-3.921", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2444.wav", "onoffCaption": "gunshot at 0.397-2.397", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2510.wav", "onoffCaption": "gunshot at 0.453-2.693, 4.329-6.459", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2545.wav", "onoffCaption": "woman laughing at 0.559-3.145, 5.591-8.177", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2637.wav", "onoffCaption": "woman laughing at 0.151-3.438, 5.006-8.293", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2736.wav", "onoffCaption": "woman laughing at 0.176-3.752, 4.722-8.298 and explosion at 0.858-2.951 and door knocking at 4.024-6.376, 7.281-9.633", "frequencyCaption": "woman laughing two times and explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2814.wav", "onoffCaption": "gunshot at 1.732-3.732, 6.222-8.222", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2841.wav", "onoffCaption": "burping belching at 0.477-3.477 and cat meowing at 2.608-4.358, 6.377-8.127", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2940.wav", "onoffCaption": "door slamming at 0.759-3.476, 5.126-6.45, 7.827-9.13", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4051.wav", "onoffCaption": "dog barking at 0.092-2.092", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4150.wav", "onoffCaption": "whistling at 0.595-5.77, 7.744-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4222.wav", "onoffCaption": "spraying at 0.814-1.755, 3.619-4.369 and cow mooing at 2.19-5.2 and gunshot at 3.053-5.053, 6.329-8.329", "frequencyCaption": "spraying two times and cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4238.wav", "onoffCaption": "whistling at 0.013-7.763 and thump thud at 0.796-3.296, 4.227-6.566 and sneeze at 4.06-6.005, 7.006-8.951", "frequencyCaption": "whistling one times and thump thud two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4288.wav", "onoffCaption": "door knocking at 0.772-3.152 and gunshot at 0.952-2.952, 3.707-5.707, 6.781-8.781", "frequencyCaption": "door knocking one times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4323.wav", "onoffCaption": "door knocking at 2.129-6.831", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4339.wav", "onoffCaption": "door knocking at 2.612-4.702", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4393.wav", "onoffCaption": "explosion at 0.73-3.73, 5.196-8.196", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4407.wav", "onoffCaption": "thump thud at 0.715-3.215 and spraying at 2.174-3.869, 5.959-6.534, 7.932-8.713", "frequencyCaption": "thump thud one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4506.wav", "onoffCaption": "duck quacking at 2.427-4.427, 5.446-7.446", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4674.wav", "onoffCaption": "woman laughing at 0.359-3.459 and gunshot at 0.473-2.473, 4.596-6.596 and cat meowing at 1.726-3.914", "frequencyCaption": "woman laughing one times and gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4691.wav", "onoffCaption": "dog barking at 3.46-5.86", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4775.wav", "onoffCaption": "dog barking at 1.351-3.351, 5.849-7.849", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4790.wav", "onoffCaption": "spraying at 2.35-3.434, 3.962-5.046, 5.902-6.986 and car horn honking at 3.116-7.438", "frequencyCaption": "spraying three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4802.wav", "onoffCaption": "cow mooing at 0.621-3.59, 5.64-8.609 and woman laughing at 1.406-6.445, 7.25-9.348", "frequencyCaption": "cow mooing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4818.wav", "onoffCaption": "burping belching at 1.606-8.774 and gunshot at 5.63-7.63", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4903.wav", "onoffCaption": "train horn at 3.661-8.101", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4919.wav", "onoffCaption": "door knocking at 2.36-5.207, 6.01-8.74", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2079.wav", "onoffCaption": "gunshot at 2.234-4.234", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2137.wav", "onoffCaption": "cow mooing at 2.569-5.538", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2162.wav", "onoffCaption": "thump thud at 0.332-4.707, 5.963-8.463 and car horn honking at 2.236-4.701", "frequencyCaption": "thump thud two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2335.wav", "onoffCaption": "tapping clicking clanking at 0.144-3.584, 5.766-8.185", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2344.wav", "onoffCaption": "whistling at 3.642-8.126", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2534.wav", "onoffCaption": "burping belching at 2.018-4.049, 4.623-6.654, 7.545-9.576", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2561.wav", "onoffCaption": "thump thud at 2.0-5.667, 6.541-8.88", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2584.wav", "onoffCaption": "explosion at 1.234-5.828, 6.719-10.0 and burping belching at 2.001-4.036", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2609.wav", "onoffCaption": "door knocking at 0.956-5.106", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2678.wav", "onoffCaption": "whistling at 2.467-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2712.wav", "onoffCaption": "car horn honking at 2.13-6.53, 7.444-9.957", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2763.wav", "onoffCaption": "sneeze at 2.235-3.466, 4.736-7.053, 7.596-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2786.wav", "onoffCaption": "thump thud at 0.526-3.297, 5.681-7.846 and whistling at 3.429-6.404", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2830.wav", "onoffCaption": "burping belching at 1.964-4.167", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2865.wav", "onoffCaption": "sheep goat bleating at 0.352-2.352 and cow mooing at 1.577-4.559, 5.826-8.808", "frequencyCaption": "sheep goat bleating one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2880.wav", "onoffCaption": "thump thud at 0.527-3.027 and spraying at 5.201-6.896, 9.043-9.543", "frequencyCaption": "thump thud one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4105.wav", "onoffCaption": "thump thud at 2.466-6.916", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4174.wav", "onoffCaption": "train horn at 3.731-5.886 and cat meowing at 6.148-7.719", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4191.wav", "onoffCaption": "tapping clicking clanking at 1.128-4.568, 5.787-8.419", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4249.wav", "onoffCaption": "woman laughing at 3.737-6.809", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4307.wav", "onoffCaption": "woman laughing at 2.521-9.255", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4352.wav", "onoffCaption": "duck quacking at 2.31-4.31, 5.752-7.752 and door knocking at 6.221-8.684", "frequencyCaption": "duck quacking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4376.wav", "onoffCaption": "door slamming at 3.066-4.39, 4.96-6.284, 7.039-8.363", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4439.wav", "onoffCaption": "spraying at 1.833-2.7, 3.222-4.089, 5.309-6.176", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4448.wav", "onoffCaption": "train horn at 2.19-8.25", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4522.wav", "onoffCaption": "door knocking at 0.198-2.575, 4.561-6.938 and woman laughing at 1.012-8.024", "frequencyCaption": "door knocking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4553.wav", "onoffCaption": "cat meowing at 0.582-2.576", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4720.wav", "onoffCaption": "sneeze at 0.726-4.374 and spraying at 1.105-2.352, 3.415-4.662, 5.838-7.085", "frequencyCaption": "sneeze one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4751.wav", "onoffCaption": "cow mooing at 0.27-3.28 and burping belching at 6.449-9.449", "frequencyCaption": "cow mooing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4826.wav", "onoffCaption": "dog barking at 0.773-2.773 and cow mooing at 1.422-6.402, 7.024-10.0 and tapping clicking clanking at 3.237-6.677", "frequencyCaption": "dog barking one times and cow mooing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4857.wav", "onoffCaption": "burping belching at 0.561-2.596, 3.115-5.15, 6.224-8.259", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2036.wav", "onoffCaption": "whistling at 0.608-3.483, 4.251-6.383", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2063.wav", "onoffCaption": "cow mooing at 0.154-3.123 and car horn honking at 5.831-10.0", "frequencyCaption": "cow mooing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2086.wav", "onoffCaption": "thump thud at 2.044-4.815, 6.182-8.692", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2178.wav", "onoffCaption": "cow mooing at 2.45-7.43", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2187.wav", "onoffCaption": "thump thud at 2.111-4.339, 5.423-8.194", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2210.wav", "onoffCaption": "dog barking at 0.214-2.214, 3.126-5.126, 6.017-8.017 and burping belching at 1.459-5.003, 6.152-8.659", "frequencyCaption": "dog barking three times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2245.wav", "onoffCaption": "gunshot at 0.323-2.323", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2311.wav", "onoffCaption": "duck quacking at 0.64-2.64, 4.525-6.525, 7.803-9.803", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2435.wav", "onoffCaption": "gunshot at 2.92-4.92, 6.165-8.165", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2613.wav", "onoffCaption": "sneeze at 3.626-4.79", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2646.wav", "onoffCaption": "sneeze at 1.137-2.431, 3.879-6.36", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2708.wav", "onoffCaption": "woman laughing at 1.941-6.98", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2747.wav", "onoffCaption": "train horn at 2.336-6.536, 7.35-9.66", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2931.wav", "onoffCaption": "woman laughing at 2.301-4.495", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4020.wav", "onoffCaption": "dog barking at 2.093-4.093, 6.532-8.532 and gunshot at 5.252-7.252 and sneeze at 5.483-7.19", "frequencyCaption": "dog barking two times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4075.wav", "onoffCaption": "cow mooing at 0.295-3.277", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4090.wav", "onoffCaption": "sheep goat bleating at 3.411-5.411, 6.681-8.681", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4121.wav", "onoffCaption": "thump thud at 0.085-4.46, 5.692-8.463 and door slamming at 0.091-2.091", "frequencyCaption": "thump thud two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4253.wav", "onoffCaption": "woman laughing at 3.8-7.087", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4348.wav", "onoffCaption": "sneeze at 2.276-3.564, 4.675-6.358", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4423.wav", "onoffCaption": "door slamming at 2.555-3.858 and spraying at 7.213-9.341", "frequencyCaption": "door slamming one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4476.wav", "onoffCaption": "burping belching at 0.592-4.928, 6.205-8.838", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4489.wav", "onoffCaption": "duck quacking at 1.562-3.562, 4.618-6.618, 7.146-9.146", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4493.wav", "onoffCaption": "burping belching at 1.073-4.332, 5.408-8.408 and cat meowing at 4.695-6.585, 7.258-9.148", "frequencyCaption": "burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4538.wav", "onoffCaption": "thump thud at 0.955-3.294, 4.777-7.332 and cat meowing at 1.094-2.989, 5.376-7.316", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4577.wav", "onoffCaption": "train horn at 0.185-4.366 and gunshot at 7.36-9.36", "frequencyCaption": "train horn one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4588.wav", "onoffCaption": "sneeze at 0.785-3.024, 5.081-7.194, 7.912-9.975", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4592.wav", "onoffCaption": "duck quacking at 2.429-4.429, 6.607-8.607", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4605.wav", "onoffCaption": "cat meowing at 0.056-1.056, 2.31-4.677, 5.816-7.352 and door slamming at 1.937-2.437, 3.086-3.586", "frequencyCaption": "cat meowing three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4704.wav", "onoffCaption": "spraying at 2.106-2.681, 3.676-6.112 and cat meowing at 2.174-3.44, 4.07-5.336, 6.08-7.346", "frequencyCaption": "spraying two times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4869.wav", "onoffCaption": "gunshot at 2.43-4.43 and sneeze at 6.425-10.0", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4873.wav", "onoffCaption": "woman laughing at 2.935-6.02, 6.817-9.1", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4896.wav", "onoffCaption": "cow mooing at 1.844-5.142, 6.268-9.566", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4927.wav", "onoffCaption": "cow mooing at 2.408-5.418, 7.326-9.414", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4968.wav", "onoffCaption": "duck quacking at 0.788-2.788, 4.242-6.242 and cow mooing at 2.553-5.522, 7.683-9.783", "frequencyCaption": "duck quacking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4972.wav", "onoffCaption": "gunshot at 0.697-2.697, 4.279-6.279", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4997.wav", "onoffCaption": "sneeze at 0.037-1.283 and dog barking at 4.35-6.35, 7.332-9.332", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2051.wav", "onoffCaption": "burping belching at 0.475-3.296, 5.659-8.092", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2222.wav", "onoffCaption": "explosion at 3.458-5.551 and cat meowing at 5.386-6.386, 7.208-8.208", "frequencyCaption": "explosion one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2277.wav", "onoffCaption": "explosion at 1.115-3.208, 4.127-6.189", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2292.wav", "onoffCaption": "gunshot at 1.525-3.525", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2339.wav", "onoffCaption": "tapping clicking clanking at 3.039-6.479", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2389.wav", "onoffCaption": "explosion at 1.037-6.037, 6.705-9.214", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2407.wav", "onoffCaption": "duck quacking at 0.106-2.106, 3.694-5.694", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2452.wav", "onoffCaption": "gunshot at 2.289-4.289, 5.735-7.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2460.wav", "onoffCaption": "duck quacking at 0.55-2.55 and tapping clicking clanking at 4.483-7.923", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2485.wav", "onoffCaption": "tapping clicking clanking at 2.678-6.118", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2549.wav", "onoffCaption": "thump thud at 1.047-4.094, 4.803-7.303", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2621.wav", "onoffCaption": "car horn honking at 1.926-4.839, 6.023-8.936", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2674.wav", "onoffCaption": "door knocking at 0.096-3.846, 5.923-9.673", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2691.wav", "onoffCaption": "door knocking at 0.614-4.91, 5.98-8.229 and cow mooing at 4.553-7.535", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2818.wav", "onoffCaption": "spraying at 2.945-3.945, 5.674-6.674, 7.659-8.659", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2903.wav", "onoffCaption": "cow mooing at 2.138-5.436, 6.232-9.201", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2964.wav", "onoffCaption": "duck quacking at 3.185-5.185, 6.724-8.724", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2981.wav", "onoffCaption": "woman laughing at 0.149-2.349, 3.471-6.171", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4012.wav", "onoffCaption": "sheep goat bleating at 2.964-6.044", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4047.wav", "onoffCaption": "whistling at 1.46-9.065", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4109.wav", "onoffCaption": "gunshot at 2.838-4.838, 6.567-8.567", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4206.wav", "onoffCaption": "car horn honking at 1.392-6.299, 7.213-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4261.wav", "onoffCaption": "gunshot at 0.303-2.809, 5.036-7.036", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4284.wav", "onoffCaption": "spraying at 0.458-1.062 and cat meowing at 3.769-5.044, 5.973-7.248, 7.752-9.027", "frequencyCaption": "spraying one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4411.wav", "onoffCaption": "train horn at 0.37-2.507", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4444.wav", "onoffCaption": "duck quacking at 0.591-2.591", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4637.wav", "onoffCaption": "cow mooing at 3.246-6.228", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4650.wav", "onoffCaption": "dog barking at 3.055-5.055", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4687.wav", "onoffCaption": "car horn honking at 2.758-4.758, 7.034-9.034", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4915.wav", "onoffCaption": "gunshot at 0.044-2.044, 2.611-4.611, 5.506-7.506", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4940.wav", "onoffCaption": "explosion at 0.203-3.203, 5.192-8.192", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2004.wav", "onoffCaption": "burping belching at 2.243-4.337, 5.194-8.075", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2105.wav", "onoffCaption": "gunshot at 0.096-2.336, 3.644-5.774, 6.28-8.28 and door slamming at 0.203-0.703", "frequencyCaption": "gunshot three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2121.wav", "onoffCaption": "door knocking at 3.089-5.713, 7.505-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2150.wav", "onoffCaption": "tapping clicking clanking at 3.558-6.998 and duck quacking at 6.009-8.009", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2174.wav", "onoffCaption": "dog barking at 0.653-2.653, 4.141-6.141", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2238.wav", "onoffCaption": "sheep goat bleating at 2.407-6.047, 6.802-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2249.wav", "onoffCaption": "spraying at 0.058-0.925", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2288.wav", "onoffCaption": "thump thud at 3.546-5.885 and spraying at 7.512-8.687", "frequencyCaption": "thump thud one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2323.wav", "onoffCaption": "door slamming at 0.137-2.137, 2.976-4.976, 6.469-8.469", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2352.wav", "onoffCaption": "whistling at 1.105-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2376.wav", "onoffCaption": "woman laughing at 2.081-5.368, 6.66-9.298 and gunshot at 3.24-5.24, 5.749-7.749", "frequencyCaption": "woman laughing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2393.wav", "onoffCaption": "sneeze at 0.13-4.63, 5.693-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2439.wav", "onoffCaption": "thump thud at 1.519-5.186 and woman laughing at 3.888-5.97, 7.447-9.529", "frequencyCaption": "thump thud one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2448.wav", "onoffCaption": "whistling at 2.912-8.412 and tapping clicking clanking at 5.008-8.448 and gunshot at 7.107-9.107", "frequencyCaption": "whistling one times and tapping clicking clanking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2489.wav", "onoffCaption": "tapping clicking clanking at 1.137-4.577, 6.143-8.158", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2506.wav", "onoffCaption": "tapping clicking clanking at 0.235-3.675, 5.904-9.344", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2522.wav", "onoffCaption": "thump thud at 0.022-2.793 and car horn honking at 1.65-6.557", "frequencyCaption": "thump thud one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2553.wav", "onoffCaption": "spraying at 0.053-0.628, 1.756-2.82, 3.538-4.795 and train horn at 0.586-5.026, 7.084-9.849", "frequencyCaption": "spraying three times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2577.wav", "onoffCaption": "whistling at 0.231-5.731, 6.91-8.919", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2592.wav", "onoffCaption": "explosion at 2.034-4.562, 5.841-8.369", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2704.wav", "onoffCaption": "spraying at 1.223-1.792, 2.643-3.393 and dog barking at 4.984-6.984", "frequencyCaption": "spraying two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2720.wav", "onoffCaption": "whistling at 2.643-4.872, 5.859-8.467", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2775.wav", "onoffCaption": "sheep goat bleating at 2.154-4.154, 6.08-8.08", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2790.wav", "onoffCaption": "sneeze at 2.831-4.894, 6.753-8.816", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2802.wav", "onoffCaption": "whistling at 3.011-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2857.wav", "onoffCaption": "sneeze at 3.411-6.025, 6.998-9.45", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2873.wav", "onoffCaption": "cow mooing at 0.596-3.578, 6.07-9.052", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2896.wav", "onoffCaption": "explosion at 3.203-8.203", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2919.wav", "onoffCaption": "thump thud at 0.206-2.977, 3.705-6.476", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2956.wav", "onoffCaption": "woman laughing at 1.045-3.637, 4.725-7.363", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2968.wav", "onoffCaption": "sneeze at 0.034-1.36 and tapping clicking clanking at 5.197-8.637", "frequencyCaption": "sneeze one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4008.wav", "onoffCaption": "woman laughing at 1.326-3.52, 5.953-8.147", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4079.wav", "onoffCaption": "spraying at 0.063-1.549 and car horn honking at 3.496-6.422", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4113.wav", "onoffCaption": "car horn honking at 0.329-3.983, 4.793-7.611 and cow mooing at 5.071-8.04", "frequencyCaption": "car horn honking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4146.wav", "onoffCaption": "sheep goat bleating at 3.618-5.618, 6.355-8.355", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4162.wav", "onoffCaption": "tapping clicking clanking at 1.893-5.333, 7.249-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4187.wav", "onoffCaption": "cow mooing at 0.198-3.496, 5.511-7.81", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4234.wav", "onoffCaption": "door knocking at 1.232-3.695 and duck quacking at 7.38-9.38", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4311.wav", "onoffCaption": "gunshot at 0.178-2.178, 3.114-5.114, 5.762-8.002 and explosion at 2.45-7.45", "frequencyCaption": "gunshot three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4335.wav", "onoffCaption": "sheep goat bleating at 0.499-2.499, 4.063-6.063, 7.651-9.651", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4360.wav", "onoffCaption": "train horn at 2.318-6.648", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4385.wav", "onoffCaption": "burping belching at 0.088-5.088, 5.724-7.818", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4510.wav", "onoffCaption": "sneeze at 0.848-2.807 and car horn honking at 7.218-10.0", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4534.wav", "onoffCaption": "duck quacking at 0.48-2.48, 3.415-5.415, 7.851-9.851", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4545.wav", "onoffCaption": "train horn at 1.683-4.883, 6.583-9.433", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4662.wav", "onoffCaption": "thump thud at 2.995-5.334, 5.838-8.209 and door knocking at 4.48-8.033", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4678.wav", "onoffCaption": "cow mooing at 1.341-4.351, 5.937-8.017", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4736.wav", "onoffCaption": "door knocking at 2.538-7.24 and sheep goat bleating at 5.123-7.123", "frequencyCaption": "door knocking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4747.wav", "onoffCaption": "train horn at 2.593-4.748, 7.117-9.772", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4763.wav", "onoffCaption": "door knocking at 0.231-3.981", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4779.wav", "onoffCaption": "cow mooing at 3.146-6.156 and door knocking at 3.765-6.265, 7.273-9.33", "frequencyCaption": "cow mooing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4786.wav", "onoffCaption": "train horn at 1.798-6.875 and dog barking at 5.284-7.284", "frequencyCaption": "train horn one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4814.wav", "onoffCaption": "woman laughing at 2.443-5.238, 7.538-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4830.wav", "onoffCaption": "spraying at 1.24-2.107, 4.396-5.263, 6.077-6.944", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4841.wav", "onoffCaption": "explosion at 0.673-5.673", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_7.wav", "onoffCaption": "sheep goat bleating at 1.687-3.687, 4.963-6.963", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2075.wav", "onoffCaption": "sneeze at 2.897-7.397", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2090.wav", "onoffCaption": "cat meowing at 3.18-4.393, 6.703-7.916", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2191.wav", "onoffCaption": "door knocking at 1.462-3.839 and thump thud at 5.248-9.623", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2206.wav", "onoffCaption": "sneeze at 2.72-4.008, 4.819-6.502, 7.747-9.073", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2253.wav", "onoffCaption": "sneeze at 0.069-1.161, 1.844-4.595, 5.283-7.491", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2307.wav", "onoffCaption": "train horn at 0.621-3.061, 5.358-8.033", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2348.wav", "onoffCaption": "door slamming at 2.104-3.869, 5.812-8.786 and dog barking at 2.325-4.325", "frequencyCaption": "door slamming two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2423.wav", "onoffCaption": "gunshot at 2.171-4.672, 7.152-9.152", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2538.wav", "onoffCaption": "door knocking at 2.954-6.329", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2605.wav", "onoffCaption": "sheep goat bleating at 2.344-4.344, 6.559-8.559", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2650.wav", "onoffCaption": "burping belching at 1.323-4.829, 5.63-7.737", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2751.wav", "onoffCaption": "cow mooing at 0.385-3.354, 5.307-7.518", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2826.wav", "onoffCaption": "thump thud at 2.171-6.546, 7.936-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2927.wav", "onoffCaption": "sheep goat bleating at 2.616-4.616, 5.942-7.942", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4036.wav", "onoffCaption": "door slamming at 1.098-3.294 and explosion at 2.32-5.32, 6.136-9.136 and sneeze at 8.145-9.309", "frequencyCaption": "door slamming one times and explosion two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4086.wav", "onoffCaption": "thump thud at 0.396-2.735, 3.807-6.146, 7.197-9.536", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4137.wav", "onoffCaption": "explosion at 0.362-3.362, 5.024-8.024 and tapping clicking clanking at 3.969-7.409", "frequencyCaption": "explosion two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4245.wav", "onoffCaption": "car horn honking at 0.888-4.729, 6.844-9.24", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4344.wav", "onoffCaption": "gunshot at 2.648-4.648, 5.271-7.312", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4460.wav", "onoffCaption": "gunshot at 2.04-4.04, 4.742-6.742", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4485.wav", "onoffCaption": "cat meowing at 0.363-2.303, 3.082-5.022", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4561.wav", "onoffCaption": "cat meowing at 0.256-5.256", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4584.wav", "onoffCaption": "thump thud at 0.001-4.451, 5.57-10.0 and sneeze at 0.165-1.665", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4609.wav", "onoffCaption": "duck quacking at 0.204-2.204 and sneeze at 4.938-7.341", "frequencyCaption": "duck quacking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4613.wav", "onoffCaption": "cat meowing at 0.077-1.104, 1.625-2.652 and woman laughing at 5.845-8.94", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4708.wav", "onoffCaption": "burping belching at 2.626-4.751, 5.854-7.979", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4712.wav", "onoffCaption": "sneeze at 0.454-2.066, 4.387-5.915 and sheep goat bleating at 4.63-6.63", "frequencyCaption": "sneeze two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4865.wav", "onoffCaption": "thump thud at 2.49-4.718, 6.791-9.291", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4880.wav", "onoffCaption": "sheep goat bleating at 0.069-2.069 and spraying at 1.372-2.122", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4931.wav", "onoffCaption": "door slamming at 2.992-4.516, 5.7-7.224", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4964.wav", "onoffCaption": "door knocking at 0.461-4.9, 5.822-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4981.wav", "onoffCaption": "train horn at 0.592-8.792", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2020.wav", "onoffCaption": "burping belching at 0.805-4.805, 7.041-9.152", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2072.wav", "onoffCaption": "gunshot at 0.343-2.343, 4.093-6.093 and whistling at 0.77-6.27", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2097.wav", "onoffCaption": "cat meowing at 0.499-2.07 and woman laughing at 4.3-6.382", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2169.wav", "onoffCaption": "thump thud at 1.556-4.327 and gunshot at 3.176-5.176", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2201.wav", "onoffCaption": "door slamming at 3.56-4.365, 6.282-7.087, 8.662-9.467", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2424.wav", "onoffCaption": "cow mooing at 1.181-4.191 and explosion at 1.485-3.573 and gunshot at 2.08-4.08", "frequencyCaption": "cow mooing one times and explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2471.wav", "onoffCaption": "tapping clicking clanking at 0.354-3.794 and car horn honking at 0.545-5.452", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2476.wav", "onoffCaption": "woman laughing at 2.307-4.59", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2493.wav", "onoffCaption": "gunshot at 2.205-4.205 and thump thud at 6.122-9.789", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2588.wav", "onoffCaption": "sheep goat bleating at 0.077-4.957, 5.948-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2657.wav", "onoffCaption": "door slamming at 0.527-2.747, 3.423-4.947, 6.797-8.71", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2869.wav", "onoffCaption": "burping belching at 0.032-3.472, 4.014-6.414 and explosion at 0.35-3.35, 4.571-7.435 and whistling at 3.319-8.494", "frequencyCaption": "burping belching two times and explosion two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2920.wav", "onoffCaption": "burping belching at 1.183-5.183, 6.898-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2972.wav", "onoffCaption": "thump thud at 0.306-2.806, 3.519-6.279, 6.88-9.06", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2997.wav", "onoffCaption": "burping belching at 1.791-5.791, 7.164-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4031.wav", "onoffCaption": "sheep goat bleating at 0.181-2.181 and door knocking at 0.401-3.457", "frequencyCaption": "sheep goat bleating one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4063.wav", "onoffCaption": "gunshot at 0.06-2.101, 3.276-5.317, 6.463-8.504", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4178.wav", "onoffCaption": "gunshot at 1.149-3.149, 3.925-5.925, 6.522-8.522 and tapping clicking clanking at 5.417-8.857", "frequencyCaption": "gunshot three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4210.wav", "onoffCaption": "cow mooing at 2.521-5.49", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4242.wav", "onoffCaption": "duck quacking at 2.613-4.613", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4359.wav", "onoffCaption": "door slamming at 2.052-3.169, 3.806-4.923, 5.493-6.61 and woman laughing at 3.52-6.106", "frequencyCaption": "door slamming three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4435.wav", "onoffCaption": "spraying at 4.083-4.934", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4467.wav", "onoffCaption": "sheep goat bleating at 1.74-3.74 and thump thud at 6.828-9.328", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4482.wav", "onoffCaption": "explosion at 2.5-7.5 and duck quacking at 4.221-6.221", "frequencyCaption": "explosion one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4599.wav", "onoffCaption": "sneeze at 0.066-1.16 and door knocking at 5.548-9.101", "frequencyCaption": "sneeze one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4614.wav", "onoffCaption": "duck quacking at 2.031-4.031 and train horn at 2.112-4.267, 5.729-8.564", "frequencyCaption": "duck quacking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4646.wav", "onoffCaption": "sneeze at 0.795-4.443 and sheep goat bleating at 6.423-8.423", "frequencyCaption": "sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4878.wav", "onoffCaption": "dog barking at 0.779-2.779", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4963.wav", "onoffCaption": "door knocking at 1.64-4.14, 5.135-7.775", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4986.wav", "onoffCaption": "explosion at 0.519-2.612, 3.807-5.9, 7.053-9.146", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_2027.wav", "onoffCaption": "cat meowing at 2.271-5.512, 7.359-8.499", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2126.wav", "onoffCaption": "spraying at 1.5-3.628, 4.53-6.658, 7.824-9.952", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2254.wav", "onoffCaption": "spraying at 3.935-4.51, 5.563-6.344, 7.392-9.784", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2300.wav", "onoffCaption": "sneeze at 3.324-5.8, 6.929-9.405", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2355.wav", "onoffCaption": "duck quacking at 0.3-2.3, 3.841-5.841, 6.667-8.667", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2494.wav", "onoffCaption": "car horn honking at 1.413-3.926, 4.97-7.349 and tapping clicking clanking at 2.467-5.907, 7.887-10.0", "frequencyCaption": "car horn honking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2570.wav", "onoffCaption": "door slamming at 0.573-1.876, 2.84-4.143", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2595.wav", "onoffCaption": "woman laughing at 0.777-3.145, 5.247-7.615 and sheep goat bleating at 1.517-3.517", "frequencyCaption": "woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2602.wav", "onoffCaption": "burping belching at 3.235-6.235, 7.539-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2618.wav", "onoffCaption": "thump thud at 1.002-3.773", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2703.wav", "onoffCaption": "cow mooing at 0.447-5.427, 7.486-9.883", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2719.wav", "onoffCaption": "burping belching at 0.966-4.246, 5.414-7.779 and cat meowing at 3.195-4.812", "frequencyCaption": "burping belching two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2756.wav", "onoffCaption": "thump thud at 1.042-3.27, 4.464-6.692", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2874.wav", "onoffCaption": "cow mooing at 2.648-5.63, 6.95-9.932", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2891.wav", "onoffCaption": "explosion at 0.143-3.143", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2975.wav", "onoffCaption": "burping belching at 2.339-5.619, 6.718-9.701", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2990.wav", "onoffCaption": "burping belching at 0.11-2.141, 2.953-4.984 and woman laughing at 7.255-9.538", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4064.wav", "onoffCaption": "whistling at 1.653-6.137, 7.449-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4081.wav", "onoffCaption": "door knocking at 0.283-4.722 and woman laughing at 1.01-3.71, 4.329-6.947", "frequencyCaption": "door knocking one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4165.wav", "onoffCaption": "tapping clicking clanking at 1.022-4.462, 5.978-9.418 and gunshot at 1.274-3.274, 5.12-7.12", "frequencyCaption": "tapping clicking clanking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4180.wav", "onoffCaption": "cow mooing at 1.464-6.444", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4217.wav", "onoffCaption": "dog barking at 1.46-3.46, 4.833-6.833 and cat meowing at 1.471-3.411, 5.693-7.237", "frequencyCaption": "dog barking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4316.wav", "onoffCaption": "car horn honking at 0.347-3.165 and cow mooing at 5.694-10.0", "frequencyCaption": "car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4428.wav", "onoffCaption": "door knocking at 1.117-3.338, 4.015-6.392, 7.322-9.807", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4432.wav", "onoffCaption": "door knocking at 0.13-2.977 and train horn at 1.697-4.577, 5.557-8.437", "frequencyCaption": "door knocking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4529.wav", "onoffCaption": "sheep goat bleating at 1.449-3.449, 4.761-6.761 and door knocking at 4.19-6.69, 7.357-9.857", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4533.wav", "onoffCaption": "door slamming at 0.536-1.555, 2.537-3.537, 5.811-6.616", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4641.wav", "onoffCaption": "explosion at 0.631-5.631 and sheep goat bleating at 2.075-5.995", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4740.wav", "onoffCaption": "tapping clicking clanking at 0.306-3.746, 5.611-8.049", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4837.wav", "onoffCaption": "thump thud at 1.079-5.454", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4936.wav", "onoffCaption": "spraying at 0.214-2.798, 3.672-6.256, 7.074-9.658", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2019.wav", "onoffCaption": "dog barking at 0.089-2.089, 2.72-4.72, 5.569-7.569 and train horn at 4.343-7.877 and door slamming at 5.704-7.924", "frequencyCaption": "dog barking three times and train horn one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2068.wav", "onoffCaption": "duck quacking at 0.047-2.047, 2.698-4.698, 5.474-7.474 and cat meowing at 2.237-3.377 and dog barking at 6.219-8.219", "frequencyCaption": "duck quacking three times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2102.wav", "onoffCaption": "spraying at 1.298-3.06, 3.89-5.065, 6.352-8.788", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2173.wav", "onoffCaption": "car horn honking at 0.126-3.78 and cat meowing at 0.557-4.917, 6.481-7.628", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2196.wav", "onoffCaption": "sneeze at 1.355-4.43, 5.374-8.449", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2371.wav", "onoffCaption": "cat meowing at 1.912-3.178, 4.339-5.605, 6.882-8.148", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2394.wav", "onoffCaption": "tapping clicking clanking at 2.112-5.552, 6.278-9.718", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2525.wav", "onoffCaption": "train horn at 0.234-3.768, 4.385-7.919 and sneeze at 5.081-8.729", "frequencyCaption": "train horn two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2554.wav", "onoffCaption": "thump thud at 2.155-4.655", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2669.wav", "onoffCaption": "door knocking at 0.303-3.856 and train horn at 5.632-9.7", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2727.wav", "onoffCaption": "train horn at 0.066-4.066, 4.98-8.98 and cat meowing at 1.727-2.727", "frequencyCaption": "train horn two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2772.wav", "onoffCaption": "burping belching at 1.808-4.808, 5.553-8.183 and car horn honking at 3.667-8.067", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2821.wav", "onoffCaption": "door knocking at 0.226-3.073, 4.879-7.422", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2850.wav", "onoffCaption": "dog barking at 0.215-2.215, 3.162-5.162, 6.429-8.429", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4130.wav", "onoffCaption": "dog barking at 0.005-2.005 and explosion at 3.331-6.387, 7.394-10.0", "frequencyCaption": "dog barking one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_4141.wav", "onoffCaption": "burping belching at 1.866-6.202, 7.316-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4229.wav", "onoffCaption": "cow mooing at 2.388-5.398, 7.519-9.548", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4258.wav", "onoffCaption": "explosion at 2.276-5.276, 6.95-9.95", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4332.wav", "onoffCaption": "cat meowing at 1.898-3.482, 5.203-6.935", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4343.wav", "onoffCaption": "sheep goat bleating at 0.153-2.153, 2.766-4.766", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4498.wav", "onoffCaption": "door slamming at 1.793-2.966, 4.605-6.129, 8.222-9.162", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4517.wav", "onoffCaption": "cat meowing at 0.438-1.578, 3.851-6.376 and door slamming at 2.388-4.388, 4.942-6.942", "frequencyCaption": "cat meowing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4566.wav", "onoffCaption": "whistling at 2.552-7.727 and car horn honking at 3.898-8.41", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4583.wav", "onoffCaption": "spraying at 3.279-5.671, 6.189-7.056, 9.134-9.866", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4715.wav", "onoffCaption": "duck quacking at 1.168-3.168, 3.913-5.913, 7.163-9.163 and dog barking at 1.171-3.171", "frequencyCaption": "duck quacking three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4764.wav", "onoffCaption": "woman laughing at 1.085-3.368 and door slamming at 5.864-6.702, 7.848-8.987", "frequencyCaption": "woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4781.wav", "onoffCaption": "woman laughing at 0.104-2.709, 3.749-5.831, 6.999-9.199", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_4813.wav", "onoffCaption": "burping belching at 0.089-3.089, 4.311-7.192", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4862.wav", "onoffCaption": "cow mooing at 1.786-4.768, 5.558-8.527", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4887.wav", "onoffCaption": "woman laughing at 0.005-3.292", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4908.wav", "onoffCaption": "gunshot at 0.567-2.608, 3.838-5.879, 6.969-9.01", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4979.wav", "onoffCaption": "cow mooing at 0.323-4.752, 5.985-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2003.wav", "onoffCaption": "door slamming at 2.017-4.475 and dog barking at 6.941-8.941", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2056.wav", "onoffCaption": "whistling at 2.868-4.877 and burping belching at 2.97-6.149, 7.093-10.0", "frequencyCaption": "whistling one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2118.wav", "onoffCaption": "door slamming at 1.955-3.104, 4.415-5.564, 6.925-8.074 and car horn honking at 4.279-8.679", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2157.wav", "onoffCaption": "door slamming at 3.384-4.617, 5.78-7.013, 7.976-9.209", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2225.wav", "onoffCaption": "duck quacking at 2.673-4.673, 6.37-8.37 and spraying at 3.012-5.596", "frequencyCaption": "duck quacking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2324.wav", "onoffCaption": "cow mooing at 2.87-6.168, 7.884-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2400.wav", "onoffCaption": "cat meowing at 0.016-1.229", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2455.wav", "onoffCaption": "cat meowing at 1.766-2.778, 3.833-4.973, 5.712-7.444", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2501.wav", "onoffCaption": "gunshot at 0.124-2.124, 2.96-4.96 and door slamming at 0.289-1.438", "frequencyCaption": "gunshot two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2673.wav", "onoffCaption": "woman laughing at 0.241-2.466", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2696.wav", "onoffCaption": "cow mooing at 2.987-6.285", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2768.wav", "onoffCaption": "sneeze at 0.124-4.653, 5.641-6.935", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2797.wav", "onoffCaption": "sheep goat bleating at 0.044-2.044, 3.147-5.147, 6.598-8.598", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2805.wav", "onoffCaption": "sneeze at 0.029-1.123, 3.204-5.521", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2904.wav", "onoffCaption": "whistling at 0.037-9.702 and dog barking at 0.082-2.082 and burping belching at 6.199-9.199", "frequencyCaption": "whistling one times and dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2951.wav", "onoffCaption": "whistling at 2.043-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4015.wav", "onoffCaption": "car horn honking at 2.146-6.468, 7.709-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4114.wav", "onoffCaption": "sneeze at 0.7-3.103, 5.122-7.525", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4233.wav", "onoffCaption": "whistling at 1.754-4.729 and woman laughing at 2.721-5.076", "frequencyCaption": "whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4266.wav", "onoffCaption": "burping belching at 1.694-5.563, 6.453-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4283.wav", "onoffCaption": "cow mooing at 0.564-3.862", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4299.wav", "onoffCaption": "car horn honking at 0.783-5.69, 7.139-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4367.wav", "onoffCaption": "train horn at 0.133-3.373, 5.09-7.557 and dog barking at 2.429-4.429, 6.478-8.878", "frequencyCaption": "train horn two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4382.wav", "onoffCaption": "gunshot at 0.052-2.553, 3.667-5.667, 6.354-8.354", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4398.wav", "onoffCaption": "cat meowing at 0.031-1.18 and gunshot at 0.572-2.742, 5.068-7.238", "frequencyCaption": "cat meowing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4443.wav", "onoffCaption": "whistling at 0.006-7.756", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4459.wav", "onoffCaption": "cat meowing at 0.808-2.363, 3.023-4.035 and tapping clicking clanking at 6.989-10.0", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4542.wav", "onoffCaption": "explosion at 0.903-4.03, 6.338-9.465", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4558.wav", "onoffCaption": "whistling at 1.603-9.258", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4630.wav", "onoffCaption": "cow mooing at 3.094-6.076, 7.671-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4665.wav", "onoffCaption": "burping belching at 4.194-7.453", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4680.wav", "onoffCaption": "cow mooing at 0.292-3.302, 4.366-7.335 and duck quacking at 6.865-8.865", "frequencyCaption": "cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4731.wav", "onoffCaption": "burping belching at 0.174-3.174, 4.471-7.471 and woman laughing at 1.095-3.733", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4846.wav", "onoffCaption": "sheep goat bleating at 3.216-5.216, 7.445-9.445", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4947.wav", "onoffCaption": "cow mooing at 0.181-3.15, 4.35-7.332 and gunshot at 2.445-4.946, 5.779-7.779 and whistling at 4.017-6.892, 7.993-10.0", "frequencyCaption": "cow mooing two times and gunshot two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_2042.wav", "onoffCaption": "sneeze at 2.179-4.175 and thump thud at 2.763-5.81, 7.173-9.414", "frequencyCaption": "sneeze one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2159.wav", "onoffCaption": "door knocking at 2.29-5.137, 6.138-8.906", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2231.wav", "onoffCaption": "cow mooing at 3.414-6.396, 7.031-9.087", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2264.wav", "onoffCaption": "car horn honking at 3.121-5.468", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2270.wav", "onoffCaption": "train horn at 1.497-5.678, 6.589-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2281.wav", "onoffCaption": "spraying at 0.544-2.306, 3.813-5.575 and train horn at 2.976-6.976", "frequencyCaption": "spraying two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2295.wav", "onoffCaption": "spraying at 3.423-4.05, 5.308-5.912, 7.048-7.632", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2414.wav", "onoffCaption": "sheep goat bleating at 1.213-3.213, 5.402-7.402", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2626.wav", "onoffCaption": "cow mooing at 0.604-3.902, 5.015-7.984", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2632.wav", "onoffCaption": "duck quacking at 2.26-4.26, 6.028-8.028", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2667.wav", "onoffCaption": "sheep goat bleating at 0.259-2.259, 3.658-5.658 and cat meowing at 2.779-4.511", "frequencyCaption": "sheep goat bleating two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2682.wav", "onoffCaption": "cat meowing at 3.69-5.106, 7.025-8.965", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2729.wav", "onoffCaption": "woman laughing at 0.702-3.774, 4.528-7.6 and burping belching at 3.235-5.329", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2799.wav", "onoffCaption": "sheep goat bleating at 0.527-2.527", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2910.wav", "onoffCaption": "sneeze at 1.626-3.309, 3.833-5.516, 6.281-7.964", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2945.wav", "onoffCaption": "thump thud at 0.075-2.846, 3.421-6.192 and burping belching at 7.915-10.0", "frequencyCaption": "thump thud two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4001.wav", "onoffCaption": "train horn at 0.358-6.073, 7.942-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4040.wav", "onoffCaption": "door slamming at 1.741-4.522", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4054.wav", "onoffCaption": "explosion at 0.122-5.122", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4227.wav", "onoffCaption": "tapping clicking clanking at 0.449-3.889, 4.56-6.915, 7.848-9.866", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_4272.wav", "onoffCaption": "sheep goat bleating at 0.103-2.103 and thump thud at 5.105-9.48", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4297.wav", "onoffCaption": "door knocking at 1.869-3.891, 4.557-6.579, 7.312-9.334", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4328.wav", "onoffCaption": "door slamming at 0.071-2.5, 3.7-5.618, 6.407-8.32", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4369.wav", "onoffCaption": "thump thud at 4.339-6.839", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4402.wav", "onoffCaption": "train horn at 3.277-6.597", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4416.wav", "onoffCaption": "sneeze at 0.054-1.641, 2.766-4.353, 5.195-6.782", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4457.wav", "onoffCaption": "tapping clicking clanking at 1.373-4.813, 5.85-8.035", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4519.wav", "onoffCaption": "burping belching at 2.319-8.999 and door slamming at 3.957-5.957, 7.089-10.0", "frequencyCaption": "burping belching one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4624.wav", "onoffCaption": "train horn at 0.444-4.884, 5.959-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4671.wav", "onoffCaption": "door knocking at 3.288-6.125", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4694.wav", "onoffCaption": "thump thud at 2.224-4.724, 6.344-8.844", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4809.wav", "onoffCaption": "door slamming at 0.796-3.513, 5.446-6.424", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4848.wav", "onoffCaption": "spraying at 1.562-2.643, 3.178-4.259, 5.809-6.89", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4906.wav", "onoffCaption": "thump thud at 1.126-3.588 and cow mooing at 2.735-6.033", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4912.wav", "onoffCaption": "thump thud at 1.033-5.408 and explosion at 6.642-9.514", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4953.wav", "onoffCaption": "sneeze at 2.441-3.767 and dog barking at 2.683-4.683, 6.445-8.445", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2017.wav", "onoffCaption": "duck quacking at 2.591-4.591, 6.473-8.473", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2029.wav", "onoffCaption": "thump thud at 0.12-4.57 and dog barking at 0.898-2.898, 4.555-6.555", "frequencyCaption": "thump thud one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2058.wav", "onoffCaption": "door slamming at 0.54-1.713, 2.823-3.996, 5.038-6.211", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2099.wav", "onoffCaption": "spraying at 1.117-1.625", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2116.wav", "onoffCaption": "sneeze at 3.4-4.634, 5.74-6.974", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2132.wav", "onoffCaption": "tapping clicking clanking at 2.156-5.596", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2143.wav", "onoffCaption": "door slamming at 0.016-1.319, 2.87-4.173, 5.945-7.248 and spraying at 4.635-5.135, 7.255-7.882", "frequencyCaption": "door slamming three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2167.wav", "onoffCaption": "tapping clicking clanking at 0.555-3.995, 4.982-7.959 and car horn honking at 2.45-4.915 and dog barking at 3.612-5.612", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2182.wav", "onoffCaption": "tapping clicking clanking at 2.81-6.25", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2330.wav", "onoffCaption": "explosion at 0.7-3.572 and thump thud at 5.734-9.652", "frequencyCaption": "explosion one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2341.wav", "onoffCaption": "duck quacking at 0.59-2.59 and burping belching at 4.988-7.611", "frequencyCaption": "duck quacking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2365.wav", "onoffCaption": "tapping clicking clanking at 1.395-4.835 and door knocking at 3.354-5.731", "frequencyCaption": "tapping clicking clanking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2380.wav", "onoffCaption": "explosion at 3.921-6.915", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2441.wav", "onoffCaption": "sneeze at 0.431-1.523, 2.574-3.666 and gunshot at 0.854-2.854, 4.67-6.67", "frequencyCaption": "sneeze two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2515.wav", "onoffCaption": "dog barking at 3.882-5.882, 6.574-8.574", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2540.wav", "onoffCaption": "sheep goat bleating at 0.606-2.606, 5.052-7.052 and door slamming at 5.485-6.624, 8.528-9.781", "frequencyCaption": "sheep goat bleating two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2564.wav", "onoffCaption": "burping belching at 1.28-3.387, 4.428-6.535, 7.767-9.874", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2581.wav", "onoffCaption": "cat meowing at 0.248-1.388, 3.452-4.727, 6.932-8.926", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2628.wav", "onoffCaption": "thump thud at 0.371-4.038", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2698.wav", "onoffCaption": "explosion at 1.193-3.911, 5.087-7.427", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2717.wav", "onoffCaption": "door slamming at 0.151-2.151", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2733.wav", "onoffCaption": "sheep goat bleating at 2.91-4.91", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2766.wav", "onoffCaption": "woman laughing at 1.206-3.288, 3.961-6.868 and car horn honking at 2.938-6.113", "frequencyCaption": "woman laughing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2783.wav", "onoffCaption": "door knocking at 2.496-4.736, 6.515-8.755", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2811.wav", "onoffCaption": "whistling at 3.028-8.203", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2835.wav", "onoffCaption": "whistling at 0.039-9.704 and spraying at 0.105-1.362, 3.426-4.33", "frequencyCaption": "whistling one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2844.wav", "onoffCaption": "thump thud at 3.178-6.225, 7.297-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2860.wav", "onoffCaption": "whistling at 0.213-3.088, 3.926-6.389 and door knocking at 3.404-7.554", "frequencyCaption": "whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2885.wav", "onoffCaption": "duck quacking at 0.313-2.313, 4.547-6.547", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4100.wav", "onoffCaption": "tapping clicking clanking at 3.081-6.521", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4155.wav", "onoffCaption": "door slamming at 0.152-1.052, 1.604-2.584, 3.551-4.668 and woman laughing at 7.554-10.0", "frequencyCaption": "door slamming three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4171.wav", "onoffCaption": "car horn honking at 0.875-3.661", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4194.wav", "onoffCaption": "duck quacking at 0.045-2.045, 4.189-6.189 and woman laughing at 0.11-7.122", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4219.wav", "onoffCaption": "whistling at 3.867-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4268.wav", "onoffCaption": "train horn at 0.779-3.246, 4.852-6.993, 7.924-10.0 and car horn honking at 4.157-7.376", "frequencyCaption": "train horn three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4302.wav", "onoffCaption": "whistling at 1.87-9.525", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4326.wav", "onoffCaption": "sheep goat bleating at 0.284-4.204, 5.202-9.122 and tapping clicking clanking at 2.879-6.319", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4357.wav", "onoffCaption": "car horn honking at 0.415-2.762 and spraying at 1.964-2.591, 3.444-3.944, 5.15-6.407 and duck quacking at 2.399-4.399", "frequencyCaption": "car horn honking one times and spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4373.wav", "onoffCaption": "woman laughing at 2.689-9.423", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4396.wav", "onoffCaption": "woman laughing at 2.564-5.202, 6.278-8.561", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4418.wav", "onoffCaption": "sneeze at 0.199-1.363, 2.216-3.38, 3.88-5.044", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4503.wav", "onoffCaption": "cat meowing at 2.088-3.115 and whistling at 2.403-5.278, 6.471-9.346", "frequencyCaption": "cat meowing one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_4527.wav", "onoffCaption": "whistling at 1.299-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4556.wav", "onoffCaption": "sneeze at 0.693-2.193, 3.282-5.395, 6.632-9.449", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4701.wav", "onoffCaption": "cat meowing at 1.878-3.153", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4725.wav", "onoffCaption": "door slamming at 0.723-2.026 and whistling at 1.016-8.671 and thump thud at 4.221-8.671", "frequencyCaption": "door slamming one times and whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4754.wav", "onoffCaption": "car horn honking at 0.071-2.536, 4.312-6.777", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4770.wav", "onoffCaption": "cow mooing at 0.271-5.251, 6.816-9.785", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4795.wav", "onoffCaption": "sheep goat bleating at 1.536-3.536 and train horn at 7.424-10.0", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4807.wav", "onoffCaption": "whistling at 0.378-7.496 and duck quacking at 0.978-2.978", "frequencyCaption": "whistling one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4823.wav", "onoffCaption": "door knocking at 0.464-2.776, 3.809-6.112, 6.749-8.876", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4852.wav", "onoffCaption": "cat meowing at 0.345-2.235 and door knocking at 5.757-7.978", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4938.wav", "onoffCaption": "woman laughing at 3.72-6.774", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4949.wav", "onoffCaption": "cat meowing at 2.909-3.994 and dog barking at 6.602-8.602", "frequencyCaption": "cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2033.wav", "onoffCaption": "cow mooing at 3.185-7.614", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2061.wav", "onoffCaption": "door slamming at 1.789-4.668 and door knocking at 2.337-4.586, 6.731-9.083 and dog barking at 3.416-5.416", "frequencyCaption": "door slamming one times and door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2066.wav", "onoffCaption": "woman laughing at 0.892-3.247 and spraying at 8.202-9.283", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2083.wav", "onoffCaption": "spraying at 0.188-1.092", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2084.wav", "onoffCaption": "dog barking at 0.685-2.685 and tapping clicking clanking at 6.937-10.0", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2128.wav", "onoffCaption": "tapping clicking clanking at 1.229-4.669 and spraying at 6.867-7.471, 8.495-9.099", "frequencyCaption": "tapping clicking clanking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2198.wav", "onoffCaption": "dog barking at 0.7-2.7, 4.34-7.261 and gunshot at 2.054-4.073, 4.78-6.78, 7.354-9.354", "frequencyCaption": "dog barking two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2215.wav", "onoffCaption": "whistling at 3.911-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2240.wav", "onoffCaption": "cat meowing at 0.114-3.144, 4.645-6.585 and sheep goat bleating at 0.204-2.204", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2309.wav", "onoffCaption": "cow mooing at 1.54-5.969, 6.587-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2314.wav", "onoffCaption": "whistling at 0.043-2.052, 3.186-5.195, 6.106-8.115 and tapping clicking clanking at 1.55-4.99, 5.612-9.052", "frequencyCaption": "whistling three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2430.wav", "onoffCaption": "spraying at 3.021-3.872, 4.802-5.706 and cow mooing at 7.834-10.0", "frequencyCaption": "spraying two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2437.wav", "onoffCaption": "gunshot at 2.885-4.885 and door slamming at 7.153-8.093", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2465.wav", "onoffCaption": "cow mooing at 3.343-6.353", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2480.wav", "onoffCaption": "spraying at 1.693-2.544, 3.905-4.756", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2531.wav", "onoffCaption": "thump thud at 0.081-4.531, 6.844-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2616.wav", "onoffCaption": "whistling at 0.165-8.792 and door slamming at 0.317-1.168, 2.4-3.251, 4.02-4.871", "frequencyCaption": "whistling one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2643.wav", "onoffCaption": "gunshot at 0.789-2.789, 3.662-5.662 and train horn at 2.218-6.399, 7.69-10.0", "frequencyCaption": "gunshot two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2644.wav", "onoffCaption": "cat meowing at 0.168-1.195, 1.774-3.751, 4.349-6.416", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2659.wav", "onoffCaption": "whistling at 0.1-5.275", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2742.wav", "onoffCaption": "train horn at 2.253-5.653, 7.221-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2758.wav", "onoffCaption": "car horn honking at 0.059-3.234, 5.075-8.25", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2828.wav", "onoffCaption": "whistling at 0.57-8.32", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2933.wav", "onoffCaption": "explosion at 0.276-3.148, 3.917-6.789", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2934.wav", "onoffCaption": "burping belching at 0.338-7.315", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2961.wav", "onoffCaption": "spraying at 4.013-6.473, 8.516-9.266", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2984.wav", "onoffCaption": "sheep goat bleating at 0.618-2.618 and gunshot at 5.151-7.151, 7.731-9.731", "frequencyCaption": "sheep goat bleating one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4025.wav", "onoffCaption": "train horn at 3.629-6.096", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4070.wav", "onoffCaption": "door slamming at 0.029-1.32 and cow mooing at 2.727-6.025, 7.22-9.384 and sneeze at 4.803-7.279", "frequencyCaption": "door slamming one times and cow mooing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4095.wav", "onoffCaption": "thump thud at 3.766-6.537", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4124.wav", "onoffCaption": "duck quacking at 0.025-2.025", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4203.wav", "onoffCaption": "spraying at 0.356-2.816, 4.552-5.202", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4251.wav", "onoffCaption": "woman laughing at 0.47-3.062, 5.365-7.586", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4256.wav", "onoffCaption": "sheep goat bleating at 0.135-2.135, 3.296-5.296, 6.567-8.567", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4318.wav", "onoffCaption": "gunshot at 3.162-5.162, 5.706-7.706 and duck quacking at 3.962-5.962, 7.3-9.3", "frequencyCaption": "gunshot two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4426.wav", "onoffCaption": "door knocking at 2.595-4.783, 5.891-8.728", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4469.wav", "onoffCaption": "gunshot at 0.545-2.545 and woman laughing at 0.563-3.358, 4.196-6.991 and sheep goat bleating at 1.829-3.829", "frequencyCaption": "gunshot one times and woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4473.wav", "onoffCaption": "sneeze at 3.586-8.086", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4496.wav", "onoffCaption": "explosion at 2.265-5.265, 6.072-8.944", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4568.wav", "onoffCaption": "cat meowing at 0.568-1.653, 2.506-3.591 and car horn honking at 7.41-10.0", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4572.wav", "onoffCaption": "duck quacking at 0.41-2.41, 4.509-6.509 and gunshot at 1.916-3.916", "frequencyCaption": "duck quacking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4597.wav", "onoffCaption": "cow mooing at 1.045-4.027, 4.606-7.588", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4600.wav", "onoffCaption": "explosion at 0.464-4.464, 6.669-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4607.wav", "onoffCaption": "cat meowing at 2.122-3.739, 4.759-6.069, 7.198-9.192", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4655.wav", "onoffCaption": "train horn at 0.649-3.889, 6.249-9.489", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4839.wav", "onoffCaption": "door knocking at 3.565-6.94, 7.981-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4876.wav", "onoffCaption": "sneeze at 0.389-3.499 and whistling at 0.703-5.878", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4889.wav", "onoffCaption": "spraying at 2.837-3.412 and car horn honking at 6.908-10.0", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4893.wav", "onoffCaption": "door slamming at 3.845-4.864, 5.397-6.416 and duck quacking at 4.624-6.624", "frequencyCaption": "door slamming two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4922.wav", "onoffCaption": "door slamming at 0.069-0.75", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4977.wav", "onoffCaption": "duck quacking at 1.623-3.623, 4.63-6.63, 7.915-9.915", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4988.wav", "onoffCaption": "spraying at 2.075-2.807, 4.719-5.57", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4992.wav", "onoffCaption": "door knocking at 0.681-5.056 and gunshot at 3.1-5.1, 7.191-9.191", "frequencyCaption": "door knocking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2034.wav", "onoffCaption": "duck quacking at 0.504-2.504, 3.307-5.307", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2212.wav", "onoffCaption": "train horn at 2.927-6.697", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2247.wav", "onoffCaption": "dog barking at 0.081-2.081 and car horn honking at 0.117-4.366 and cat meowing at 6.32-7.904", "frequencyCaption": "dog barking one times and car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2346.wav", "onoffCaption": "door knocking at 0.207-3.582", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2462.wav", "onoffCaption": "tapping clicking clanking at 0.119-3.559, 4.558-6.763", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2487.wav", "onoffCaption": "burping belching at 0.648-7.328 and door knocking at 2.358-6.126, 7.457-9.846", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2563.wav", "onoffCaption": "door slamming at 0.181-1.472 and explosion at 3.64-6.168 and dog barking at 4.954-6.954", "frequencyCaption": "door slamming one times and explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2579.wav", "onoffCaption": "cat meowing at 0.093-2.087", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2611.wav", "onoffCaption": "tapping clicking clanking at 0.362-3.802, 6.013-9.453", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2710.wav", "onoffCaption": "duck quacking at 3.339-5.339, 7.228-9.228", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2898.wav", "onoffCaption": "car horn honking at 2.151-6.473, 6.993-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2966.wav", "onoffCaption": "train horn at 0.045-3.365, 4.328-7.648", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2983.wav", "onoffCaption": "gunshot at 0.554-2.554, 3.272-5.272 and train horn at 3.312-7.312", "frequencyCaption": "gunshot two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4022.wav", "onoffCaption": "car horn honking at 0.689-3.908", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4077.wav", "onoffCaption": "sheep goat bleating at 0.949-2.949, 3.937-5.937, 6.447-8.447", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4088.wav", "onoffCaption": "spraying at 2.992-5.452", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4092.wav", "onoffCaption": "explosion at 2.473-5.073", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4139.wav", "onoffCaption": "whistling at 0.518-8.173", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4176.wav", "onoffCaption": "car horn honking at 0.455-4.042", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4189.wav", "onoffCaption": "cat meowing at 0.138-1.498, 2.544-3.904, 5.026-6.386", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4193.wav", "onoffCaption": "woman laughing at 0.153-5.192 and cow mooing at 3.326-7.755", "frequencyCaption": "woman laughing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4204.wav", "onoffCaption": "train horn at 3.537-7.977 and duck quacking at 6.259-8.259", "frequencyCaption": "train horn one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4421.wav", "onoffCaption": "cat meowing at 0.141-1.288 and sheep goat bleating at 0.715-4.035 and door knocking at 6.083-8.583", "frequencyCaption": "cat meowing one times and sheep goat bleating one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4474.wav", "onoffCaption": "burping belching at 2.963-5.378, 6.525-8.94", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4491.wav", "onoffCaption": "tapping clicking clanking at 2.783-6.223, 7.925-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4520.wav", "onoffCaption": "dog barking at 1.345-4.665, 6.247-9.567", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4652.wav", "onoffCaption": "thump thud at 0.178-2.678, 4.247-6.747 and gunshot at 0.859-2.859, 3.774-5.774, 6.922-8.922", "frequencyCaption": "thump thud two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4749.wav", "onoffCaption": "duck quacking at 1.65-3.65, 5.955-7.955", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4824.wav", "onoffCaption": "woman laughing at 0.292-2.408 and tapping clicking clanking at 4.885-8.325", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4925.wav", "onoffCaption": "spraying at 0.314-1.314 and cow mooing at 5.093-8.075", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4970.wav", "onoffCaption": "gunshot at 1.78-3.78, 5.766-7.766", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4995.wav", "onoffCaption": "duck quacking at 1.467-3.467, 5.668-7.668", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_9.wav", "onoffCaption": "gunshot at 0.635-2.635", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2135.wav", "onoffCaption": "cat meowing at 4.338-5.423, 7.709-8.794", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2160.wav", "onoffCaption": "car horn honking at 0.225-5.132, 6.96-9.007 and burping belching at 2.253-4.776", "frequencyCaption": "car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2185.wav", "onoffCaption": "thump thud at 2.942-7.392", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2208.wav", "onoffCaption": "woman laughing at 1.114-8.559", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2279.wav", "onoffCaption": "thump thud at 1.323-5.773, 7.153-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2313.wav", "onoffCaption": "duck quacking at 1.898-3.898, 4.892-6.892 and sheep goat bleating at 4.573-6.573", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2478.wav", "onoffCaption": "dog barking at 0.087-2.087, 3.299-5.299, 6.614-8.614", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2536.wav", "onoffCaption": "cat meowing at 0.427-3.668, 5.917-9.158 and woman laughing at 0.858-3.627", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2586.wav", "onoffCaption": "dog barking at 1.358-3.358, 5.489-7.489", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2734.wav", "onoffCaption": "burping belching at 1.247-3.354 and spraying at 2.437-3.064, 4.131-5.215", "frequencyCaption": "burping belching one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2745.wav", "onoffCaption": "cow mooing at 0.875-3.857, 5.932-8.914", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2832.wav", "onoffCaption": "explosion at 0.339-2.341, 4.836-7.714 and car horn honking at 1.971-4.757", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2867.wav", "onoffCaption": "tapping clicking clanking at 0.598-4.038, 5.634-9.074", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2882.wav", "onoffCaption": "cow mooing at 2.804-7.784", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2929.wav", "onoffCaption": "tapping clicking clanking at 1.97-5.41 and sheep goat bleating at 2.461-4.461, 4.975-6.975", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2999.wav", "onoffCaption": "thump thud at 0.135-2.597, 3.618-6.08", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4038.wav", "onoffCaption": "cow mooing at 2.757-7.186", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4123.wav", "onoffCaption": "thump thud at 0.25-4.7, 5.564-8.064", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4305.wav", "onoffCaption": "whistling at 2.362-7.862", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4350.wav", "onoffCaption": "burping belching at 3.881-6.207, 6.742-9.742", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4575.wav", "onoffCaption": "explosion at 2.14-7.14, 7.84-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4590.wav", "onoffCaption": "thump thud at 1.677-4.724", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4648.wav", "onoffCaption": "gunshot at 3.838-6.078", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4706.wav", "onoffCaption": "explosion at 0.24-2.993, 4.501-7.501 and duck quacking at 2.746-4.746, 6.768-8.768", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4753.wav", "onoffCaption": "gunshot at 2.893-4.893", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4871.wav", "onoffCaption": "car horn honking at 1.632-3.979, 5.537-8.355", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4894.wav", "onoffCaption": "gunshot at 1.001-3.001", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2111.wav", "onoffCaption": "duck quacking at 0.295-2.295, 2.851-4.851 and woman laughing at 1.563-5.615, 7.148-10.0", "frequencyCaption": "duck quacking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2144.wav", "onoffCaption": "dog barking at 0.587-2.587, 3.375-5.375, 6.652-8.652", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2337.wav", "onoffCaption": "cow mooing at 0.134-5.114, 6.062-8.242", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2362.wav", "onoffCaption": "sheep goat bleating at 3.271-5.271, 7.111-9.111", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2387.wav", "onoffCaption": "sheep goat bleating at 2.984-4.984", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2409.wav", "onoffCaption": "cow mooing at 1.935-5.233, 6.077-9.059", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2413.wav", "onoffCaption": "gunshot at 0.733-2.733", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2512.wav", "onoffCaption": "whistling at 3.548-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2547.wav", "onoffCaption": "explosion at 0.219-2.221, 3.47-5.534 and sheep goat bleating at 6.765-8.765", "frequencyCaption": "explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2761.wav", "onoffCaption": "duck quacking at 2.498-4.498, 5.686-7.686 and explosion at 5.745-7.809", "frequencyCaption": "duck quacking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2784.wav", "onoffCaption": "thump thud at 1.867-4.206, 5.36-7.699", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2816.wav", "onoffCaption": "door slamming at 2.818-5.599 and cow mooing at 2.905-7.885", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2843.wav", "onoffCaption": "burping belching at 2.029-7.973", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2958.wav", "onoffCaption": "spraying at 0.929-3.321", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4049.wav", "onoffCaption": "woman laughing at 0.476-7.921 and car horn honking at 3.279-5.779", "frequencyCaption": "woman laughing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4107.wav", "onoffCaption": "cow mooing at 0.839-5.268, 6.236-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4152.wav", "onoffCaption": "burping belching at 1.002-3.625, 6.031-8.654", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4321.wav", "onoffCaption": "sneeze at 0.133-1.379, 2.353-3.965 and whistling at 1.027-7.317", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4374.wav", "onoffCaption": "explosion at 0.919-3.672, 5.793-8.522 and burping belching at 5.774-8.976", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4391.wav", "onoffCaption": "sheep goat bleating at 2.629-4.629", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4504.wav", "onoffCaption": "cat meowing at 0.188-1.374 and dog barking at 0.587-2.587, 3.86-5.86, 6.47-8.47", "frequencyCaption": "cat meowing one times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4551.wav", "onoffCaption": "woman laughing at 3.557-6.642", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4639.wav", "onoffCaption": "car horn honking at 1.638-4.813 and burping belching at 7.602-9.633", "frequencyCaption": "car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4689.wav", "onoffCaption": "sheep goat bleating at 1.268-3.268, 5.556-7.556", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4722.wav", "onoffCaption": "whistling at 1.885-7.385, 7.979-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4777.wav", "onoffCaption": "train horn at 1.37-5.81, 7.867-10.0 and gunshot at 2.589-4.589", "frequencyCaption": "train horn two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4792.wav", "onoffCaption": "woman laughing at 1.201-3.395, 5.62-7.814", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4800.wav", "onoffCaption": "train horn at 0.444-3.924 and sneeze at 2.515-4.198 and duck quacking at 6.469-8.469", "frequencyCaption": "train horn one times and sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4855.wav", "onoffCaption": "door slamming at 0.924-3.353 and tapping clicking clanking at 6.856-10.0", "frequencyCaption": "door slamming one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2010.wav", "onoffCaption": "thump thud at 0.321-2.783 and cat meowing at 4.756-9.756", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2045.wav", "onoffCaption": "door slamming at 0.243-2.469 and cow mooing at 4.409-7.707", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2236.wav", "onoffCaption": "thump thud at 1.714-3.942, 6.215-8.443", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2263.wav", "onoffCaption": "car horn honking at 2.479-6.801 and door slamming at 2.505-3.524, 5.637-6.656 and door knocking at 3.955-6.455", "frequencyCaption": "car horn honking one times and door slamming two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2286.wav", "onoffCaption": "sneeze at 0.115-1.349, 1.861-3.155, 3.884-6.727", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2378.wav", "onoffCaption": "duck quacking at 1.898-3.898, 4.706-6.706, 7.974-9.974 and car horn honking at 3.594-6.38", "frequencyCaption": "duck quacking three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2446.wav", "onoffCaption": "door knocking at 3.375-5.999, 7.052-9.129", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2508.wav", "onoffCaption": "woman laughing at 1.307-3.532, 4.108-6.316 and explosion at 3.427-6.427", "frequencyCaption": "woman laughing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2635.wav", "onoffCaption": "sheep goat bleating at 0.781-2.781", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2660.wav", "onoffCaption": "woman laughing at 3.713-7.765", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2685.wav", "onoffCaption": "sheep goat bleating at 0.99-2.99, 5.032-7.088", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2859.wav", "onoffCaption": "burping belching at 2.044-5.55, 7.897-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2917.wav", "onoffCaption": "tapping clicking clanking at 2.146-5.586, 6.162-9.602", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2942.wav", "onoffCaption": "duck quacking at 0.672-2.672, 5.106-7.106, 7.837-9.837", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4006.wav", "onoffCaption": "tapping clicking clanking at 2.901-6.341", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4053.wav", "onoffCaption": "cat meowing at 0.589-1.599, 2.313-3.34, 4.373-5.52", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4148.wav", "onoffCaption": "spraying at 1.484-3.217, 3.795-5.052, 5.923-7.007", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4220.wav", "onoffCaption": "cow mooing at 0.893-4.191", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4275.wav", "onoffCaption": "train horn at 0.043-3.443 and sheep goat bleating at 5.656-7.656", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4290.wav", "onoffCaption": "dog barking at 0.053-2.053 and sneeze at 1.421-3.008 and cat meowing at 1.901-3.087", "frequencyCaption": "dog barking one times and sneeze one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4405.wav", "onoffCaption": "explosion at 0.051-3.051, 3.851-6.574", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4450.wav", "onoffCaption": "sneeze at 2.386-3.62, 4.447-6.154, 7.403-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4623.wav", "onoffCaption": "cat meowing at 2.217-7.217", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4676.wav", "onoffCaption": "sheep goat bleating at 0.572-2.572", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4693.wav", "onoffCaption": "cow mooing at 2.486-5.455, 7.47-9.667", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4738.wav", "onoffCaption": "door slamming at 3.65-4.628, 6.613-8.613", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4788.wav", "onoffCaption": "explosion at 0.578-4.447, 5.213-9.082 and car horn honking at 0.746-3.564", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4901.wav", "onoffCaption": "car horn honking at 2.729-5.642, 6.454-8.801", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4954.wav", "onoffCaption": "cat meowing at 0.74-3.77", "frequencyCaption": "cat meowing one times"}
diff --git a/picoaudio/audioldm/__init__.py b/picoaudio/audioldm/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..2bbf85f01ccc72b6f18e7405d940adf07a26b500
--- /dev/null
+++ b/picoaudio/audioldm/__init__.py
@@ -0,0 +1,8 @@
+from .ldm import LatentDiffusion
+from .utils import seed_everything, save_wave, get_time, get_duration
+from .pipeline import *
+
+
+
+
+
diff --git a/picoaudio/audioldm/__main__.py b/picoaudio/audioldm/__main__.py
new file mode 100644
index 0000000000000000000000000000000000000000..13f8bafa839f512a156dd6380d2cf43c573a970a
--- /dev/null
+++ b/picoaudio/audioldm/__main__.py
@@ -0,0 +1,183 @@
+#!/usr/bin/python3
+import os
+from audioldm import text_to_audio, style_transfer, build_model, save_wave, get_time, round_up_duration, get_duration
+import argparse
+
+CACHE_DIR = os.getenv(
+    "AUDIOLDM_CACHE_DIR",
+    os.path.join(os.path.expanduser("~"), ".cache/audioldm"))
+
+parser = argparse.ArgumentParser()
+
+parser.add_argument(
+    "--mode",
+    type=str,
+    required=False,
+    default="generation",
+    help="generation: text-to-audio generation; transfer: style transfer",
+    choices=["generation", "transfer"]
+)
+
+parser.add_argument(
+    "-t",
+    "--text",
+    type=str,
+    required=False,
+    default="",
+    help="Text prompt to the model for audio generation",
+)
+
+parser.add_argument(
+    "-f",
+    "--file_path",
+    type=str,
+    required=False,
+    default=None,
+    help="(--mode transfer): Original audio file for style transfer; Or (--mode generation): the guidance audio file for generating simialr audio",
+)
+
+parser.add_argument(
+    "--transfer_strength",
+    type=float,
+    required=False,
+    default=0.5,
+    help="A value between 0 and 1. 0 means original audio without transfer, 1 means completely transfer to the audio indicated by text",
+)
+
+parser.add_argument(
+    "-s",
+    "--save_path",
+    type=str,
+    required=False,
+    help="The path to save model output",
+    default="./output",
+)
+
+parser.add_argument(
+    "--model_name",
+    type=str,
+    required=False,
+    help="The checkpoint you gonna use",
+    default="audioldm-s-full",
+    choices=["audioldm-s-full", "audioldm-l-full", "audioldm-s-full-v2"]
+)
+
+parser.add_argument(
+    "-ckpt",
+    "--ckpt_path",
+    type=str,
+    required=False,
+    help="The path to the pretrained .ckpt model",
+    default=None,
+)
+
+parser.add_argument(
+    "-b",
+    "--batchsize",
+    type=int,
+    required=False,
+    default=1,
+    help="Generate how many samples at the same time",
+)
+
+parser.add_argument(
+    "--ddim_steps",
+    type=int,
+    required=False,
+    default=200,
+    help="The sampling step for DDIM",
+)
+
+parser.add_argument(
+    "-gs",
+    "--guidance_scale",
+    type=float,
+    required=False,
+    default=2.5,
+    help="Guidance scale (Large => better quality and relavancy to text; Small => better diversity)",
+)
+
+parser.add_argument(
+    "-dur",
+    "--duration",
+    type=float,
+    required=False,
+    default=10.0,
+    help="The duration of the samples",
+)
+
+parser.add_argument(
+    "-n",
+    "--n_candidate_gen_per_text",
+    type=int,
+    required=False,
+    default=3,
+    help="Automatic quality control. This number control the number of candidates (e.g., generate three audios and choose the best to show you). A Larger value usually lead to better quality with heavier computation",
+)
+
+parser.add_argument(
+    "--seed",
+    type=int,
+    required=False,
+    default=42,
+    help="Change this value (any integer number) will lead to a different generation result.",
+)
+
+args = parser.parse_args()
+
+if(args.ckpt_path is not None):
+    print("Warning: ckpt_path has no effect after version 0.0.20.")
+    
+assert args.duration % 2.5 == 0, "Duration must be a multiple of 2.5"
+
+mode = args.mode
+if(mode == "generation" and args.file_path is not None):
+    mode = "generation_audio_to_audio"
+    if(len(args.text) > 0):
+        print("Warning: You have specified the --file_path. --text will be ignored")
+        args.text = ""
+        
+save_path = os.path.join(args.save_path, mode)
+
+if(args.file_path is not None):
+    save_path = os.path.join(save_path, os.path.basename(args.file_path.split(".")[0]))
+
+text = args.text
+random_seed = args.seed
+duration = args.duration
+guidance_scale = args.guidance_scale
+n_candidate_gen_per_text = args.n_candidate_gen_per_text
+
+os.makedirs(save_path, exist_ok=True)
+audioldm = build_model(model_name=args.model_name)
+
+if(args.mode == "generation"):
+    waveform = text_to_audio(
+        audioldm,
+        text,
+        args.file_path,
+        random_seed,
+        duration=duration,
+        guidance_scale=guidance_scale,
+        ddim_steps=args.ddim_steps,
+        n_candidate_gen_per_text=n_candidate_gen_per_text,
+        batchsize=args.batchsize,
+    )
+    
+elif(args.mode == "transfer"):
+    assert args.file_path is not None
+    assert os.path.exists(args.file_path), "The original audio file \'%s\' for style transfer does not exist." % args.file_path
+    waveform = style_transfer(
+        audioldm,
+        text,
+        args.file_path,
+        args.transfer_strength,
+        random_seed,
+        duration=duration,
+        guidance_scale=guidance_scale,
+        ddim_steps=args.ddim_steps,
+        batchsize=args.batchsize,
+    )
+    waveform = waveform[:,None,:]
+
+save_wave(waveform, save_path, name="%s_%s" % (get_time(), text))
diff --git a/picoaudio/audioldm/audio/__init__.py b/picoaudio/audioldm/audio/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..56902e96f041bc4ba6bfadd7a7742023b9560233
--- /dev/null
+++ b/picoaudio/audioldm/audio/__init__.py
@@ -0,0 +1,2 @@
+from .tools import wav_to_fbank, read_wav_file
+from .stft import TacotronSTFT
diff --git a/picoaudio/audioldm/audio/audio_processing.py b/picoaudio/audioldm/audio/audio_processing.py
new file mode 100644
index 0000000000000000000000000000000000000000..77a4057aa82f226f68474f4c2a19eba84510d663
--- /dev/null
+++ b/picoaudio/audioldm/audio/audio_processing.py
@@ -0,0 +1,100 @@
+import torch
+import numpy as np
+import librosa.util as librosa_util
+from scipy.signal import get_window
+
+
+def window_sumsquare(
+    window,
+    n_frames,
+    hop_length,
+    win_length,
+    n_fft,
+    dtype=np.float32,
+    norm=None,
+):
+    """
+    # from librosa 0.6
+    Compute the sum-square envelope of a window function at a given hop length.
+
+    This is used to estimate modulation effects induced by windowing
+    observations in short-time fourier transforms.
+
+    Parameters
+    ----------
+    window : string, tuple, number, callable, or list-like
+        Window specification, as in `get_window`
+
+    n_frames : int > 0
+        The number of analysis frames
+
+    hop_length : int > 0
+        The number of samples to advance between frames
+
+    win_length : [optional]
+        The length of the window function.  By default, this matches `n_fft`.
+
+    n_fft : int > 0
+        The length of each analysis frame.
+
+    dtype : np.dtype
+        The data type of the output
+
+    Returns
+    -------
+    wss : np.ndarray, shape=`(n_fft + hop_length * (n_frames - 1))`
+        The sum-squared envelope of the window function
+    """
+    if win_length is None:
+        win_length = n_fft
+
+    n = n_fft + hop_length * (n_frames - 1)
+    x = np.zeros(n, dtype=dtype)
+
+    # Compute the squared window at the desired length
+    win_sq = get_window(window, win_length, fftbins=True)
+    win_sq = librosa_util.normalize(win_sq, norm=norm) ** 2
+    win_sq = librosa_util.pad_center(win_sq, n_fft)
+
+    # Fill the envelope
+    for i in range(n_frames):
+        sample = i * hop_length
+        x[sample : min(n, sample + n_fft)] += win_sq[: max(0, min(n_fft, n - sample))]
+    return x
+
+
+def griffin_lim(magnitudes, stft_fn, n_iters=30):
+    """
+    PARAMS
+    ------
+    magnitudes: spectrogram magnitudes
+    stft_fn: STFT class with transform (STFT) and inverse (ISTFT) methods
+    """
+
+    angles = np.angle(np.exp(2j * np.pi * np.random.rand(*magnitudes.size())))
+    angles = angles.astype(np.float32)
+    angles = torch.autograd.Variable(torch.from_numpy(angles))
+    signal = stft_fn.inverse(magnitudes, angles).squeeze(1)
+
+    for i in range(n_iters):
+        _, angles = stft_fn.transform(signal)
+        signal = stft_fn.inverse(magnitudes, angles).squeeze(1)
+    return signal
+
+
+def dynamic_range_compression(x, normalize_fun=torch.log, C=1, clip_val=1e-5):
+    """
+    PARAMS
+    ------
+    C: compression factor
+    """
+    return normalize_fun(torch.clamp(x, min=clip_val) * C)
+
+
+def dynamic_range_decompression(x, C=1):
+    """
+    PARAMS
+    ------
+    C: compression factor used to compress
+    """
+    return torch.exp(x) / C
diff --git a/picoaudio/audioldm/audio/stft.py b/picoaudio/audioldm/audio/stft.py
new file mode 100644
index 0000000000000000000000000000000000000000..71d69793b3721a8179c1fb77afb427cfd869da40
--- /dev/null
+++ b/picoaudio/audioldm/audio/stft.py
@@ -0,0 +1,186 @@
+import torch
+import torch.nn.functional as F
+import numpy as np
+from scipy.signal import get_window
+from librosa.util import pad_center, tiny
+from librosa.filters import mel as librosa_mel_fn
+
+from audioldm.audio.audio_processing import (
+    dynamic_range_compression,
+    dynamic_range_decompression,
+    window_sumsquare,
+)
+
+
+class STFT(torch.nn.Module):
+    """adapted from Prem Seetharaman's https://github.com/pseeth/pytorch-stft"""
+
+    def __init__(self, filter_length, hop_length, win_length, window="hann"):
+        super(STFT, self).__init__()
+        self.filter_length = filter_length
+        self.hop_length = hop_length
+        self.win_length = win_length
+        self.window = window
+        self.forward_transform = None
+        scale = self.filter_length / self.hop_length
+        fourier_basis = np.fft.fft(np.eye(self.filter_length))
+
+        cutoff = int((self.filter_length / 2 + 1))
+        fourier_basis = np.vstack(
+            [np.real(fourier_basis[:cutoff, :]), np.imag(fourier_basis[:cutoff, :])]
+        )
+
+        forward_basis = torch.FloatTensor(fourier_basis[:, None, :])
+        inverse_basis = torch.FloatTensor(
+            np.linalg.pinv(scale * fourier_basis).T[:, None, :]
+        )
+
+        if window is not None:
+            assert filter_length >= win_length
+            # get window and zero center pad it to filter_length
+            fft_window = get_window(window, win_length, fftbins=True)
+            fft_window = pad_center(fft_window, size=filter_length)
+            fft_window = torch.from_numpy(fft_window).float()
+
+            # window the bases
+            forward_basis *= fft_window
+            inverse_basis *= fft_window
+
+        self.register_buffer("forward_basis", forward_basis.float())
+        self.register_buffer("inverse_basis", inverse_basis.float())
+
+    def transform(self, input_data):
+        device = self.forward_basis.device
+        input_data = input_data.to(device)
+        
+        num_batches = input_data.size(0)
+        num_samples = input_data.size(1)
+
+        self.num_samples = num_samples
+
+        # similar to librosa, reflect-pad the input
+        input_data = input_data.view(num_batches, 1, num_samples)
+        input_data = F.pad(
+            input_data.unsqueeze(1),
+            (int(self.filter_length / 2), int(self.filter_length / 2), 0, 0),
+            mode="reflect",
+        )
+        input_data = input_data.squeeze(1)
+
+        forward_transform = F.conv1d(
+            input_data,
+            torch.autograd.Variable(self.forward_basis, requires_grad=False),
+            stride=self.hop_length,
+            padding=0,
+        )#.cpu()
+
+        cutoff = int((self.filter_length / 2) + 1)
+        real_part = forward_transform[:, :cutoff, :]
+        imag_part = forward_transform[:, cutoff:, :]
+
+        magnitude = torch.sqrt(real_part**2 + imag_part**2)
+        phase = torch.autograd.Variable(torch.atan2(imag_part.data, real_part.data))
+
+        return magnitude, phase
+
+    def inverse(self, magnitude, phase):
+        device = self.forward_basis.device
+        magnitude, phase = magnitude.to(device), phase.to(device)
+        
+        recombine_magnitude_phase = torch.cat(
+            [magnitude * torch.cos(phase), magnitude * torch.sin(phase)], dim=1
+        )
+
+        inverse_transform = F.conv_transpose1d(
+            recombine_magnitude_phase,
+            torch.autograd.Variable(self.inverse_basis, requires_grad=False),
+            stride=self.hop_length,
+            padding=0,
+        )
+
+        if self.window is not None:
+            window_sum = window_sumsquare(
+                self.window,
+                magnitude.size(-1),
+                hop_length=self.hop_length,
+                win_length=self.win_length,
+                n_fft=self.filter_length,
+                dtype=np.float32,
+            )
+            # remove modulation effects
+            approx_nonzero_indices = torch.from_numpy(
+                np.where(window_sum > tiny(window_sum))[0]
+            )
+            window_sum = torch.autograd.Variable(
+                torch.from_numpy(window_sum), requires_grad=False
+            )
+            window_sum = window_sum
+            inverse_transform[:, :, approx_nonzero_indices] /= window_sum[
+                approx_nonzero_indices
+            ]
+
+            # scale by hop ratio
+            inverse_transform *= float(self.filter_length) / self.hop_length
+
+        inverse_transform = inverse_transform[:, :, int(self.filter_length / 2) :]
+        inverse_transform = inverse_transform[:, :, : -int(self.filter_length / 2) :]
+
+        return inverse_transform
+
+    def forward(self, input_data):
+        self.magnitude, self.phase = self.transform(input_data)
+        reconstruction = self.inverse(self.magnitude, self.phase)
+        return reconstruction
+
+
+class TacotronSTFT(torch.nn.Module):
+    def __init__(
+        self,
+        filter_length,
+        hop_length,
+        win_length,
+        n_mel_channels,
+        sampling_rate,
+        mel_fmin,
+        mel_fmax,
+    ):
+        super(TacotronSTFT, self).__init__()
+        self.n_mel_channels = n_mel_channels
+        self.sampling_rate = sampling_rate
+        self.stft_fn = STFT(filter_length, hop_length, win_length)
+        mel_basis = librosa_mel_fn(
+            sr=sampling_rate, n_fft=filter_length, n_mels=n_mel_channels, fmin=mel_fmin, fmax=mel_fmax
+        )
+        mel_basis = torch.from_numpy(mel_basis).float()
+        self.register_buffer("mel_basis", mel_basis)
+
+    def spectral_normalize(self, magnitudes, normalize_fun):
+        output = dynamic_range_compression(magnitudes, normalize_fun)
+        return output
+
+    def spectral_de_normalize(self, magnitudes):
+        output = dynamic_range_decompression(magnitudes)
+        return output
+
+    def mel_spectrogram(self, y, normalize_fun=torch.log):
+        """Computes mel-spectrograms from a batch of waves
+        PARAMS
+        ------
+        y: Variable(torch.FloatTensor) with shape (B, T) in range [-1, 1]
+
+        RETURNS
+        -------
+        mel_output: torch.FloatTensor of shape (B, n_mel_channels, T)
+        """
+        assert torch.min(y.data) >= -1, torch.min(y.data)
+        assert torch.max(y.data) <= 1, torch.max(y.data)
+
+        magnitudes, phases = self.stft_fn.transform(y)
+        magnitudes = magnitudes.data
+        mel_output = torch.matmul(self.mel_basis, magnitudes)
+        mel_output = self.spectral_normalize(mel_output, normalize_fun)
+        energy = torch.norm(magnitudes, dim=1)
+
+        log_magnitudes = self.spectral_normalize(magnitudes, normalize_fun)
+
+        return mel_output, log_magnitudes, energy
diff --git a/picoaudio/audioldm/audio/tools.py b/picoaudio/audioldm/audio/tools.py
new file mode 100644
index 0000000000000000000000000000000000000000..d641a982664b6673822c8528a1929c593f011b11
--- /dev/null
+++ b/picoaudio/audioldm/audio/tools.py
@@ -0,0 +1,85 @@
+import torch
+import numpy as np
+import torchaudio
+
+
+def get_mel_from_wav(audio, _stft):
+    audio = torch.clip(torch.FloatTensor(audio).unsqueeze(0), -1, 1)
+    audio = torch.autograd.Variable(audio, requires_grad=False)
+    melspec, log_magnitudes_stft, energy = _stft.mel_spectrogram(audio)
+    melspec = torch.squeeze(melspec, 0).numpy().astype(np.float32)
+    log_magnitudes_stft = (
+        torch.squeeze(log_magnitudes_stft, 0).numpy().astype(np.float32)
+    )
+    energy = torch.squeeze(energy, 0).numpy().astype(np.float32)
+    return melspec, log_magnitudes_stft, energy
+
+
+def _pad_spec(fbank, target_length=1024):
+    n_frames = fbank.shape[0]
+    p = target_length - n_frames
+    # cut and pad
+    if p > 0:
+        m = torch.nn.ZeroPad2d((0, 0, 0, p))
+        fbank = m(fbank)
+    elif p < 0:
+        fbank = fbank[0:target_length, :]
+
+    if fbank.size(-1) % 2 != 0:
+        fbank = fbank[..., :-1]
+
+    return fbank
+
+
+def pad_wav(waveform, segment_length):
+    waveform_length = waveform.shape[-1]
+    assert waveform_length > 100, "Waveform is too short, %s" % waveform_length
+    if segment_length is None or waveform_length == segment_length:
+        return waveform
+    elif waveform_length > segment_length:
+        return waveform[:segment_length]
+    elif waveform_length < segment_length:
+        temp_wav = np.zeros((1, segment_length))
+        temp_wav[:, :waveform_length] = waveform
+    return temp_wav
+
+def normalize_wav(waveform):
+    waveform = waveform - np.mean(waveform)
+    waveform = waveform / (np.max(np.abs(waveform)) + 1e-8)
+    return waveform * 0.5
+
+
+def read_wav_file(filename, segment_length):
+    # waveform, sr = librosa.load(filename, sr=None, mono=True) # 4 times slower
+    waveform, sr = torchaudio.load(filename)  # Faster!!!
+    waveform = torchaudio.functional.resample(waveform, orig_freq=sr, new_freq=16000)
+    waveform = waveform.numpy()[0, ...]
+    waveform = normalize_wav(waveform)
+    waveform = waveform[None, ...]
+    waveform = pad_wav(waveform, segment_length)
+    
+    waveform = waveform / np.max(np.abs(waveform))
+    waveform = 0.5 * waveform
+    
+    return waveform
+
+
+def wav_to_fbank(filename, target_length=1024, fn_STFT=None):
+    assert fn_STFT is not None
+
+    # mixup
+    waveform = read_wav_file(filename, target_length * 160)  # hop size is 160
+
+    waveform = waveform[0, ...]
+    waveform = torch.FloatTensor(waveform)
+
+    fbank, log_magnitudes_stft, energy = get_mel_from_wav(waveform, fn_STFT)
+
+    fbank = torch.FloatTensor(fbank.T)
+    log_magnitudes_stft = torch.FloatTensor(log_magnitudes_stft.T)
+
+    fbank, log_magnitudes_stft = _pad_spec(fbank, target_length), _pad_spec(
+        log_magnitudes_stft, target_length
+    )
+
+    return fbank, log_magnitudes_stft, waveform
diff --git a/picoaudio/audioldm/clap/__init__.py b/picoaudio/audioldm/clap/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/picoaudio/audioldm/clap/encoders.py b/picoaudio/audioldm/clap/encoders.py
new file mode 100644
index 0000000000000000000000000000000000000000..77d5e5c47c9dacf44406e9d00a831bfa051f4214
--- /dev/null
+++ b/picoaudio/audioldm/clap/encoders.py
@@ -0,0 +1,170 @@
+import torch
+import torch.nn as nn
+from audioldm.clap.open_clip import create_model
+from audioldm.clap.training.data import get_audio_features
+import torchaudio
+from transformers import RobertaTokenizer
+import torch.nn.functional as F
+
+
+class CLAPAudioEmbeddingClassifierFreev2(nn.Module):
+    def __init__(
+        self,
+        pretrained_path="",
+        key="class",
+        sampling_rate=16000,
+        embed_mode="audio",
+        amodel = "HTSAT-tiny",
+        unconditional_prob=0.1,
+        random_mute=False,
+        max_random_mute_portion=0.5,
+        training_mode=True,
+    ):
+        super().__init__()
+
+        self.key = key
+        self.device = "cpu"
+        self.precision = "fp32"
+        self.amodel = amodel  # or 'PANN-14'
+        self.tmodel = "roberta"  # the best text encoder in our training
+        self.enable_fusion = False  # False if you do not want to use the fusion model
+        self.fusion_type = "aff_2d"
+        self.pretrained = pretrained_path
+        self.embed_mode = embed_mode
+        self.embed_mode_orig = embed_mode
+        self.sampling_rate = sampling_rate
+        self.unconditional_prob = unconditional_prob
+        self.random_mute = random_mute
+        self.tokenize = RobertaTokenizer.from_pretrained("roberta-base")
+        self.max_random_mute_portion = max_random_mute_portion
+        self.training_mode = training_mode
+        self.model, self.model_cfg = create_model(
+            self.amodel,
+            self.tmodel,
+            self.pretrained,
+            precision=self.precision,
+            device=self.device,
+            enable_fusion=self.enable_fusion,
+            fusion_type=self.fusion_type,
+        )
+        for p in self.model.parameters():
+            p.requires_grad = False
+
+        self.model.eval()
+
+    def get_unconditional_condition(self, batchsize):
+        self.unconditional_token = self.model.get_text_embedding(
+            self.tokenizer(["", ""])
+        )[0:1]
+        return torch.cat([self.unconditional_token.unsqueeze(0)] * batchsize, dim=0)
+
+    def batch_to_list(self, batch):
+        ret = []
+        for i in range(batch.size(0)):
+            ret.append(batch[i])
+        return ret
+
+    def make_decision(self, probability):
+        if float(torch.rand(1)) < probability:
+            return True
+        else:
+            return False
+
+    def random_uniform(self, start, end):
+        val = torch.rand(1).item()
+        return start + (end - start) * val
+
+    def _random_mute(self, waveform):
+        # waveform: [bs, t-steps]
+        t_steps = waveform.size(-1)
+        for i in range(waveform.size(0)):
+            mute_size = int(
+                self.random_uniform(0, end=int(t_steps * self.max_random_mute_portion))
+            )
+            mute_start = int(self.random_uniform(0, t_steps - mute_size))
+            waveform[i, mute_start : mute_start + mute_size] = 0
+        return waveform
+
+    def cos_similarity(self, waveform, text):
+        # waveform: [bs, t_steps]
+        with torch.no_grad():
+            self.embed_mode = "audio"
+            audio_emb = self(waveform.cuda())
+            self.embed_mode = "text"
+            text_emb = self(text)
+            similarity = F.cosine_similarity(audio_emb, text_emb, dim=2), audio_emb, text_emb
+            return similarity.squeeze()
+
+    def forward(self, batch, key=None):
+        # If you want this conditioner to be unconditional, set self.unconditional_prob = 1.0
+        # If you want this conditioner to be fully conditional, set self.unconditional_prob = 0.0
+        if self.model.training == True and not self.training_mode:
+            print(
+                "The pretrained CLAP model should always be in eval mode. Reloading model just in case you change the parameters."
+            )
+            self.model, self.model_cfg = create_model(
+                self.amodel,
+                self.tmodel,
+                self.pretrained,
+                precision=self.precision,
+                device="cuda",
+                enable_fusion=self.enable_fusion,
+                fusion_type=self.fusion_type,
+            )
+            for p in self.model.parameters():
+                p.requires_grad = False
+            self.model.eval()
+
+        # the 'fusion' truncate mode can be changed to 'rand_trunc' if run in unfusion mode
+        if self.embed_mode == "audio":
+            with torch.no_grad():
+                audio_dict_list = []
+                assert (
+                    self.sampling_rate == 16000
+                ), "We only support 16000 sampling rate"
+                if self.random_mute:
+                    batch = self._random_mute(batch)
+                # batch: [bs, 1, t-samples]
+                batch = torchaudio.functional.resample(
+                    batch, orig_freq=self.sampling_rate, new_freq=48000
+                )
+                for waveform in self.batch_to_list(batch):
+                    audio_dict = {}
+                    audio_dict = get_audio_features(
+                        audio_dict,
+                        waveform,
+                        480000,
+                        data_truncating="fusion",
+                        data_filling="repeatpad",
+                        audio_cfg=self.model_cfg["audio_cfg"],
+                    )
+                    audio_dict_list.append(audio_dict)
+                # [bs, 512]
+                embed = self.model.get_audio_embedding(audio_dict_list)
+        elif self.embed_mode == "text":
+            with torch.no_grad():
+                # the 'fusion' truncate mode can be changed to 'rand_trunc' if run in unfusion mode
+                text_data = self.tokenizer(batch)
+                embed = self.model.get_text_embedding(text_data)
+
+        embed = embed.unsqueeze(1)
+        self.unconditional_token = self.model.get_text_embedding(
+            self.tokenizer(["", ""])
+        )[0:1]
+
+        for i in range(embed.size(0)):
+            if self.make_decision(self.unconditional_prob):
+                embed[i] = self.unconditional_token
+
+        # [bs, 1, 512]
+        return embed.detach()
+
+    def tokenizer(self, text):
+        result = self.tokenize(
+            text,
+            padding="max_length",
+            truncation=True,
+            max_length=512,
+            return_tensors="pt",
+        )
+        return {k: v.squeeze(0) for k, v in result.items()}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/__init__.py b/picoaudio/audioldm/clap/open_clip/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e9f728f2f273be5d5fdbec6c6cc41d737176a8c0
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/__init__.py
@@ -0,0 +1,25 @@
+from .factory import (
+    list_models,
+    create_model,
+    create_model_and_transforms,
+    add_model_config,
+)
+from .loss import ClipLoss, gather_features, LPLoss, lp_gather_features, LPMetrics
+from .model import (
+    CLAP,
+    CLAPTextCfg,
+    CLAPVisionCfg,
+    CLAPAudioCfp,
+    convert_weights_to_fp16,
+    trace_model,
+)
+from .openai import load_openai_model, list_openai_models
+from .pretrained import (
+    list_pretrained,
+    list_pretrained_tag_models,
+    list_pretrained_model_tags,
+    get_pretrained_url,
+    download_pretrained,
+)
+from .tokenizer import SimpleTokenizer, tokenize
+from .transform import image_transform
diff --git a/picoaudio/audioldm/clap/open_clip/bert.py b/picoaudio/audioldm/clap/open_clip/bert.py
new file mode 100644
index 0000000000000000000000000000000000000000..a83d96d2a77ed05198efc05837522bc88d2499cc
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/bert.py
@@ -0,0 +1,40 @@
+from transformers import BertTokenizer, BertModel
+
+tokenizer = BertTokenizer.from_pretrained("bert-base-uncased")
+model = BertModel.from_pretrained("bert-base-uncased")
+text = "Replace me by any text you'd like."
+
+
+def bert_embeddings(text):
+    # text = "Replace me by any text you'd like."
+    encoded_input = tokenizer(text, return_tensors="pt")
+    output = model(**encoded_input)
+    return output
+
+
+from transformers import RobertaTokenizer, RobertaModel
+
+tokenizer = RobertaTokenizer.from_pretrained("roberta-base")
+model = RobertaModel.from_pretrained("roberta-base")
+text = "Replace me by any text you'd like."
+
+
+def Roberta_embeddings(text):
+    # text = "Replace me by any text you'd like."
+    encoded_input = tokenizer(text, return_tensors="pt")
+    output = model(**encoded_input)
+    return output
+
+
+from transformers import BartTokenizer, BartModel
+
+tokenizer = BartTokenizer.from_pretrained("facebook/bart-base")
+model = BartModel.from_pretrained("facebook/bart-base")
+text = "Replace me by any text you'd like."
+
+
+def bart_embeddings(text):
+    # text = "Replace me by any text you'd like."
+    encoded_input = tokenizer(text, return_tensors="pt")
+    output = model(**encoded_input)
+    return output
diff --git a/picoaudio/audioldm/clap/open_clip/bpe_simple_vocab_16e6.txt.gz b/picoaudio/audioldm/clap/open_clip/bpe_simple_vocab_16e6.txt.gz
new file mode 100644
index 0000000000000000000000000000000000000000..36a15856e00a06a9fbed8cdd34d2393fea4a3113
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/bpe_simple_vocab_16e6.txt.gz
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:924691ac288e54409236115652ad4aa250f48203de50a9e4722a6ecd48d6804a
+size 1356917
diff --git a/picoaudio/audioldm/clap/open_clip/factory.py b/picoaudio/audioldm/clap/open_clip/factory.py
new file mode 100644
index 0000000000000000000000000000000000000000..64d5368bf4f14bd9592472de73d6162a93b16d73
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/factory.py
@@ -0,0 +1,279 @@
+import json
+import logging
+import os
+import pathlib
+import re
+from copy import deepcopy
+from pathlib import Path
+
+import torch
+
+from .model import CLAP, convert_weights_to_fp16
+from .openai import load_openai_model
+from .pretrained import get_pretrained_url, download_pretrained
+from .transform import image_transform
+
+_MODEL_CONFIG_PATHS = [Path(__file__).parent / f"model_configs/"]
+_MODEL_CONFIGS = {}  # directory (model_name: config) of model architecture configs
+CACHE_DIR = os.getenv("AUDIOLDM_CACHE_DIR", "~/.cache/audioldm")
+
+
+
+def _natural_key(string_):
+    return [int(s) if s.isdigit() else s for s in re.split(r"(\d+)", string_.lower())]
+
+
+def _rescan_model_configs():
+    global _MODEL_CONFIGS
+
+    config_ext = (".json",)
+    config_files = []
+    for config_path in _MODEL_CONFIG_PATHS:
+        if config_path.is_file() and config_path.suffix in config_ext:
+            config_files.append(config_path)
+        elif config_path.is_dir():
+            for ext in config_ext:
+                config_files.extend(config_path.glob(f"*{ext}"))
+
+    for cf in config_files:
+        if os.path.basename(cf)[0] == ".":
+            continue  # Ignore hidden files
+
+        with open(cf, "r") as f:
+            model_cfg = json.load(f)
+            if all(a in model_cfg for a in ("embed_dim", "audio_cfg", "text_cfg")):
+                _MODEL_CONFIGS[cf.stem] = model_cfg
+
+    _MODEL_CONFIGS = {
+        k: v
+        for k, v in sorted(_MODEL_CONFIGS.items(), key=lambda x: _natural_key(x[0]))
+    }
+
+
+_rescan_model_configs()  # initial populate of model config registry
+
+
+def load_state_dict(checkpoint_path: str, map_location="cpu", skip_params=True):
+    checkpoint = torch.load(checkpoint_path, map_location=map_location)
+    if isinstance(checkpoint, dict) and "state_dict" in checkpoint:
+        state_dict = checkpoint["state_dict"]
+    else:
+        state_dict = checkpoint
+    if skip_params:
+        if next(iter(state_dict.items()))[0].startswith("module"):
+            state_dict = {k[7:]: v for k, v in state_dict.items()}
+    # for k in state_dict:
+    #     if k.startswith('transformer'):
+    #         v = state_dict.pop(k)
+    #         state_dict['text_branch.' + k[12:]] = v
+    return state_dict
+
+
+def create_model(
+    amodel_name: str,
+    tmodel_name: str,
+    pretrained: str = "",
+    precision: str = "fp32",
+    device: torch.device = torch.device("cpu"),
+    jit: bool = False,
+    force_quick_gelu: bool = False,
+    openai_model_cache_dir: str = os.path.expanduser(f"{CACHE_DIR}/clip"),
+    skip_params=True,
+    pretrained_audio: str = "",
+    pretrained_text: str = "",
+    enable_fusion: bool = False,
+    fusion_type: str = "None"
+    # pretrained_image: bool = False,
+):
+    amodel_name = amodel_name.replace(
+        "/", "-"
+    )  # for callers using old naming with / in ViT names
+    pretrained_orig = pretrained
+    pretrained = pretrained.lower()
+    if pretrained == "openai":
+        if amodel_name in _MODEL_CONFIGS:
+            logging.info(f"Loading {amodel_name} model config.")
+            model_cfg = deepcopy(_MODEL_CONFIGS[amodel_name])
+        else:
+            logging.error(
+                f"Model config for {amodel_name} not found; available models {list_models()}."
+            )
+            raise RuntimeError(f"Model config for {amodel_name} not found.")
+
+        logging.info(f"Loading pretrained ViT-B-16 text encoder from OpenAI.")
+        # Hard Code in model name
+        model_cfg["text_cfg"]["model_type"] = tmodel_name
+        model = load_openai_model(
+            "ViT-B-16",
+            model_cfg,
+            device=device,
+            jit=jit,
+            cache_dir=openai_model_cache_dir,
+            enable_fusion=enable_fusion,
+            fusion_type=fusion_type,
+        )
+        # See https://discuss.pytorch.org/t/valueerror-attemting-to-unscale-fp16-gradients/81372
+        if precision == "amp" or precision == "fp32":
+            model = model.float()
+    else:
+        if amodel_name in _MODEL_CONFIGS:
+            logging.info(f"Loading {amodel_name} model config.")
+            model_cfg = deepcopy(_MODEL_CONFIGS[amodel_name])
+        else:
+            logging.error(
+                f"Model config for {amodel_name} not found; available models {list_models()}."
+            )
+            raise RuntimeError(f"Model config for {amodel_name} not found.")
+
+        if force_quick_gelu:
+            # override for use of QuickGELU on non-OpenAI transformer models
+            model_cfg["quick_gelu"] = True
+
+        # if pretrained_image:
+        #     if 'timm_amodel_name' in model_cfg.get('vision_cfg', {}):
+        #         # pretrained weight loading for timm models set via vision_cfg
+        #         model_cfg['vision_cfg']['timm_model_pretrained'] = True
+        #     else:
+        #         assert False, 'pretrained image towers currently only supported for timm models'
+        model_cfg["text_cfg"]["model_type"] = tmodel_name
+        model_cfg["enable_fusion"] = enable_fusion
+        model_cfg["fusion_type"] = fusion_type
+        model = CLAP(**model_cfg)
+
+        if pretrained:
+            checkpoint_path = ""
+            url = get_pretrained_url(amodel_name, pretrained)
+            if url:
+                checkpoint_path = download_pretrained(url, root=openai_model_cache_dir)
+            elif os.path.exists(pretrained_orig):
+                checkpoint_path = pretrained_orig
+            if checkpoint_path:
+                logging.info(
+                    f"Loading pretrained {amodel_name}-{tmodel_name} weights ({pretrained})."
+                )
+                ckpt = load_state_dict(checkpoint_path, skip_params=True)
+                model.load_state_dict(ckpt)
+                param_names = [n for n, p in model.named_parameters()]
+                # for n in param_names:
+                #     print(n, "\t", "Loaded" if n in ckpt else "Unloaded")
+            else:
+                logging.warning(
+                    f"Pretrained weights ({pretrained}) not found for model {amodel_name}."
+                )
+                raise RuntimeError(
+                    f"Pretrained weights ({pretrained}) not found for model {amodel_name}."
+                )
+
+        if pretrained_audio:
+            if amodel_name.startswith("PANN"):
+                if "Cnn14_mAP" in pretrained_audio:  # official checkpoint
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                    audio_ckpt = audio_ckpt["model"]
+                    keys = list(audio_ckpt.keys())
+                    for key in keys:
+                        if (
+                            "spectrogram_extractor" not in key
+                            and "logmel_extractor" not in key
+                        ):
+                            v = audio_ckpt.pop(key)
+                            audio_ckpt["audio_branch." + key] = v
+                elif os.path.basename(pretrained_audio).startswith(
+                    "PANN"
+                ):  # checkpoint trained via HTSAT codebase
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                    audio_ckpt = audio_ckpt["state_dict"]
+                    keys = list(audio_ckpt.keys())
+                    for key in keys:
+                        if key.startswith("sed_model"):
+                            v = audio_ckpt.pop(key)
+                            audio_ckpt["audio_branch." + key[10:]] = v
+                elif os.path.basename(pretrained_audio).startswith(
+                    "finetuned"
+                ):  # checkpoint trained via linear probe codebase
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                else:
+                    raise ValueError("Unknown audio checkpoint")
+            elif amodel_name.startswith("HTSAT"):
+                if "HTSAT_AudioSet_Saved" in pretrained_audio:  # official checkpoint
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                    audio_ckpt = audio_ckpt["state_dict"]
+                    keys = list(audio_ckpt.keys())
+                    for key in keys:
+                        if key.startswith("sed_model") and (
+                            "spectrogram_extractor" not in key
+                            and "logmel_extractor" not in key
+                        ):
+                            v = audio_ckpt.pop(key)
+                            audio_ckpt["audio_branch." + key[10:]] = v
+                elif os.path.basename(pretrained_audio).startswith(
+                    "HTSAT"
+                ):  # checkpoint trained via HTSAT codebase
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                    audio_ckpt = audio_ckpt["state_dict"]
+                    keys = list(audio_ckpt.keys())
+                    for key in keys:
+                        if key.startswith("sed_model"):
+                            v = audio_ckpt.pop(key)
+                            audio_ckpt["audio_branch." + key[10:]] = v
+                elif os.path.basename(pretrained_audio).startswith(
+                    "finetuned"
+                ):  # checkpoint trained via linear probe codebase
+                    audio_ckpt = torch.load(pretrained_audio, map_location="cpu")
+                else:
+                    raise ValueError("Unknown audio checkpoint")
+            else:
+                raise f"this audio encoder pretrained checkpoint is not support"
+
+            model.load_state_dict(audio_ckpt, strict=False)
+            logging.info(
+                f"Loading pretrained {amodel_name} weights ({pretrained_audio})."
+            )
+            param_names = [n for n, p in model.named_parameters()]
+            for n in param_names:
+                print(n, "\t", "Loaded" if n in audio_ckpt else "Unloaded")
+
+        model.to(device=device)
+        if precision == "fp16":
+            assert device.type != "cpu"
+            convert_weights_to_fp16(model)
+
+        if jit:
+            model = torch.jit.script(model)
+
+    return model, model_cfg
+
+
+def create_model_and_transforms(
+    model_name: str,
+    pretrained: str = "",
+    precision: str = "fp32",
+    device: torch.device = torch.device("cpu"),
+    jit: bool = False,
+    force_quick_gelu: bool = False,
+    # pretrained_image: bool = False,
+):
+    model = create_model(
+        model_name,
+        pretrained,
+        precision,
+        device,
+        jit,
+        force_quick_gelu=force_quick_gelu,
+        # pretrained_image=pretrained_image
+    )
+    preprocess_train = image_transform(model.visual.image_size, is_train=True)
+    preprocess_val = image_transform(model.visual.image_size, is_train=False)
+    return model, preprocess_train, preprocess_val
+
+
+def list_models():
+    """enumerate available model architectures based on config files"""
+    return list(_MODEL_CONFIGS.keys())
+
+
+def add_model_config(path):
+    """add model config path or file and update registry"""
+    if not isinstance(path, Path):
+        path = Path(path)
+    _MODEL_CONFIG_PATHS.append(path)
+    _rescan_model_configs()
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/feature_fusion.py b/picoaudio/audioldm/clap/open_clip/feature_fusion.py
new file mode 100644
index 0000000000000000000000000000000000000000..dbe4e170e05894c12ebdc36ba1dc1de65e441b89
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/feature_fusion.py
@@ -0,0 +1,192 @@
+"""
+Feature Fusion for Varible-Length Data Processing
+AFF/iAFF is referred and modified from https://github.com/YimianDai/open-aff/blob/master/aff_pytorch/aff_net/fusion.py
+According to the paper: Yimian Dai et al, Attentional Feature Fusion, IEEE Winter Conference on Applications of Computer Vision, WACV 2021
+"""
+
+import torch
+import torch.nn as nn
+
+
+class DAF(nn.Module):
+    """
+    直接相加 DirectAddFuse
+    """
+
+    def __init__(self):
+        super(DAF, self).__init__()
+
+    def forward(self, x, residual):
+        return x + residual
+
+
+class iAFF(nn.Module):
+    """
+    多特征融合 iAFF
+    """
+
+    def __init__(self, channels=64, r=4, type="2D"):
+        super(iAFF, self).__init__()
+        inter_channels = int(channels // r)
+
+        if type == "1D":
+            # 本地注意力
+            self.local_att = nn.Sequential(
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+
+            # 全局注意力
+            self.global_att = nn.Sequential(
+                nn.AdaptiveAvgPool1d(1),
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+
+            # 第二次本地注意力
+            self.local_att2 = nn.Sequential(
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+            # 第二次全局注意力
+            self.global_att2 = nn.Sequential(
+                nn.AdaptiveAvgPool1d(1),
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+        elif type == "2D":
+            # 本地注意力
+            self.local_att = nn.Sequential(
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+
+            # 全局注意力
+            self.global_att = nn.Sequential(
+                nn.AdaptiveAvgPool2d(1),
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+
+            # 第二次本地注意力
+            self.local_att2 = nn.Sequential(
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+            # 第二次全局注意力
+            self.global_att2 = nn.Sequential(
+                nn.AdaptiveAvgPool2d(1),
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+        else:
+            raise f"the type is not supported"
+
+        self.sigmoid = nn.Sigmoid()
+
+    def forward(self, x, residual):
+        flag = False
+        xa = x + residual
+        if xa.size(0) == 1:
+            xa = torch.cat([xa, xa], dim=0)
+            flag = True
+        xl = self.local_att(xa)
+        xg = self.global_att(xa)
+        xlg = xl + xg
+        wei = self.sigmoid(xlg)
+        xi = x * wei + residual * (1 - wei)
+
+        xl2 = self.local_att2(xi)
+        xg2 = self.global_att(xi)
+        xlg2 = xl2 + xg2
+        wei2 = self.sigmoid(xlg2)
+        xo = x * wei2 + residual * (1 - wei2)
+        if flag:
+            xo = xo[0].unsqueeze(0)
+        return xo
+
+
+class AFF(nn.Module):
+    """
+    多特征融合 AFF
+    """
+
+    def __init__(self, channels=64, r=4, type="2D"):
+        super(AFF, self).__init__()
+        inter_channels = int(channels // r)
+
+        if type == "1D":
+            self.local_att = nn.Sequential(
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+            self.global_att = nn.Sequential(
+                nn.AdaptiveAvgPool1d(1),
+                nn.Conv1d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv1d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm1d(channels),
+            )
+        elif type == "2D":
+            self.local_att = nn.Sequential(
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+            self.global_att = nn.Sequential(
+                nn.AdaptiveAvgPool2d(1),
+                nn.Conv2d(channels, inter_channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(inter_channels),
+                nn.ReLU(inplace=True),
+                nn.Conv2d(inter_channels, channels, kernel_size=1, stride=1, padding=0),
+                nn.BatchNorm2d(channels),
+            )
+        else:
+            raise f"the type is not supported."
+
+        self.sigmoid = nn.Sigmoid()
+
+    def forward(self, x, residual):
+        flag = False
+        xa = x + residual
+        if xa.size(0) == 1:
+            xa = torch.cat([xa, xa], dim=0)
+            flag = True
+        xl = self.local_att(xa)
+        xg = self.global_att(xa)
+        xlg = xl + xg
+        wei = self.sigmoid(xlg)
+        xo = 2 * x * wei + 2 * residual * (1 - wei)
+        if flag:
+            xo = xo[0].unsqueeze(0)
+        return xo
diff --git a/picoaudio/audioldm/clap/open_clip/htsat.py b/picoaudio/audioldm/clap/open_clip/htsat.py
new file mode 100644
index 0000000000000000000000000000000000000000..3b856c6a43df162116a941f1b5c76e93713b276a
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/htsat.py
@@ -0,0 +1,1308 @@
+# Ke Chen
+# knutchen@ucsd.edu
+# HTS-AT: A HIERARCHICAL TOKEN-SEMANTIC AUDIO TRANSFORMER FOR SOUND CLASSIFICATION AND DETECTION
+# Some layers designed on the model
+# below codes are based and referred from https://github.com/microsoft/Swin-Transformer
+# Swin Transformer for Computer Vision: https://arxiv.org/pdf/2103.14030.pdf
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from itertools import repeat
+import collections.abc
+import math
+import warnings
+
+from torch.nn.init import _calculate_fan_in_and_fan_out
+import torch.utils.checkpoint as checkpoint
+
+import random
+
+from torchlibrosa.stft import Spectrogram, LogmelFilterBank
+from torchlibrosa.augmentation import SpecAugmentation
+
+from itertools import repeat
+from .utils import do_mixup, interpolate
+
+from .feature_fusion import iAFF, AFF, DAF
+
+# from PyTorch internals
+def _ntuple(n):
+    def parse(x):
+        if isinstance(x, collections.abc.Iterable):
+            return x
+        return tuple(repeat(x, n))
+
+    return parse
+
+
+to_1tuple = _ntuple(1)
+to_2tuple = _ntuple(2)
+to_3tuple = _ntuple(3)
+to_4tuple = _ntuple(4)
+to_ntuple = _ntuple
+
+
+def drop_path(x, drop_prob: float = 0.0, training: bool = False):
+    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
+    This is the same as the DropConnect impl I created for EfficientNet, etc networks, however,
+    the original name is misleading as 'Drop Connect' is a different form of dropout in a separate paper...
+    See discussion: https://github.com/tensorflow/tpu/issues/494#issuecomment-532968956 ... I've opted for
+    changing the layer and argument names to 'drop path' rather than mix DropConnect as a layer name and use
+    'survival rate' as the argument.
+    """
+    if drop_prob == 0.0 or not training:
+        return x
+    keep_prob = 1 - drop_prob
+    shape = (x.shape[0],) + (1,) * (
+        x.ndim - 1
+    )  # work with diff dim tensors, not just 2D ConvNets
+    random_tensor = keep_prob + torch.rand(shape, dtype=x.dtype, device=x.device)
+    random_tensor.floor_()  # binarize
+    output = x.div(keep_prob) * random_tensor
+    return output
+
+
+class DropPath(nn.Module):
+    """Drop paths (Stochastic Depth) per sample  (when applied in main path of residual blocks)."""
+
+    def __init__(self, drop_prob=None):
+        super(DropPath, self).__init__()
+        self.drop_prob = drop_prob
+
+    def forward(self, x):
+        return drop_path(x, self.drop_prob, self.training)
+
+
+class PatchEmbed(nn.Module):
+    """2D Image to Patch Embedding"""
+
+    def __init__(
+        self,
+        img_size=224,
+        patch_size=16,
+        in_chans=3,
+        embed_dim=768,
+        norm_layer=None,
+        flatten=True,
+        patch_stride=16,
+        enable_fusion=False,
+        fusion_type="None",
+    ):
+        super().__init__()
+        img_size = to_2tuple(img_size)
+        patch_size = to_2tuple(patch_size)
+        patch_stride = to_2tuple(patch_stride)
+        self.img_size = img_size
+        self.patch_size = patch_size
+        self.patch_stride = patch_stride
+        self.grid_size = (
+            img_size[0] // patch_stride[0],
+            img_size[1] // patch_stride[1],
+        )
+        self.num_patches = self.grid_size[0] * self.grid_size[1]
+        self.flatten = flatten
+        self.in_chans = in_chans
+        self.embed_dim = embed_dim
+
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+
+        padding = (
+            (patch_size[0] - patch_stride[0]) // 2,
+            (patch_size[1] - patch_stride[1]) // 2,
+        )
+
+        if (self.enable_fusion) and (self.fusion_type == "channel_map"):
+            self.proj = nn.Conv2d(
+                in_chans * 4,
+                embed_dim,
+                kernel_size=patch_size,
+                stride=patch_stride,
+                padding=padding,
+            )
+        else:
+            self.proj = nn.Conv2d(
+                in_chans,
+                embed_dim,
+                kernel_size=patch_size,
+                stride=patch_stride,
+                padding=padding,
+            )
+        self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
+
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d"]
+        ):
+            self.mel_conv2d = nn.Conv2d(
+                in_chans,
+                embed_dim,
+                kernel_size=(patch_size[0], patch_size[1] * 3),
+                stride=(patch_stride[0], patch_stride[1] * 3),
+                padding=padding,
+            )
+            if self.fusion_type == "daf_2d":
+                self.fusion_model = DAF()
+            elif self.fusion_type == "aff_2d":
+                self.fusion_model = AFF(channels=embed_dim, type="2D")
+            elif self.fusion_type == "iaff_2d":
+                self.fusion_model = iAFF(channels=embed_dim, type="2D")
+
+    def forward(self, x, longer_idx=None):
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d"]
+        ):
+            global_x = x[:, 0:1, :, :]
+
+            # global processing
+            B, C, H, W = global_x.shape
+            assert (
+                H == self.img_size[0] and W == self.img_size[1]
+            ), f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."
+            global_x = self.proj(global_x)
+            TW = global_x.size(-1)
+            if len(longer_idx) > 0:
+                # local processing
+                local_x = x[longer_idx, 1:, :, :].contiguous()
+                B, C, H, W = local_x.shape
+                local_x = local_x.view(B * C, 1, H, W)
+                local_x = self.mel_conv2d(local_x)
+                local_x = local_x.view(
+                    B, C, local_x.size(1), local_x.size(2), local_x.size(3)
+                )
+                local_x = local_x.permute((0, 2, 3, 1, 4)).contiguous().flatten(3)
+                TB, TC, TH, _ = local_x.size()
+                if local_x.size(-1) < TW:
+                    local_x = torch.cat(
+                        [
+                            local_x,
+                            torch.zeros(
+                                (TB, TC, TH, TW - local_x.size(-1)),
+                                device=global_x.device,
+                            ),
+                        ],
+                        dim=-1,
+                    )
+                else:
+                    local_x = local_x[:, :, :, :TW]
+
+                global_x[longer_idx] = self.fusion_model(global_x[longer_idx], local_x)
+            x = global_x
+        else:
+            B, C, H, W = x.shape
+            assert (
+                H == self.img_size[0] and W == self.img_size[1]
+            ), f"Input image size ({H}*{W}) doesn't match model ({self.img_size[0]}*{self.img_size[1]})."
+            x = self.proj(x)
+
+        if self.flatten:
+            x = x.flatten(2).transpose(1, 2)  # BCHW -> BNC
+        x = self.norm(x)
+        return x
+
+
+class Mlp(nn.Module):
+    """MLP as used in Vision Transformer, MLP-Mixer and related networks"""
+
+    def __init__(
+        self,
+        in_features,
+        hidden_features=None,
+        out_features=None,
+        act_layer=nn.GELU,
+        drop=0.0,
+    ):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        self.fc1 = nn.Linear(in_features, hidden_features)
+        self.act = act_layer()
+        self.fc2 = nn.Linear(hidden_features, out_features)
+        self.drop = nn.Dropout(drop)
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.drop(x)
+        x = self.fc2(x)
+        x = self.drop(x)
+        return x
+
+
+def _no_grad_trunc_normal_(tensor, mean, std, a, b):
+    # Cut & paste from PyTorch official master until it's in a few official releases - RW
+    # Method based on https://people.sc.fsu.edu/~jburkardt/presentations/truncated_normal.pdf
+    def norm_cdf(x):
+        # Computes standard normal cumulative distribution function
+        return (1.0 + math.erf(x / math.sqrt(2.0))) / 2.0
+
+    if (mean < a - 2 * std) or (mean > b + 2 * std):
+        warnings.warn(
+            "mean is more than 2 std from [a, b] in nn.init.trunc_normal_. "
+            "The distribution of values may be incorrect.",
+            stacklevel=2,
+        )
+
+    with torch.no_grad():
+        # Values are generated by using a truncated uniform distribution and
+        # then using the inverse CDF for the normal distribution.
+        # Get upper and lower cdf values
+        l = norm_cdf((a - mean) / std)
+        u = norm_cdf((b - mean) / std)
+
+        # Uniformly fill tensor with values from [l, u], then translate to
+        # [2l-1, 2u-1].
+        tensor.uniform_(2 * l - 1, 2 * u - 1)
+
+        # Use inverse cdf transform for normal distribution to get truncated
+        # standard normal
+        tensor.erfinv_()
+
+        # Transform to proper mean, std
+        tensor.mul_(std * math.sqrt(2.0))
+        tensor.add_(mean)
+
+        # Clamp to ensure it's in the proper range
+        tensor.clamp_(min=a, max=b)
+        return tensor
+
+
+def trunc_normal_(tensor, mean=0.0, std=1.0, a=-2.0, b=2.0):
+    # type: (Tensor, float, float, float, float) -> Tensor
+    r"""Fills the input Tensor with values drawn from a truncated
+    normal distribution. The values are effectively drawn from the
+    normal distribution :math:`\mathcal{N}(\text{mean}, \text{std}^2)`
+    with values outside :math:`[a, b]` redrawn until they are within
+    the bounds. The method used for generating the random values works
+    best when :math:`a \leq \text{mean} \leq b`.
+    Args:
+        tensor: an n-dimensional `torch.Tensor`
+        mean: the mean of the normal distribution
+        std: the standard deviation of the normal distribution
+        a: the minimum cutoff value
+        b: the maximum cutoff value
+    Examples:
+        >>> w = torch.empty(3, 5)
+        >>> nn.init.trunc_normal_(w)
+    """
+    return _no_grad_trunc_normal_(tensor, mean, std, a, b)
+
+
+def variance_scaling_(tensor, scale=1.0, mode="fan_in", distribution="normal"):
+    fan_in, fan_out = _calculate_fan_in_and_fan_out(tensor)
+    if mode == "fan_in":
+        denom = fan_in
+    elif mode == "fan_out":
+        denom = fan_out
+    elif mode == "fan_avg":
+        denom = (fan_in + fan_out) / 2
+
+    variance = scale / denom
+
+    if distribution == "truncated_normal":
+        # constant is stddev of standard normal truncated to (-2, 2)
+        trunc_normal_(tensor, std=math.sqrt(variance) / 0.87962566103423978)
+    elif distribution == "normal":
+        tensor.normal_(std=math.sqrt(variance))
+    elif distribution == "uniform":
+        bound = math.sqrt(3 * variance)
+        tensor.uniform_(-bound, bound)
+    else:
+        raise ValueError(f"invalid distribution {distribution}")
+
+
+def lecun_normal_(tensor):
+    variance_scaling_(tensor, mode="fan_in", distribution="truncated_normal")
+
+
+def window_partition(x, window_size):
+    """
+    Args:
+        x: (B, H, W, C)
+        window_size (int): window size
+    Returns:
+        windows: (num_windows*B, window_size, window_size, C)
+    """
+    B, H, W, C = x.shape
+    x = x.view(B, H // window_size, window_size, W // window_size, window_size, C)
+    windows = (
+        x.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, window_size, window_size, C)
+    )
+    return windows
+
+
+def window_reverse(windows, window_size, H, W):
+    """
+    Args:
+        windows: (num_windows*B, window_size, window_size, C)
+        window_size (int): Window size
+        H (int): Height of image
+        W (int): Width of image
+    Returns:
+        x: (B, H, W, C)
+    """
+    B = int(windows.shape[0] / (H * W / window_size / window_size))
+    x = windows.view(
+        B, H // window_size, W // window_size, window_size, window_size, -1
+    )
+    x = x.permute(0, 1, 3, 2, 4, 5).contiguous().view(B, H, W, -1)
+    return x
+
+
+class WindowAttention(nn.Module):
+    r"""Window based multi-head self attention (W-MSA) module with relative position bias.
+    It supports both of shifted and non-shifted window.
+    Args:
+        dim (int): Number of input channels.
+        window_size (tuple[int]): The height and width of the window.
+        num_heads (int): Number of attention heads.
+        qkv_bias (bool, optional):  If True, add a learnable bias to query, key, value. Default: True
+        qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set
+        attn_drop (float, optional): Dropout ratio of attention weight. Default: 0.0
+        proj_drop (float, optional): Dropout ratio of output. Default: 0.0
+    """
+
+    def __init__(
+        self,
+        dim,
+        window_size,
+        num_heads,
+        qkv_bias=True,
+        qk_scale=None,
+        attn_drop=0.0,
+        proj_drop=0.0,
+    ):
+
+        super().__init__()
+        self.dim = dim
+        self.window_size = window_size  # Wh, Ww
+        self.num_heads = num_heads
+        head_dim = dim // num_heads
+        self.scale = qk_scale or head_dim**-0.5
+
+        # define a parameter table of relative position bias
+        self.relative_position_bias_table = nn.Parameter(
+            torch.zeros((2 * window_size[0] - 1) * (2 * window_size[1] - 1), num_heads)
+        )  # 2*Wh-1 * 2*Ww-1, nH
+
+        # get pair-wise relative position index for each token inside the window
+        coords_h = torch.arange(self.window_size[0])
+        coords_w = torch.arange(self.window_size[1])
+        coords = torch.stack(torch.meshgrid([coords_h, coords_w]))  # 2, Wh, Ww
+        coords_flatten = torch.flatten(coords, 1)  # 2, Wh*Ww
+        relative_coords = (
+            coords_flatten[:, :, None] - coords_flatten[:, None, :]
+        )  # 2, Wh*Ww, Wh*Ww
+        relative_coords = relative_coords.permute(
+            1, 2, 0
+        ).contiguous()  # Wh*Ww, Wh*Ww, 2
+        relative_coords[:, :, 0] += self.window_size[0] - 1  # shift to start from 0
+        relative_coords[:, :, 1] += self.window_size[1] - 1
+        relative_coords[:, :, 0] *= 2 * self.window_size[1] - 1
+        relative_position_index = relative_coords.sum(-1)  # Wh*Ww, Wh*Ww
+        self.register_buffer("relative_position_index", relative_position_index)
+
+        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+
+        trunc_normal_(self.relative_position_bias_table, std=0.02)
+        self.softmax = nn.Softmax(dim=-1)
+
+    def forward(self, x, mask=None):
+        """
+        Args:
+            x: input features with shape of (num_windows*B, N, C)
+            mask: (0/-inf) mask with shape of (num_windows, Wh*Ww, Wh*Ww) or None
+        """
+        B_, N, C = x.shape
+        qkv = (
+            self.qkv(x)
+            .reshape(B_, N, 3, self.num_heads, C // self.num_heads)
+            .permute(2, 0, 3, 1, 4)
+        )
+        q, k, v = (
+            qkv[0],
+            qkv[1],
+            qkv[2],
+        )  # make torchscript happy (cannot use tensor as tuple)
+
+        q = q * self.scale
+        attn = q @ k.transpose(-2, -1)
+
+        relative_position_bias = self.relative_position_bias_table[
+            self.relative_position_index.view(-1)
+        ].view(
+            self.window_size[0] * self.window_size[1],
+            self.window_size[0] * self.window_size[1],
+            -1,
+        )  # Wh*Ww,Wh*Ww,nH
+        relative_position_bias = relative_position_bias.permute(
+            2, 0, 1
+        ).contiguous()  # nH, Wh*Ww, Wh*Ww
+        attn = attn + relative_position_bias.unsqueeze(0)
+
+        if mask is not None:
+            nW = mask.shape[0]
+            attn = attn.view(B_ // nW, nW, self.num_heads, N, N) + mask.unsqueeze(
+                1
+            ).unsqueeze(0)
+            attn = attn.view(-1, self.num_heads, N, N)
+            attn = self.softmax(attn)
+        else:
+            attn = self.softmax(attn)
+
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v).transpose(1, 2).reshape(B_, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x, attn
+
+    def extra_repr(self):
+        return f"dim={self.dim}, window_size={self.window_size}, num_heads={self.num_heads}"
+
+
+# We use the model based on Swintransformer Block, therefore we can use the swin-transformer pretrained model
+class SwinTransformerBlock(nn.Module):
+    r"""Swin Transformer Block.
+    Args:
+        dim (int): Number of input channels.
+        input_resolution (tuple[int]): Input resulotion.
+        num_heads (int): Number of attention heads.
+        window_size (int): Window size.
+        shift_size (int): Shift size for SW-MSA.
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+        qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.
+        drop (float, optional): Dropout rate. Default: 0.0
+        attn_drop (float, optional): Attention dropout rate. Default: 0.0
+        drop_path (float, optional): Stochastic depth rate. Default: 0.0
+        act_layer (nn.Module, optional): Activation layer. Default: nn.GELU
+        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
+    """
+
+    def __init__(
+        self,
+        dim,
+        input_resolution,
+        num_heads,
+        window_size=7,
+        shift_size=0,
+        mlp_ratio=4.0,
+        qkv_bias=True,
+        qk_scale=None,
+        drop=0.0,
+        attn_drop=0.0,
+        drop_path=0.0,
+        act_layer=nn.GELU,
+        norm_layer=nn.LayerNorm,
+        norm_before_mlp="ln",
+    ):
+        super().__init__()
+        self.dim = dim
+        self.input_resolution = input_resolution
+        self.num_heads = num_heads
+        self.window_size = window_size
+        self.shift_size = shift_size
+        self.mlp_ratio = mlp_ratio
+        self.norm_before_mlp = norm_before_mlp
+        if min(self.input_resolution) <= self.window_size:
+            # if window size is larger than input resolution, we don't partition windows
+            self.shift_size = 0
+            self.window_size = min(self.input_resolution)
+        assert (
+            0 <= self.shift_size < self.window_size
+        ), "shift_size must in 0-window_size"
+
+        self.norm1 = norm_layer(dim)
+        self.attn = WindowAttention(
+            dim,
+            window_size=to_2tuple(self.window_size),
+            num_heads=num_heads,
+            qkv_bias=qkv_bias,
+            qk_scale=qk_scale,
+            attn_drop=attn_drop,
+            proj_drop=drop,
+        )
+
+        self.drop_path = DropPath(drop_path) if drop_path > 0.0 else nn.Identity()
+        if self.norm_before_mlp == "ln":
+            self.norm2 = nn.LayerNorm(dim)
+        elif self.norm_before_mlp == "bn":
+            self.norm2 = lambda x: nn.BatchNorm1d(dim)(x.transpose(1, 2)).transpose(
+                1, 2
+            )
+        else:
+            raise NotImplementedError
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp(
+            in_features=dim,
+            hidden_features=mlp_hidden_dim,
+            act_layer=act_layer,
+            drop=drop,
+        )
+
+        if self.shift_size > 0:
+            # calculate attention mask for SW-MSA
+            H, W = self.input_resolution
+            img_mask = torch.zeros((1, H, W, 1))  # 1 H W 1
+            h_slices = (
+                slice(0, -self.window_size),
+                slice(-self.window_size, -self.shift_size),
+                slice(-self.shift_size, None),
+            )
+            w_slices = (
+                slice(0, -self.window_size),
+                slice(-self.window_size, -self.shift_size),
+                slice(-self.shift_size, None),
+            )
+            cnt = 0
+            for h in h_slices:
+                for w in w_slices:
+                    img_mask[:, h, w, :] = cnt
+                    cnt += 1
+
+            mask_windows = window_partition(
+                img_mask, self.window_size
+            )  # nW, window_size, window_size, 1
+            mask_windows = mask_windows.view(-1, self.window_size * self.window_size)
+            attn_mask = mask_windows.unsqueeze(1) - mask_windows.unsqueeze(2)
+            attn_mask = attn_mask.masked_fill(
+                attn_mask != 0, float(-100.0)
+            ).masked_fill(attn_mask == 0, float(0.0))
+        else:
+            attn_mask = None
+
+        self.register_buffer("attn_mask", attn_mask)
+
+    def forward(self, x):
+        # pdb.set_trace()
+        H, W = self.input_resolution
+        # print("H: ", H)
+        # print("W: ", W)
+        # pdb.set_trace()
+        B, L, C = x.shape
+        # assert L == H * W, "input feature has wrong size"
+
+        shortcut = x
+        x = self.norm1(x)
+        x = x.view(B, H, W, C)
+
+        # cyclic shift
+        if self.shift_size > 0:
+            shifted_x = torch.roll(
+                x, shifts=(-self.shift_size, -self.shift_size), dims=(1, 2)
+            )
+        else:
+            shifted_x = x
+
+        # partition windows
+        x_windows = window_partition(
+            shifted_x, self.window_size
+        )  # nW*B, window_size, window_size, C
+        x_windows = x_windows.view(
+            -1, self.window_size * self.window_size, C
+        )  # nW*B, window_size*window_size, C
+
+        # W-MSA/SW-MSA
+        attn_windows, attn = self.attn(
+            x_windows, mask=self.attn_mask
+        )  # nW*B, window_size*window_size, C
+
+        # merge windows
+        attn_windows = attn_windows.view(-1, self.window_size, self.window_size, C)
+        shifted_x = window_reverse(attn_windows, self.window_size, H, W)  # B H' W' C
+
+        # reverse cyclic shift
+        if self.shift_size > 0:
+            x = torch.roll(
+                shifted_x, shifts=(self.shift_size, self.shift_size), dims=(1, 2)
+            )
+        else:
+            x = shifted_x
+        x = x.view(B, H * W, C)
+
+        # FFN
+        x = shortcut + self.drop_path(x)
+        x = x + self.drop_path(self.mlp(self.norm2(x)))
+
+        return x, attn
+
+    def extra_repr(self):
+        return (
+            f"dim={self.dim}, input_resolution={self.input_resolution}, num_heads={self.num_heads}, "
+            f"window_size={self.window_size}, shift_size={self.shift_size}, mlp_ratio={self.mlp_ratio}"
+        )
+
+
+class PatchMerging(nn.Module):
+    r"""Patch Merging Layer.
+    Args:
+        input_resolution (tuple[int]): Resolution of input feature.
+        dim (int): Number of input channels.
+        norm_layer (nn.Module, optional): Normalization layer.  Default: nn.LayerNorm
+    """
+
+    def __init__(self, input_resolution, dim, norm_layer=nn.LayerNorm):
+        super().__init__()
+        self.input_resolution = input_resolution
+        self.dim = dim
+        self.reduction = nn.Linear(4 * dim, 2 * dim, bias=False)
+        self.norm = norm_layer(4 * dim)
+
+    def forward(self, x):
+        """
+        x: B, H*W, C
+        """
+        H, W = self.input_resolution
+        B, L, C = x.shape
+        assert L == H * W, "input feature has wrong size"
+        assert H % 2 == 0 and W % 2 == 0, f"x size ({H}*{W}) are not even."
+
+        x = x.view(B, H, W, C)
+
+        x0 = x[:, 0::2, 0::2, :]  # B H/2 W/2 C
+        x1 = x[:, 1::2, 0::2, :]  # B H/2 W/2 C
+        x2 = x[:, 0::2, 1::2, :]  # B H/2 W/2 C
+        x3 = x[:, 1::2, 1::2, :]  # B H/2 W/2 C
+        x = torch.cat([x0, x1, x2, x3], -1)  # B H/2 W/2 4*C
+        x = x.view(B, -1, 4 * C)  # B H/2*W/2 4*C
+
+        x = self.norm(x)
+        x = self.reduction(x)
+
+        return x
+
+    def extra_repr(self):
+        return f"input_resolution={self.input_resolution}, dim={self.dim}"
+
+
+class BasicLayer(nn.Module):
+    """A basic Swin Transformer layer for one stage.
+    Args:
+        dim (int): Number of input channels.
+        input_resolution (tuple[int]): Input resolution.
+        depth (int): Number of blocks.
+        num_heads (int): Number of attention heads.
+        window_size (int): Local window size.
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim.
+        qkv_bias (bool, optional): If True, add a learnable bias to query, key, value. Default: True
+        qk_scale (float | None, optional): Override default qk scale of head_dim ** -0.5 if set.
+        drop (float, optional): Dropout rate. Default: 0.0
+        attn_drop (float, optional): Attention dropout rate. Default: 0.0
+        drop_path (float | tuple[float], optional): Stochastic depth rate. Default: 0.0
+        norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm
+        downsample (nn.Module | None, optional): Downsample layer at the end of the layer. Default: None
+        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False.
+    """
+
+    def __init__(
+        self,
+        dim,
+        input_resolution,
+        depth,
+        num_heads,
+        window_size,
+        mlp_ratio=4.0,
+        qkv_bias=True,
+        qk_scale=None,
+        drop=0.0,
+        attn_drop=0.0,
+        drop_path=0.0,
+        norm_layer=nn.LayerNorm,
+        downsample=None,
+        use_checkpoint=False,
+        norm_before_mlp="ln",
+    ):
+
+        super().__init__()
+        self.dim = dim
+        self.input_resolution = input_resolution
+        self.depth = depth
+        self.use_checkpoint = use_checkpoint
+
+        # build blocks
+        self.blocks = nn.ModuleList(
+            [
+                SwinTransformerBlock(
+                    dim=dim,
+                    input_resolution=input_resolution,
+                    num_heads=num_heads,
+                    window_size=window_size,
+                    shift_size=0 if (i % 2 == 0) else window_size // 2,
+                    mlp_ratio=mlp_ratio,
+                    qkv_bias=qkv_bias,
+                    qk_scale=qk_scale,
+                    drop=drop,
+                    attn_drop=attn_drop,
+                    drop_path=drop_path[i]
+                    if isinstance(drop_path, list)
+                    else drop_path,
+                    norm_layer=norm_layer,
+                    norm_before_mlp=norm_before_mlp,
+                )
+                for i in range(depth)
+            ]
+        )
+
+        # patch merging layer
+        if downsample is not None:
+            self.downsample = downsample(
+                input_resolution, dim=dim, norm_layer=norm_layer
+            )
+        else:
+            self.downsample = None
+
+    def forward(self, x):
+        attns = []
+        for blk in self.blocks:
+            if self.use_checkpoint:
+                x = checkpoint.checkpoint(blk, x)
+            else:
+                x, attn = blk(x)
+                if not self.training:
+                    attns.append(attn.unsqueeze(0))
+        if self.downsample is not None:
+            x = self.downsample(x)
+        if not self.training:
+            attn = torch.cat(attns, dim=0)
+            attn = torch.mean(attn, dim=0)
+        return x, attn
+
+    def extra_repr(self):
+        return f"dim={self.dim}, input_resolution={self.input_resolution}, depth={self.depth}"
+
+
+# The Core of HTSAT
+class HTSAT_Swin_Transformer(nn.Module):
+    r"""HTSAT based on the Swin Transformer
+    Args:
+        spec_size (int | tuple(int)): Input Spectrogram size. Default 256
+        patch_size (int | tuple(int)): Patch size. Default: 4
+        path_stride (iot | tuple(int)): Patch Stride for Frequency and Time Axis. Default: 4
+        in_chans (int): Number of input image channels. Default: 1 (mono)
+        num_classes (int): Number of classes for classification head. Default: 527
+        embed_dim (int): Patch embedding dimension. Default: 96
+        depths (tuple(int)): Depth of each HTSAT-Swin Transformer layer.
+        num_heads (tuple(int)): Number of attention heads in different layers.
+        window_size (int): Window size. Default: 8
+        mlp_ratio (float): Ratio of mlp hidden dim to embedding dim. Default: 4
+        qkv_bias (bool): If True, add a learnable bias to query, key, value. Default: True
+        qk_scale (float): Override default qk scale of head_dim ** -0.5 if set. Default: None
+        drop_rate (float): Dropout rate. Default: 0
+        attn_drop_rate (float): Attention dropout rate. Default: 0
+        drop_path_rate (float): Stochastic depth rate. Default: 0.1
+        norm_layer (nn.Module): Normalization layer. Default: nn.LayerNorm.
+        ape (bool): If True, add absolute position embedding to the patch embedding. Default: False
+        patch_norm (bool): If True, add normalization after patch embedding. Default: True
+        use_checkpoint (bool): Whether to use checkpointing to save memory. Default: False
+        config (module): The configuration Module from config.py
+    """
+
+    def __init__(
+        self,
+        spec_size=256,
+        patch_size=4,
+        patch_stride=(4, 4),
+        in_chans=1,
+        num_classes=527,
+        embed_dim=96,
+        depths=[2, 2, 6, 2],
+        num_heads=[4, 8, 16, 32],
+        window_size=8,
+        mlp_ratio=4.0,
+        qkv_bias=True,
+        qk_scale=None,
+        drop_rate=0.0,
+        attn_drop_rate=0.0,
+        drop_path_rate=0.1,
+        norm_layer=nn.LayerNorm,
+        ape=False,
+        patch_norm=True,
+        use_checkpoint=False,
+        norm_before_mlp="ln",
+        config=None,
+        enable_fusion=False,
+        fusion_type="None",
+        **kwargs,
+    ):
+        super(HTSAT_Swin_Transformer, self).__init__()
+
+        self.config = config
+        self.spec_size = spec_size
+        self.patch_stride = patch_stride
+        self.patch_size = patch_size
+        self.window_size = window_size
+        self.embed_dim = embed_dim
+        self.depths = depths
+        self.ape = ape
+        self.in_chans = in_chans
+        self.num_classes = num_classes
+        self.num_heads = num_heads
+        self.num_layers = len(self.depths)
+        self.num_features = int(self.embed_dim * 2 ** (self.num_layers - 1))
+
+        self.drop_rate = drop_rate
+        self.attn_drop_rate = attn_drop_rate
+        self.drop_path_rate = drop_path_rate
+
+        self.qkv_bias = qkv_bias
+        self.qk_scale = None
+
+        self.patch_norm = patch_norm
+        self.norm_layer = norm_layer if self.patch_norm else None
+        self.norm_before_mlp = norm_before_mlp
+        self.mlp_ratio = mlp_ratio
+
+        self.use_checkpoint = use_checkpoint
+
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+
+        #  process mel-spec ; used only once
+        self.freq_ratio = self.spec_size // self.config.mel_bins
+        window = "hann"
+        center = True
+        pad_mode = "reflect"
+        ref = 1.0
+        amin = 1e-10
+        top_db = None
+        self.interpolate_ratio = 32  # Downsampled ratio
+        # Spectrogram extractor
+        self.spectrogram_extractor = Spectrogram(
+            n_fft=config.window_size,
+            hop_length=config.hop_size,
+            win_length=config.window_size,
+            window=window,
+            center=center,
+            pad_mode=pad_mode,
+            freeze_parameters=True,
+        )
+        # Logmel feature extractor
+        self.logmel_extractor = LogmelFilterBank(
+            sr=config.sample_rate,
+            n_fft=config.window_size,
+            n_mels=config.mel_bins,
+            fmin=config.fmin,
+            fmax=config.fmax,
+            ref=ref,
+            amin=amin,
+            top_db=top_db,
+            freeze_parameters=True,
+        )
+        # Spec augmenter
+        self.spec_augmenter = SpecAugmentation(
+            time_drop_width=64,
+            time_stripes_num=2,
+            freq_drop_width=8,
+            freq_stripes_num=2,
+        )  # 2 2
+        self.bn0 = nn.BatchNorm2d(self.config.mel_bins)
+
+        # split spctrogram into non-overlapping patches
+        self.patch_embed = PatchEmbed(
+            img_size=self.spec_size,
+            patch_size=self.patch_size,
+            in_chans=self.in_chans,
+            embed_dim=self.embed_dim,
+            norm_layer=self.norm_layer,
+            patch_stride=patch_stride,
+            enable_fusion=self.enable_fusion,
+            fusion_type=self.fusion_type,
+        )
+
+        num_patches = self.patch_embed.num_patches
+        patches_resolution = self.patch_embed.grid_size
+        self.patches_resolution = patches_resolution
+
+        # absolute position embedding
+        if self.ape:
+            self.absolute_pos_embed = nn.Parameter(
+                torch.zeros(1, num_patches, self.embed_dim)
+            )
+            trunc_normal_(self.absolute_pos_embed, std=0.02)
+
+        self.pos_drop = nn.Dropout(p=self.drop_rate)
+
+        # stochastic depth
+        dpr = [
+            x.item() for x in torch.linspace(0, self.drop_path_rate, sum(self.depths))
+        ]  # stochastic depth decay rule
+
+        # build layers
+        self.layers = nn.ModuleList()
+        for i_layer in range(self.num_layers):
+            layer = BasicLayer(
+                dim=int(self.embed_dim * 2**i_layer),
+                input_resolution=(
+                    patches_resolution[0] // (2**i_layer),
+                    patches_resolution[1] // (2**i_layer),
+                ),
+                depth=self.depths[i_layer],
+                num_heads=self.num_heads[i_layer],
+                window_size=self.window_size,
+                mlp_ratio=self.mlp_ratio,
+                qkv_bias=self.qkv_bias,
+                qk_scale=self.qk_scale,
+                drop=self.drop_rate,
+                attn_drop=self.attn_drop_rate,
+                drop_path=dpr[
+                    sum(self.depths[:i_layer]) : sum(self.depths[: i_layer + 1])
+                ],
+                norm_layer=self.norm_layer,
+                downsample=PatchMerging if (i_layer < self.num_layers - 1) else None,
+                use_checkpoint=use_checkpoint,
+                norm_before_mlp=self.norm_before_mlp,
+            )
+            self.layers.append(layer)
+
+        self.norm = self.norm_layer(self.num_features)
+        self.avgpool = nn.AdaptiveAvgPool1d(1)
+        self.maxpool = nn.AdaptiveMaxPool1d(1)
+
+        SF = (
+            self.spec_size
+            // (2 ** (len(self.depths) - 1))
+            // self.patch_stride[0]
+            // self.freq_ratio
+        )
+        self.tscam_conv = nn.Conv2d(
+            in_channels=self.num_features,
+            out_channels=self.num_classes,
+            kernel_size=(SF, 3),
+            padding=(0, 1),
+        )
+        self.head = nn.Linear(num_classes, num_classes)
+
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_1d", "aff_1d", "iaff_1d"]
+        ):
+            self.mel_conv1d = nn.Sequential(
+                nn.Conv1d(64, 64, kernel_size=5, stride=3, padding=2),
+                nn.BatchNorm1d(64),
+            )
+            if self.fusion_type == "daf_1d":
+                self.fusion_model = DAF()
+            elif self.fusion_type == "aff_1d":
+                self.fusion_model = AFF(channels=64, type="1D")
+            elif self.fusion_type == "iaff_1d":
+                self.fusion_model = iAFF(channels=64, type="1D")
+
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            trunc_normal_(m.weight, std=0.02)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+
+    @torch.jit.ignore
+    def no_weight_decay(self):
+        return {"absolute_pos_embed"}
+
+    @torch.jit.ignore
+    def no_weight_decay_keywords(self):
+        return {"relative_position_bias_table"}
+
+    def forward_features(self, x, longer_idx=None):
+        # A deprecated optimization for using a hierarchical output from different blocks
+
+        frames_num = x.shape[2]
+        x = self.patch_embed(x, longer_idx=longer_idx)
+        if self.ape:
+            x = x + self.absolute_pos_embed
+        x = self.pos_drop(x)
+        for i, layer in enumerate(self.layers):
+            x, attn = layer(x)
+        # for x
+        x = self.norm(x)
+        B, N, C = x.shape
+        SF = frames_num // (2 ** (len(self.depths) - 1)) // self.patch_stride[0]
+        ST = frames_num // (2 ** (len(self.depths) - 1)) // self.patch_stride[1]
+        x = x.permute(0, 2, 1).contiguous().reshape(B, C, SF, ST)
+        B, C, F, T = x.shape
+        # group 2D CNN
+        c_freq_bin = F // self.freq_ratio
+        x = x.reshape(B, C, F // c_freq_bin, c_freq_bin, T)
+        x = x.permute(0, 1, 3, 2, 4).contiguous().reshape(B, C, c_freq_bin, -1)
+        # get latent_output
+        fine_grained_latent_output = torch.mean(x, dim=2)
+        fine_grained_latent_output = interpolate(
+            fine_grained_latent_output.permute(0, 2, 1).contiguous(),
+            8 * self.patch_stride[1],
+        )
+
+        latent_output = self.avgpool(torch.flatten(x, 2))
+        latent_output = torch.flatten(latent_output, 1)
+
+        # display the attention map, if needed
+
+        x = self.tscam_conv(x)
+        x = torch.flatten(x, 2)  # B, C, T
+
+        fpx = interpolate(
+            torch.sigmoid(x).permute(0, 2, 1).contiguous(), 8 * self.patch_stride[1]
+        )
+
+        x = self.avgpool(x)
+        x = torch.flatten(x, 1)
+
+        output_dict = {
+            "framewise_output": fpx,  # already sigmoided
+            "clipwise_output": torch.sigmoid(x),
+            "fine_grained_embedding": fine_grained_latent_output,
+            "embedding": latent_output,
+        }
+
+        return output_dict
+
+    def crop_wav(self, x, crop_size, spe_pos=None):
+        time_steps = x.shape[2]
+        tx = torch.zeros(x.shape[0], x.shape[1], crop_size, x.shape[3]).to(x.device)
+        for i in range(len(x)):
+            if spe_pos is None:
+                crop_pos = random.randint(0, time_steps - crop_size - 1)
+            else:
+                crop_pos = spe_pos
+            tx[i][0] = x[i, 0, crop_pos : crop_pos + crop_size, :]
+        return tx
+
+    # Reshape the wavform to a img size, if you want to use the pretrained swin transformer model
+    def reshape_wav2img(self, x):
+        B, C, T, F = x.shape
+        target_T = int(self.spec_size * self.freq_ratio)
+        target_F = self.spec_size // self.freq_ratio
+        assert (
+            T <= target_T and F <= target_F
+        ), "the wav size should less than or equal to the swin input size"
+        # to avoid bicubic zero error
+        if T < target_T:
+            x = nn.functional.interpolate(
+                x, (target_T, x.shape[3]), mode="bicubic", align_corners=True
+            )
+        if F < target_F:
+            x = nn.functional.interpolate(
+                x, (x.shape[2], target_F), mode="bicubic", align_corners=True
+            )
+        x = x.permute(0, 1, 3, 2).contiguous()
+        x = x.reshape(
+            x.shape[0],
+            x.shape[1],
+            x.shape[2],
+            self.freq_ratio,
+            x.shape[3] // self.freq_ratio,
+        )
+        # print(x.shape)
+        x = x.permute(0, 1, 3, 2, 4).contiguous()
+        x = x.reshape(x.shape[0], x.shape[1], x.shape[2] * x.shape[3], x.shape[4])
+        return x
+
+    # Repeat the wavform to a img size, if you want to use the pretrained swin transformer model
+    def repeat_wat2img(self, x, cur_pos):
+        B, C, T, F = x.shape
+        target_T = int(self.spec_size * self.freq_ratio)
+        target_F = self.spec_size // self.freq_ratio
+        assert (
+            T <= target_T and F <= target_F
+        ), "the wav size should less than or equal to the swin input size"
+        # to avoid bicubic zero error
+        if T < target_T:
+            x = nn.functional.interpolate(
+                x, (target_T, x.shape[3]), mode="bicubic", align_corners=True
+            )
+        if F < target_F:
+            x = nn.functional.interpolate(
+                x, (x.shape[2], target_F), mode="bicubic", align_corners=True
+            )
+        x = x.permute(0, 1, 3, 2).contiguous()  # B C F T
+        x = x[:, :, :, cur_pos : cur_pos + self.spec_size]
+        x = x.repeat(repeats=(1, 1, 4, 1))
+        return x
+
+    def forward(
+        self, x: torch.Tensor, mixup_lambda=None, infer_mode=False, device=None
+    ):  # out_feat_keys: List[str] = None):
+
+        if self.enable_fusion and x["longer"].sum() == 0:
+            # if no audio is longer than 10s, then randomly select one audio to be longer
+            x["longer"][torch.randint(0, x["longer"].shape[0], (1,))] = True
+
+        if not self.enable_fusion:
+            x = x["waveform"].to(device=device, non_blocking=True)
+            x = self.spectrogram_extractor(x)  # (batch_size, 1, time_steps, freq_bins)
+            x = self.logmel_extractor(x)  # (batch_size, 1, time_steps, mel_bins)
+            x = x.transpose(1, 3)
+            x = self.bn0(x)
+            x = x.transpose(1, 3)
+            if self.training:
+                x = self.spec_augmenter(x)
+
+            if self.training and mixup_lambda is not None:
+                x = do_mixup(x, mixup_lambda)
+
+            x = self.reshape_wav2img(x)
+            output_dict = self.forward_features(x)
+        else:
+            longer_list = x["longer"].to(device=device, non_blocking=True)
+            x = x["mel_fusion"].to(device=device, non_blocking=True)
+            x = x.transpose(1, 3)
+            x = self.bn0(x)
+            x = x.transpose(1, 3)
+            longer_list_idx = torch.where(longer_list)[0]
+            if self.fusion_type in ["daf_1d", "aff_1d", "iaff_1d"]:
+                new_x = x[:, 0:1, :, :].clone().contiguous()
+                if len(longer_list_idx) > 0:
+                    # local processing
+                    fusion_x_local = x[longer_list_idx, 1:, :, :].clone().contiguous()
+                    FB, FC, FT, FF = fusion_x_local.size()
+                    fusion_x_local = fusion_x_local.view(FB * FC, FT, FF)
+                    fusion_x_local = torch.permute(
+                        fusion_x_local, (0, 2, 1)
+                    ).contiguous()
+                    fusion_x_local = self.mel_conv1d(fusion_x_local)
+                    fusion_x_local = fusion_x_local.view(
+                        FB, FC, FF, fusion_x_local.size(-1)
+                    )
+                    fusion_x_local = (
+                        torch.permute(fusion_x_local, (0, 2, 1, 3))
+                        .contiguous()
+                        .flatten(2)
+                    )
+                    if fusion_x_local.size(-1) < FT:
+                        fusion_x_local = torch.cat(
+                            [
+                                fusion_x_local,
+                                torch.zeros(
+                                    (FB, FF, FT - fusion_x_local.size(-1)),
+                                    device=device,
+                                ),
+                            ],
+                            dim=-1,
+                        )
+                    else:
+                        fusion_x_local = fusion_x_local[:, :, :FT]
+                    # 1D fusion
+                    new_x = new_x.squeeze(1).permute((0, 2, 1)).contiguous()
+                    new_x[longer_list_idx] = self.fusion_model(
+                        new_x[longer_list_idx], fusion_x_local
+                    )
+                    x = new_x.permute((0, 2, 1)).contiguous()[:, None, :, :]
+                else:
+                    x = new_x
+
+            elif self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d", "channel_map"]:
+                x = x  # no change
+
+            if self.training:
+                x = self.spec_augmenter(x)
+            if self.training and mixup_lambda is not None:
+                x = do_mixup(x, mixup_lambda)
+
+            x = self.reshape_wav2img(x)
+            output_dict = self.forward_features(x, longer_idx=longer_list_idx)
+
+        # if infer_mode:
+        #     # in infer mode. we need to handle different length audio input
+        #     frame_num = x.shape[2]
+        #     target_T = int(self.spec_size * self.freq_ratio)
+        #     repeat_ratio = math.floor(target_T / frame_num)
+        #     x = x.repeat(repeats=(1,1,repeat_ratio,1))
+        #     x = self.reshape_wav2img(x)
+        #     output_dict = self.forward_features(x)
+        # else:
+        #     if x.shape[2] > self.freq_ratio * self.spec_size:
+        #         if self.training:
+        #             x = self.crop_wav(x, crop_size=self.freq_ratio * self.spec_size)
+        #             x = self.reshape_wav2img(x)
+        #             output_dict = self.forward_features(x)
+        #         else:
+        #             # Change: Hard code here
+        #             overlap_size = (x.shape[2] - 1) // 4
+        #             output_dicts = []
+        #             crop_size = (x.shape[2] - 1) // 2
+        #             for cur_pos in range(0, x.shape[2] - crop_size - 1, overlap_size):
+        #                 tx = self.crop_wav(x, crop_size = crop_size, spe_pos = cur_pos)
+        #                 tx = self.reshape_wav2img(tx)
+        #                 output_dicts.append(self.forward_features(tx))
+        #             clipwise_output = torch.zeros_like(output_dicts[0]["clipwise_output"]).float().to(x.device)
+        #             framewise_output = torch.zeros_like(output_dicts[0]["framewise_output"]).float().to(x.device)
+        #             for d in output_dicts:
+        #                 clipwise_output += d["clipwise_output"]
+        #                 framewise_output += d["framewise_output"]
+        #             clipwise_output  = clipwise_output / len(output_dicts)
+        #             framewise_output = framewise_output / len(output_dicts)
+        #             output_dict = {
+        #                 'framewise_output': framewise_output,
+        #                 'clipwise_output': clipwise_output
+        #             }
+        #     else: # this part is typically used, and most easy one
+        #         x = self.reshape_wav2img(x)
+        #         output_dict = self.forward_features(x)
+        # x = self.head(x)
+
+        # We process the data in the dataloader part, in that here we only consider the input_T < fixed_T
+
+        return output_dict
+
+
+def create_htsat_model(audio_cfg, enable_fusion=False, fusion_type="None"):
+    try:
+
+        assert audio_cfg.model_name in [
+            "tiny",
+            "base",
+            "large",
+        ], "model name for HTS-AT is wrong!"
+        if audio_cfg.model_name == "tiny":
+            model = HTSAT_Swin_Transformer(
+                spec_size=256,
+                patch_size=4,
+                patch_stride=(4, 4),
+                num_classes=audio_cfg.class_num,
+                embed_dim=96,
+                depths=[2, 2, 6, 2],
+                num_heads=[4, 8, 16, 32],
+                window_size=8,
+                config=audio_cfg,
+                enable_fusion=enable_fusion,
+                fusion_type=fusion_type,
+            )
+        elif audio_cfg.model_name == "base":
+            model = HTSAT_Swin_Transformer(
+                spec_size=256,
+                patch_size=4,
+                patch_stride=(4, 4),
+                num_classes=audio_cfg.class_num,
+                embed_dim=128,
+                depths=[2, 2, 12, 2],
+                num_heads=[4, 8, 16, 32],
+                window_size=8,
+                config=audio_cfg,
+                enable_fusion=enable_fusion,
+                fusion_type=fusion_type,
+            )
+        elif audio_cfg.model_name == "large":
+            model = HTSAT_Swin_Transformer(
+                spec_size=256,
+                patch_size=4,
+                patch_stride=(4, 4),
+                num_classes=audio_cfg.class_num,
+                embed_dim=256,
+                depths=[2, 2, 12, 2],
+                num_heads=[4, 8, 16, 32],
+                window_size=8,
+                config=audio_cfg,
+                enable_fusion=enable_fusion,
+                fusion_type=fusion_type,
+            )
+
+        return model
+    except:
+        raise RuntimeError(
+            f"Import Model for {audio_cfg.model_name} not found, or the audio cfg parameters are not enough."
+        )
diff --git a/picoaudio/audioldm/clap/open_clip/linear_probe.py b/picoaudio/audioldm/clap/open_clip/linear_probe.py
new file mode 100644
index 0000000000000000000000000000000000000000..9d7e23b6b67a53e16d050d675a99d01d7d04d581
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/linear_probe.py
@@ -0,0 +1,66 @@
+import numpy as np
+import torch.nn.functional as F
+from torch import nn
+from .model import MLPLayers
+
+
+class LinearProbe(nn.Module):
+    def __init__(self, model, mlp, freeze, in_ch, out_ch, act=None):
+        """
+        Args:
+            model: nn.Module
+            mlp: bool, if True, then use the MLP layer as the linear probe module
+            freeze: bool, if Ture, then freeze all the CLAP model's layers when training the linear probe
+            in_ch: int, the output channel from CLAP model
+            out_ch: int, the output channel from linear probe (class_num)
+            act: torch.nn.functional, the activation function before the loss function
+        """
+        super().__init__()
+        in_ch = 512
+        self.clap_model = model
+        self.clap_model.text_branch = None  # to save memory
+        self.freeze = freeze
+        if mlp:
+            self.lp_layer = MLPLayers(units=[in_ch, in_ch * 2, out_ch])
+        else:
+            self.lp_layer = nn.Linear(in_ch, out_ch)
+
+        if self.freeze:
+            for param in self.clap_model.parameters():
+                param.requires_grad = False
+
+        if act == "None":
+            self.act = None
+        elif act == "relu":
+            self.act = nn.ReLU()
+        elif act == "elu":
+            self.act = nn.ELU()
+        elif act == "prelu":
+            self.act = nn.PReLU(num_parameters=in_ch)
+        elif act == "softmax":
+            self.act = nn.Softmax(dim=-1)
+        elif act == "sigmoid":
+            self.act = nn.Sigmoid()
+
+    def forward(self, x, mix_lambda=None, device=None):
+        """
+        Args:
+            x: waveform, torch.tensor [batch, t_samples] / batch of mel_spec and longer list
+            mix_lambda: torch.tensor [batch], the mixup lambda
+        Returns:
+            class_prob: torch.tensor [batch, class_num]
+
+        """
+        # batchnorm cancel grandient
+        if self.freeze:
+            self.clap_model.eval()
+
+        x = self.clap_model.audio_projection(
+            self.clap_model.audio_branch(x, mixup_lambda=mix_lambda, device=device)[
+                "embedding"
+            ]
+        )
+        out = self.lp_layer(x)
+        if self.act is not None:
+            out = self.act(out)
+        return out
diff --git a/picoaudio/audioldm/clap/open_clip/loss.py b/picoaudio/audioldm/clap/open_clip/loss.py
new file mode 100644
index 0000000000000000000000000000000000000000..cc66298a14997da4aa2efc71e37c0a6bcda53fd1
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/loss.py
@@ -0,0 +1,398 @@
+from multiprocessing.sharedctypes import Value
+import torch
+import torch.distributed.nn
+from torch import distributed as dist, nn as nn
+from torch.nn import functional as F
+import numpy as np
+from sklearn.metrics import average_precision_score, roc_auc_score, accuracy_score
+
+try:
+    import horovod.torch as hvd
+except ImportError:
+    hvd = None
+
+
+def gather_features(
+    audio_features,
+    text_features,
+    audio_features_mlp=None,
+    text_features_mlp=None,
+    local_loss=False,
+    gather_with_grad=False,
+    rank=0,
+    world_size=1,
+    use_horovod=False,
+    mlp_loss=False,
+):
+    if use_horovod:
+        assert hvd is not None, "Please install horovod"
+        if gather_with_grad:
+            all_audio_features = hvd.allgather(audio_features)
+            all_text_features = hvd.allgather(text_features)
+            if mlp_loss:
+                all_audio_features_mlp = hvd.allgather(audio_features_mlp)
+                all_text_features_mlp = hvd.allgather(text_features_mlp)
+        else:
+            with torch.no_grad():
+                all_audio_features = hvd.allgather(audio_features)
+                all_text_features = hvd.allgather(text_features)
+                if mlp_loss:
+                    all_audio_features_mlp = hvd.allgather(audio_features_mlp)
+                    all_text_features_mlp = hvd.allgather(text_features_mlp)
+            if not local_loss:
+                # ensure grads for local rank when all_* features don't have a gradient
+                gathered_audio_features = list(
+                    all_audio_features.chunk(world_size, dim=0)
+                )
+                gathered_text_features = list(
+                    all_text_features.chunk(world_size, dim=0)
+                )
+                gathered_audio_features[rank] = audio_features
+                gathered_text_features[rank] = text_features
+                all_audio_features = torch.cat(gathered_audio_features, dim=0)
+                all_text_features = torch.cat(gathered_text_features, dim=0)
+                if mlp_loss:
+                    gathered_audio_features_mlp = list(
+                        all_audio_features_mlp.chunk(world_size, dim=0)
+                    )
+                    gathered_text_features_mlp = list(
+                        all_text_features_mlp.chunk(world_size, dim=0)
+                    )
+                    gathered_audio_features_mlp[rank] = audio_features_mlp
+                    gathered_text_features_mlp[rank] = text_features_mlp
+                    all_audio_features_mlp = torch.cat(
+                        gathered_audio_features_mlp, dim=0
+                    )
+                    all_text_features_mlp = torch.cat(gathered_text_features_mlp, dim=0)
+    else:
+        # We gather tensors from all gpus
+        if gather_with_grad:
+            all_audio_features = torch.cat(
+                torch.distributed.nn.all_gather(audio_features), dim=0
+            )
+            all_text_features = torch.cat(
+                torch.distributed.nn.all_gather(text_features), dim=0
+            )
+            if mlp_loss:
+                all_audio_features_mlp = torch.cat(
+                    torch.distributed.nn.all_gather(audio_features_mlp), dim=0
+                )
+                all_text_features_mlp = torch.cat(
+                    torch.distributed.nn.all_gather(text_features_mlp), dim=0
+                )
+        else:
+            gathered_audio_features = [
+                torch.zeros_like(audio_features) for _ in range(world_size)
+            ]
+            gathered_text_features = [
+                torch.zeros_like(text_features) for _ in range(world_size)
+            ]
+            dist.all_gather(gathered_audio_features, audio_features)
+            dist.all_gather(gathered_text_features, text_features)
+            if mlp_loss:
+                gathered_audio_features_mlp = [
+                    torch.zeros_like(audio_features_mlp) for _ in range(world_size)
+                ]
+                gathered_text_features_mlp = [
+                    torch.zeros_like(text_features_mlp) for _ in range(world_size)
+                ]
+                dist.all_gather(gathered_audio_features_mlp, audio_features_mlp)
+                dist.all_gather(gathered_text_features_mlp, text_features_mlp)
+            if not local_loss:
+                # ensure grads for local rank when all_* features don't have a gradient
+                gathered_audio_features[rank] = audio_features
+                gathered_text_features[rank] = text_features
+                if mlp_loss:
+                    gathered_audio_features_mlp[rank] = audio_features_mlp
+                    gathered_text_features_mlp[rank] = text_features_mlp
+
+            all_audio_features = torch.cat(gathered_audio_features, dim=0)
+            all_text_features = torch.cat(gathered_text_features, dim=0)
+            if mlp_loss:
+                all_audio_features_mlp = torch.cat(gathered_audio_features_mlp, dim=0)
+                all_text_features_mlp = torch.cat(gathered_text_features_mlp, dim=0)
+    if mlp_loss:
+        return (
+            all_audio_features,
+            all_text_features,
+            all_audio_features_mlp,
+            all_text_features_mlp,
+        )
+    else:
+        return all_audio_features, all_text_features
+
+
+class ClipLoss(nn.Module):
+    def __init__(
+        self,
+        local_loss=False,
+        gather_with_grad=False,
+        cache_labels=False,
+        rank=0,
+        world_size=1,
+        use_horovod=False,
+        mlp_loss=False,
+        weight_loss_kappa=0,
+    ):
+        super().__init__()
+        self.local_loss = local_loss
+        self.gather_with_grad = gather_with_grad
+        self.cache_labels = cache_labels
+        self.rank = rank
+        self.world_size = world_size
+        self.use_horovod = use_horovod
+        self.mlp_loss = mlp_loss
+        self.weighted_loss = bool(weight_loss_kappa != 0)
+        self.weight_loss_kappa = weight_loss_kappa
+        # cache state
+        self.prev_num_logits = 0
+        self.labels = {}
+
+    def forward(
+        self,
+        audio_features,
+        text_features,
+        logit_scale_a,
+        logit_scale_t=None,
+        audio_features_mlp=None,
+        text_features_mlp=None,
+    ):
+        device = audio_features.device
+        if self.mlp_loss:
+            if self.world_size > 1:
+                (
+                    all_audio_features,
+                    all_text_features,
+                    all_audio_features_mlp,
+                    all_text_features_mlp,
+                ) = gather_features(
+                    audio_features=audio_features,
+                    text_features=text_features,
+                    audio_features_mlp=audio_features_mlp,
+                    text_features_mlp=text_features_mlp,
+                    local_loss=self.local_loss,
+                    gather_with_grad=self.gather_with_grad,
+                    rank=self.rank,
+                    world_size=self.world_size,
+                    use_horovod=self.use_horovod,
+                    mlp_loss=self.mlp_loss,
+                )
+                if self.local_loss:
+                    a_logits_per_audio = (
+                        logit_scale_a * audio_features @ all_text_features_mlp.T
+                    )
+                    a_logits_per_text = (
+                        logit_scale_a * text_features_mlp @ all_audio_features.T
+                    )
+                    t_logits_per_audio = (
+                        logit_scale_t * audio_features_mlp @ all_text_features.T
+                    )
+                    t_logits_per_text = (
+                        logit_scale_t * text_features @ all_audio_features_mlp.T
+                    )
+                else:
+                    a_logits_per_audio = (
+                        logit_scale_a * all_audio_features @ all_text_features_mlp.T
+                    )
+                    a_logits_per_text = a_logits_per_audio.T
+                    t_logits_per_audio = (
+                        logit_scale_t * all_audio_features_mlp @ all_text_features.T
+                    )
+                    t_logits_per_text = t_logits_per_audio.T
+            else:
+                a_logits_per_audio = (
+                    logit_scale_a * audio_features @ text_features_mlp.T
+                )
+                a_logits_per_text = logit_scale_a * text_features_mlp @ audio_features.T
+                t_logits_per_audio = (
+                    logit_scale_t * audio_features_mlp @ text_features.T
+                )
+                t_logits_per_text = logit_scale_t * text_features @ audio_features_mlp.T
+
+            # calculated ground-truth and cache if enabled
+            num_logits = a_logits_per_audio.shape[0]
+            if self.prev_num_logits != num_logits or device not in self.labels:
+                labels = torch.arange(num_logits, device=device, dtype=torch.long)
+                if self.world_size > 1 and self.local_loss:
+                    labels = labels + num_logits * self.rank
+                if self.cache_labels:
+                    self.labels[device] = labels
+                    self.prev_num_logits = num_logits
+            else:
+                labels = self.labels[device]
+
+            if not self.weighted_loss:
+                total_loss = (
+                    F.cross_entropy(a_logits_per_audio, labels)
+                    + F.cross_entropy(a_logits_per_text, labels)
+                    + F.cross_entropy(t_logits_per_audio, labels)
+                    + F.cross_entropy(t_logits_per_text, labels)
+                ) / 4
+            else:
+                audio_weight = (audio_features @ audio_features.T).detach()
+                audio_weight = (
+                    torch.exp(
+                        torch.sum(audio_weight, axis=1)
+                        / (self.weight_loss_kappa * len(audio_weight))
+                    )
+                ).detach()
+                text_weight = (text_features @ text_features.T).detach()
+                text_weight = (
+                    torch.exp(
+                        torch.sum(text_weight, axis=1)
+                        / (self.weight_loss_kappa * len(text_features))
+                    )
+                ).detach()
+                total_loss = (
+                    F.cross_entropy(a_logits_per_audio, labels, weight=audio_weight)
+                    + F.cross_entropy(a_logits_per_text, labels, weight=audio_weight)
+                    + F.cross_entropy(t_logits_per_audio, labels, weight=text_weight)
+                    + F.cross_entropy(t_logits_per_text, labels, weight=text_weight)
+                ) / 4
+        else:
+            if self.world_size > 1:
+                all_audio_features, all_text_features = gather_features(
+                    audio_features=audio_features,
+                    text_features=text_features,
+                    local_loss=self.local_loss,
+                    gather_with_grad=self.gather_with_grad,
+                    rank=self.rank,
+                    world_size=self.world_size,
+                    use_horovod=self.use_horovod,
+                    mlp_loss=self.mlp_loss,
+                )
+
+                if self.local_loss:
+                    logits_per_audio = (
+                        logit_scale_a * audio_features @ all_text_features.T
+                    )
+                    logits_per_text = (
+                        logit_scale_a * text_features @ all_audio_features.T
+                    )
+                else:
+                    logits_per_audio = (
+                        logit_scale_a * all_audio_features @ all_text_features.T
+                    )
+                    logits_per_text = logits_per_audio.T
+            else:
+                logits_per_audio = logit_scale_a * audio_features @ text_features.T
+                logits_per_text = logit_scale_a * text_features @ audio_features.T
+
+            # calculated ground-truth and cache if enabled
+            num_logits = logits_per_audio.shape[0]
+            if self.prev_num_logits != num_logits or device not in self.labels:
+                labels = torch.arange(num_logits, device=device, dtype=torch.long)
+                if self.world_size > 1 and self.local_loss:
+                    labels = labels + num_logits * self.rank
+                if self.cache_labels:
+                    self.labels[device] = labels
+                    self.prev_num_logits = num_logits
+            else:
+                labels = self.labels[device]
+            if not self.weighted_loss:
+                total_loss = (
+                    F.cross_entropy(logits_per_audio, labels)
+                    + F.cross_entropy(logits_per_text, labels)
+                ) / 2
+            else:
+                audio_weight = (all_audio_features @ all_audio_features.T).detach()
+                audio_weight = (
+                    torch.exp(
+                        torch.sum(audio_weight, axis=1)
+                        / (self.weight_loss_kappa * len(all_audio_features))
+                    )
+                ).detach()
+                text_weight = (all_text_features @ all_text_features.T).detach()
+                text_weight = (
+                    torch.exp(
+                        torch.sum(text_weight, axis=1)
+                        / (self.weight_loss_kappa * len(all_text_features))
+                    )
+                ).detach()
+                total_loss = (
+                    F.cross_entropy(logits_per_audio, labels, weight=text_weight)
+                    + F.cross_entropy(logits_per_text, labels, weight=audio_weight)
+                ) / 2
+        return total_loss
+
+
+def lp_gather_features(pred, target, world_size=1, use_horovod=False):
+    if use_horovod:
+        assert hvd is not None, "Please install horovod"
+        with torch.no_grad():
+            all_preds = hvd.allgather(pred)
+            all_targets = hvd.allgath(target)
+    else:
+        gathered_preds = [torch.zeros_like(pred) for _ in range(world_size)]
+        gathered_targets = [torch.zeros_like(target) for _ in range(world_size)]
+
+        dist.all_gather(gathered_preds, pred)
+        dist.all_gather(gathered_targets, target)
+        all_preds = torch.cat(gathered_preds, dim=0)
+        all_targets = torch.cat(gathered_targets, dim=0)
+
+    return all_preds, all_targets
+
+
+def get_map(pred, target):
+    pred = torch.sigmoid(pred).numpy()
+    target = target.numpy()
+    return np.mean(average_precision_score(target, pred, average=None))
+
+
+def get_acc(pred, target):
+    pred = torch.argmax(pred, 1).numpy()
+    target = torch.argmax(target, 1).numpy()
+    return accuracy_score(target, pred)
+
+
+def get_mauc(pred, target):
+    pred = torch.sigmoid(pred).numpy()
+    target = target.numpy()
+    return np.mean(roc_auc_score(target, pred, average=None))
+
+
+class LPMetrics(object):
+    def __init__(self, metric_names=["map", "acc", "mauc"]):
+        self.metrics = []
+        for name in metric_names:
+            self.metrics.append(self.get_metric(name))
+        self.metric_names = metric_names
+
+    def get_metric(self, name):
+        if name == "map":
+            return get_map
+        elif name == "acc":
+            return get_acc
+        elif name == "mauc":
+            return get_mauc
+        else:
+            raise ValueError(f"the metric should be at least one of [map, acc, mauc]")
+
+    def evaluate_mertics(self, pred, target):
+        metric_dict = {}
+        for i in range(len(self.metric_names)):
+            metric_dict[self.metric_names[i]] = self.metrics[i](pred, target)
+        return metric_dict
+
+
+def calc_celoss(pred, target):
+    target = torch.argmax(target, 1).long()
+    return nn.CrossEntropyLoss()(pred, target)
+
+
+class LPLoss(nn.Module):
+    def __init__(self, loss_name):
+        super().__init__()
+        if loss_name == "bce":
+            self.loss_func = nn.BCEWithLogitsLoss()
+        elif loss_name == "ce":
+            self.loss_func = calc_celoss
+        elif loss_name == "mse":
+            self.loss_func = nn.MSELoss()
+        else:
+            raise ValueError(f"the loss func should be at least one of [bce, ce, mse]")
+
+    def forward(self, pred, target):
+        loss = self.loss_func(pred, target)
+        return loss
diff --git a/picoaudio/audioldm/clap/open_clip/model.py b/picoaudio/audioldm/clap/open_clip/model.py
new file mode 100644
index 0000000000000000000000000000000000000000..b439244f8c293a0b4263b7ac1fd553e9d0adf184
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model.py
@@ -0,0 +1,936 @@
+""" CLAP Model
+
+Adapted from CLIP: https://github.com/openai/CLIP. Originally MIT License, Copyright (c) 2021 OpenAI.
+Adapted to the Audio Task.
+"""
+
+from collections import OrderedDict
+from dataclasses import dataclass
+from email.mime import audio
+from typing import Tuple, Union, Callable, Optional
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+from torch import nn
+
+from .timm_model import TimmModel
+import logging
+from .utils import freeze_batch_norm_2d
+
+from .pann_model import create_pann_model
+from .htsat import create_htsat_model
+from transformers import BertModel, RobertaModel, BartModel
+from transformers.tokenization_utils_base import BatchEncoding
+
+
+class MLPLayers(nn.Module):
+    def __init__(self, units=[512, 512, 512], nonlin=nn.ReLU(), dropout=0.1):
+        super(MLPLayers, self).__init__()
+        self.nonlin = nonlin
+        self.dropout = dropout
+
+        sequence = []
+        for u0, u1 in zip(units[:-1], units[1:]):
+            sequence.append(nn.Linear(u0, u1))
+            sequence.append(self.nonlin)
+            sequence.append(nn.Dropout(self.dropout))
+        sequence = sequence[:-2]
+
+        self.sequential = nn.Sequential(*sequence)
+
+    def forward(self, X):
+        X = self.sequential(X)
+        return X
+
+
+class Bottleneck(nn.Module):
+    expansion = 4
+
+    def __init__(self, inplanes, planes, stride=1):
+        super().__init__()
+
+        # all conv layers have stride 1. an avgpool is performed after the second convolution when stride > 1
+        self.conv1 = nn.Conv2d(inplanes, planes, 1, bias=False)
+        self.bn1 = nn.BatchNorm2d(planes)
+
+        self.conv2 = nn.Conv2d(planes, planes, 3, padding=1, bias=False)
+        self.bn2 = nn.BatchNorm2d(planes)
+
+        self.avgpool = nn.AvgPool2d(stride) if stride > 1 else nn.Identity()
+
+        self.conv3 = nn.Conv2d(planes, planes * self.expansion, 1, bias=False)
+        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
+
+        self.relu = nn.ReLU(inplace=True)
+        self.downsample = None
+        self.stride = stride
+
+        if stride > 1 or inplanes != planes * Bottleneck.expansion:
+            # downsampling layer is prepended with an avgpool, and the subsequent convolution has stride 1
+            self.downsample = nn.Sequential(
+                OrderedDict(
+                    [
+                        ("-1", nn.AvgPool2d(stride)),
+                        (
+                            "0",
+                            nn.Conv2d(
+                                inplanes,
+                                planes * self.expansion,
+                                1,
+                                stride=1,
+                                bias=False,
+                            ),
+                        ),
+                        ("1", nn.BatchNorm2d(planes * self.expansion)),
+                    ]
+                )
+            )
+
+    def forward(self, x: torch.Tensor):
+        identity = x
+
+        out = self.relu(self.bn1(self.conv1(x)))
+        out = self.relu(self.bn2(self.conv2(out)))
+        out = self.avgpool(out)
+        out = self.bn3(self.conv3(out))
+
+        if self.downsample is not None:
+            identity = self.downsample(x)
+
+        out += identity
+        out = self.relu(out)
+        return out
+
+
+class AttentionPool2d(nn.Module):
+    def __init__(
+        self, spacial_dim: int, embed_dim: int, num_heads: int, output_dim: int = None
+    ):
+        super().__init__()
+        self.positional_embedding = nn.Parameter(
+            torch.randn(spacial_dim**2 + 1, embed_dim) / embed_dim**0.5
+        )
+        self.k_proj = nn.Linear(embed_dim, embed_dim)
+        self.q_proj = nn.Linear(embed_dim, embed_dim)
+        self.v_proj = nn.Linear(embed_dim, embed_dim)
+        self.c_proj = nn.Linear(embed_dim, output_dim or embed_dim)
+        self.num_heads = num_heads
+
+    def forward(self, x):
+        x = x.reshape(x.shape[0], x.shape[1], x.shape[2] * x.shape[3]).permute(
+            2, 0, 1
+        )  # NCHW -> (HW)NC
+        x = torch.cat([x.mean(dim=0, keepdim=True), x], dim=0)  # (HW+1)NC
+        x = x + self.positional_embedding[:, None, :].to(x.dtype)  # (HW+1)NC
+        x, _ = F.multi_head_attention_forward(
+            query=x,
+            key=x,
+            value=x,
+            embed_dim_to_check=x.shape[-1],
+            num_heads=self.num_heads,
+            q_proj_weight=self.q_proj.weight,
+            k_proj_weight=self.k_proj.weight,
+            v_proj_weight=self.v_proj.weight,
+            in_proj_weight=None,
+            in_proj_bias=torch.cat(
+                [self.q_proj.bias, self.k_proj.bias, self.v_proj.bias]
+            ),
+            bias_k=None,
+            bias_v=None,
+            add_zero_attn=False,
+            dropout_p=0,
+            out_proj_weight=self.c_proj.weight,
+            out_proj_bias=self.c_proj.bias,
+            use_separate_proj_weight=True,
+            training=self.training,
+            need_weights=False,
+        )
+
+        return x[0]
+
+
+class ModifiedResNet(nn.Module):
+    """
+    A ResNet class that is similar to torchvision's but contains the following changes:
+    - There are now 3 "stem" convolutions as opposed to 1, with an average pool instead of a max pool.
+    - Performs anti-aliasing strided convolutions, where an avgpool is prepended to convolutions with stride > 1
+    - The final pooling layer is a QKV attention instead of an average pool
+    """
+
+    def __init__(self, layers, output_dim, heads, image_size=224, width=64):
+        super().__init__()
+        self.output_dim = output_dim
+        self.image_size = image_size
+
+        # the 3-layer stem
+        self.conv1 = nn.Conv2d(
+            3, width // 2, kernel_size=3, stride=2, padding=1, bias=False
+        )
+        self.bn1 = nn.BatchNorm2d(width // 2)
+        self.conv2 = nn.Conv2d(
+            width // 2, width // 2, kernel_size=3, padding=1, bias=False
+        )
+        self.bn2 = nn.BatchNorm2d(width // 2)
+        self.conv3 = nn.Conv2d(width // 2, width, kernel_size=3, padding=1, bias=False)
+        self.bn3 = nn.BatchNorm2d(width)
+        self.avgpool = nn.AvgPool2d(2)
+        self.relu = nn.ReLU(inplace=True)
+
+        # residual layers
+        self._inplanes = width  # this is a *mutable* variable used during construction
+        self.layer1 = self._make_layer(width, layers[0])
+        self.layer2 = self._make_layer(width * 2, layers[1], stride=2)
+        self.layer3 = self._make_layer(width * 4, layers[2], stride=2)
+        self.layer4 = self._make_layer(width * 8, layers[3], stride=2)
+
+        embed_dim = width * 32  # the ResNet feature dimension
+        self.attnpool = AttentionPool2d(image_size // 32, embed_dim, heads, output_dim)
+
+        self.init_parameters()
+
+    def _make_layer(self, planes, blocks, stride=1):
+        layers = [Bottleneck(self._inplanes, planes, stride)]
+
+        self._inplanes = planes * Bottleneck.expansion
+        for _ in range(1, blocks):
+            layers.append(Bottleneck(self._inplanes, planes))
+
+        return nn.Sequential(*layers)
+
+    def init_parameters(self):
+        if self.attnpool is not None:
+            std = self.attnpool.c_proj.in_features**-0.5
+            nn.init.normal_(self.attnpool.q_proj.weight, std=std)
+            nn.init.normal_(self.attnpool.k_proj.weight, std=std)
+            nn.init.normal_(self.attnpool.v_proj.weight, std=std)
+            nn.init.normal_(self.attnpool.c_proj.weight, std=std)
+
+        for resnet_block in [self.layer1, self.layer2, self.layer3, self.layer4]:
+            for name, param in resnet_block.named_parameters():
+                if name.endswith("bn3.weight"):
+                    nn.init.zeros_(param)
+
+    def lock(self, unlocked_groups=0, freeze_bn_stats=False):
+        assert (
+            unlocked_groups == 0
+        ), "partial locking not currently supported for this model"
+        for param in self.parameters():
+            param.requires_grad = False
+        if freeze_bn_stats:
+            freeze_batch_norm_2d(self)
+
+    def stem(self, x):
+        for conv, bn in [
+            (self.conv1, self.bn1),
+            (self.conv2, self.bn2),
+            (self.conv3, self.bn3),
+        ]:
+            x = self.relu(bn(conv(x)))
+        x = self.avgpool(x)
+        return x
+
+    def forward(self, x):
+        x = self.stem(x)
+        x = self.layer1(x)
+        x = self.layer2(x)
+        x = self.layer3(x)
+        x = self.layer4(x)
+        x = self.attnpool(x)
+
+        return x
+
+
+class LayerNorm(nn.LayerNorm):
+    """Subclass torch's LayerNorm to handle fp16."""
+
+    def forward(self, x: torch.Tensor):
+        orig_type = x.dtype
+        x = F.layer_norm(x, self.normalized_shape, self.weight, self.bias, self.eps)
+        return x.to(orig_type)
+
+
+class QuickGELU(nn.Module):
+    # NOTE This is slower than nn.GELU or nn.SiLU and uses more GPU memory
+    def forward(self, x: torch.Tensor):
+        return x * torch.sigmoid(1.702 * x)
+
+
+class ResidualAttentionBlock(nn.Module):
+    def __init__(self, d_model: int, n_head: int, act_layer: Callable = nn.GELU):
+        super().__init__()
+
+        self.attn = nn.MultiheadAttention(d_model, n_head)
+        self.ln_1 = LayerNorm(d_model)
+        self.mlp = nn.Sequential(
+            OrderedDict(
+                [
+                    ("c_fc", nn.Linear(d_model, d_model * 4)),
+                    ("gelu", act_layer()),
+                    ("c_proj", nn.Linear(d_model * 4, d_model)),
+                ]
+            )
+        )
+        self.ln_2 = LayerNorm(d_model)
+
+    def attention(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None):
+        return self.attn(x, x, x, need_weights=False, attn_mask=attn_mask)[0]
+
+    def forward(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None):
+        x = x + self.attention(self.ln_1(x), attn_mask=attn_mask)
+        x = x + self.mlp(self.ln_2(x))
+        return x
+
+
+class Transformer(nn.Module):
+    def __init__(
+        self, width: int, layers: int, heads: int, act_layer: Callable = nn.GELU
+    ):
+        super().__init__()
+        self.width = width
+        self.layers = layers
+        self.resblocks = nn.ModuleList(
+            [
+                ResidualAttentionBlock(width, heads, act_layer=act_layer)
+                for _ in range(layers)
+            ]
+        )
+
+    def forward(self, x: torch.Tensor, attn_mask: Optional[torch.Tensor] = None):
+        for r in self.resblocks:
+            x = r(x, attn_mask=attn_mask)
+        return x
+
+
+class VisualTransformer(nn.Module):
+    def __init__(
+        self,
+        image_size: int,
+        patch_size: int,
+        width: int,
+        layers: int,
+        heads: int,
+        output_dim: int,
+        act_layer: Callable = nn.GELU,
+    ):
+        super().__init__()
+        self.image_size = image_size
+        self.output_dim = output_dim
+        self.conv1 = nn.Conv2d(
+            in_channels=3,
+            out_channels=width,
+            kernel_size=patch_size,
+            stride=patch_size,
+            bias=False,
+        )
+
+        scale = width**-0.5
+        self.class_embedding = nn.Parameter(scale * torch.randn(width))
+        self.positional_embedding = nn.Parameter(
+            scale * torch.randn((image_size // patch_size) ** 2 + 1, width)
+        )
+        self.ln_pre = LayerNorm(width)
+
+        self.text_branch = Transformer(width, layers, heads, act_layer=act_layer)
+
+        self.ln_post = LayerNorm(width)
+        self.proj = nn.Parameter(scale * torch.randn(width, output_dim))
+
+    def lock(self, unlocked_groups=0, freeze_bn_stats=False):
+        assert (
+            unlocked_groups == 0
+        ), "partial locking not currently supported for this model"
+        for param in self.parameters():
+            param.requires_grad = False
+
+    def forward(self, x: torch.Tensor):
+        x = self.conv1(x)  # shape = [*, width, grid, grid]
+        x = x.reshape(x.shape[0], x.shape[1], -1)  # shape = [*, width, grid ** 2]
+        x = x.permute(0, 2, 1)  # shape = [*, grid ** 2, width]
+        x = torch.cat(
+            [
+                self.class_embedding.to(x.dtype)
+                + torch.zeros(
+                    x.shape[0], 1, x.shape[-1], dtype=x.dtype, device=x.device
+                ),
+                x,
+            ],
+            dim=1,
+        )  # shape = [*, grid ** 2 + 1, width]
+        x = x + self.positional_embedding.to(x.dtype)
+        x = self.ln_pre(x)
+
+        x = x.permute(1, 0, 2)  # NLD -> LND
+        x = self.text_branch(x)
+        x = x.permute(1, 0, 2)  # LND -> NLD
+
+        x = self.ln_post(x[:, 0, :])
+
+        if self.proj is not None:
+            x = x @ self.proj
+
+        return x
+
+
+@dataclass
+class CLAPVisionCfg:
+    layers: Union[Tuple[int, int, int, int], int] = 12
+    width: int = 768
+    patch_size: int = 16
+    image_size: Union[Tuple[int, int], int] = 224
+    timm_model_name: str = (
+        None  # a valid model name overrides layers, width, patch_size
+    )
+    timm_model_pretrained: bool = (
+        False  # use (imagenet) pretrained weights for named model
+    )
+    timm_pool: str = (
+        "avg"  # feature pooling for timm model ('abs_attn', 'rot_attn', 'avg', '')
+    )
+    timm_proj: str = (
+        "linear"  # linear projection for timm model output ('linear', 'mlp', '')
+    )
+
+
+# Audio Config Class
+@dataclass
+class CLAPAudioCfp:
+    model_type: str = "PANN"
+    model_name: str = "Cnn14"
+    sample_rate: int = 48000
+    # Param
+    audio_length: int = 1024
+    window_size: int = 1024
+    hop_size: int = 1024
+    fmin: int = 50
+    fmax: int = 14000
+    class_num: int = 527
+    mel_bins: int = 64
+    clip_samples: int = 480000
+
+
+@dataclass
+class CLAPTextCfg:
+    context_length: int
+    vocab_size: int
+    width: int
+    heads: int
+    layers: int
+    model_type: str
+
+
+class CLAP(nn.Module):
+    def __init__(
+        self,
+        embed_dim: int,
+        audio_cfg: CLAPAudioCfp,
+        text_cfg: CLAPTextCfg,
+        quick_gelu: bool = False,
+        enable_fusion: bool = False,
+        fusion_type: str = "None",
+        joint_embed_shape: int = 512,
+        mlp_act: str = "relu",
+    ):
+        super().__init__()
+        if isinstance(audio_cfg, dict):
+            audio_cfg = CLAPAudioCfp(**audio_cfg)
+        if isinstance(text_cfg, dict):
+            text_cfg = CLAPTextCfg(**text_cfg)
+
+        self.audio_cfg = audio_cfg
+        self.text_cfg = text_cfg
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+        self.joint_embed_shape = joint_embed_shape
+        self.mlp_act = mlp_act
+
+        self.context_length = text_cfg.context_length
+
+        # OpenAI models are pretrained w/ QuickGELU but native nn.GELU is both faster and more
+        # memory efficient in recent PyTorch releases (>= 1.10).
+        # NOTE: timm models always use native GELU regardless of quick_gelu flag.
+        act_layer = QuickGELU if quick_gelu else nn.GELU
+
+        if mlp_act == "relu":
+            mlp_act_layer = nn.ReLU()
+        elif mlp_act == "gelu":
+            mlp_act_layer = nn.GELU()
+        else:
+            raise NotImplementedError
+
+        # audio branch
+        # audio branch parameters
+        if audio_cfg.model_type == "PANN":
+            self.audio_branch = create_pann_model(audio_cfg, enable_fusion, fusion_type)
+        elif audio_cfg.model_type == "HTSAT":
+            self.audio_branch = create_htsat_model(
+                audio_cfg, enable_fusion, fusion_type
+            )
+        else:
+            logging.error(f"Model config for {audio_cfg.model_type} not found")
+            raise RuntimeError(f"Model config for {audio_cfg.model_type} not found.")
+
+        # text branch
+        # text branch parameters
+        if text_cfg.model_type == "transformer":
+            self.text_branch = Transformer(
+                width=text_cfg.width,
+                layers=text_cfg.layers,
+                heads=text_cfg.heads,
+                act_layer=act_layer,
+            )
+            self.vocab_size = text_cfg.vocab_size
+            self.token_embedding = nn.Embedding(text_cfg.vocab_size, text_cfg.width)
+            self.positional_embedding = nn.Parameter(
+                torch.empty(self.context_length, text_cfg.width)
+            )
+            self.ln_final = LayerNorm(text_cfg.width)
+            self.text_transform = MLPLayers(
+                units=[
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                ],
+                dropout=0.1,
+            )
+            self.text_projection = nn.Sequential(
+                nn.Linear(text_cfg.width, self.joint_embed_shape),
+                mlp_act_layer,
+                nn.Linear(self.joint_embed_shape, self.joint_embed_shape),
+            )
+        elif text_cfg.model_type == "bert":
+            self.text_branch = BertModel.from_pretrained("bert-base-uncased")
+            self.text_transform = MLPLayers(
+                units=[
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                ],
+                dropout=0.1,
+            )
+            self.text_projection = nn.Sequential(
+                nn.Linear(768, self.joint_embed_shape),
+                mlp_act_layer,
+                nn.Linear(self.joint_embed_shape, self.joint_embed_shape),
+            )
+        elif text_cfg.model_type == "roberta":
+            self.text_branch = RobertaModel.from_pretrained("roberta-base")
+            self.text_transform = MLPLayers(
+                units=[
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                ],
+                dropout=0.1,
+            )
+            self.text_projection = nn.Sequential(
+                nn.Linear(768, self.joint_embed_shape),
+                mlp_act_layer,
+                nn.Linear(self.joint_embed_shape, self.joint_embed_shape),
+            )
+        elif text_cfg.model_type == "bart":
+            self.text_branch = BartModel.from_pretrained("facebook/bart-base")
+            self.text_transform = MLPLayers(
+                units=[
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                    self.joint_embed_shape,
+                ],
+                dropout=0.1,
+            )
+            self.text_projection = nn.Sequential(
+                nn.Linear(768, self.joint_embed_shape),
+                mlp_act_layer,
+                nn.Linear(self.joint_embed_shape, self.joint_embed_shape),
+            )
+        else:
+            logging.error(f"Model config for {text_cfg.model_type} not found")
+            raise RuntimeError(f"Model config for {text_cfg.model_type} not found.")
+        self.text_branch_type = text_cfg.model_type
+        # text branch parameters
+
+        # audio branch parameters
+        self.audio_transform = MLPLayers(
+            units=[
+                self.joint_embed_shape,
+                self.joint_embed_shape,
+                self.joint_embed_shape,
+            ],
+            dropout=0.1,
+        )
+
+        # below here is text branch parameters
+
+        # ============================================================================================================
+        self.audio_projection = nn.Sequential(
+            nn.Linear(embed_dim, self.joint_embed_shape),
+            mlp_act_layer,
+            nn.Linear(self.joint_embed_shape, self.joint_embed_shape),
+        )
+
+        self.logit_scale_a = nn.Parameter(torch.ones([]) * np.log(1 / 0.07))
+        self.logit_scale_t = nn.Parameter(torch.ones([]) * np.log(1 / 0.07))
+        self.register_buffer("attn_mask", self.build_attention_mask(), persistent=False)
+
+        self.init_text_branch_parameters()
+
+    def init_text_branch_parameters(self):
+        if self.text_branch_type == "transformer":
+            nn.init.normal_(self.token_embedding.weight, std=0.02)
+            nn.init.normal_(self.positional_embedding, std=0.01)
+            proj_std = (self.text_branch.width**-0.5) * (
+                (2 * self.text_branch.layers) ** -0.5
+            )
+            attn_std = self.text_branch.width**-0.5
+            fc_std = (2 * self.text_branch.width) ** -0.5
+            for block in self.text_branch.resblocks:
+                nn.init.normal_(block.attn.in_proj_weight, std=attn_std)
+                nn.init.normal_(block.attn.out_proj.weight, std=proj_std)
+                nn.init.normal_(block.mlp.c_fc.weight, std=fc_std)
+                nn.init.normal_(block.mlp.c_proj.weight, std=proj_std)
+        if self.text_branch_type == "bert" or self.text_branch_type == "roberta":
+            width = self.text_branch.embeddings.word_embeddings.weight.shape[-1]
+        elif self.text_branch_type == "bart":
+            width = self.text_branch.shared.weight.shape[-1]
+        else:
+            width = self.text_branch.width
+        nn.init.constant_(self.logit_scale_a, np.log(1 / 0.07))
+        nn.init.constant_(self.logit_scale_t, np.log(1 / 0.07))
+
+        # deprecated
+        # if hasattr(self.visual, 'init_parameters'):
+        # self.visual.init_parameters()
+
+        # if self.text_projection is not None:
+        #     nn.init.normal_(self.text_projection, std=width**-0.5)
+
+    def build_attention_mask(self):
+        # lazily create causal attention mask, with full attention between the vision tokens
+        # pytorch uses additive attention mask; fill with -inf
+        mask = torch.empty(self.context_length, self.context_length)
+        mask.fill_(float("-inf"))
+        mask.triu_(1)  # zero out the lower diagonal
+        return mask
+
+    def encode_audio(self, audio, device):
+        return self.audio_branch(
+            audio, mixup_lambda=None, device=device
+        )  # mix lambda needs to add
+
+    # def list_of_dict_of_tensor2dict_of_tensor(self, x, device):
+    #     tmp = {}
+    #     for k in x[0].keys():
+    #         tmp[k] = []
+    #         for i in range(len(x)):
+    #             tmp[k].append(x[i][k][:77])
+    #     for k in x[0].keys():
+    #         tmp[k] = torch.tensor(tmp[k]).to(device=device, non_blocking=True)
+    #     return tmp
+
+    def encode_text(self, text, device):
+        if self.text_branch_type == "transformer":
+            text = text.to(device=device, non_blocking=True)
+            x = self.token_embedding(text)  # [batch_size, n_ctx, d_model]
+
+            x = x + self.positional_embedding
+            x = x.permute(1, 0, 2)  # NLD -> LND
+            x = self.text_branch(x, attn_mask=self.attn_mask)
+            x = x.permute(1, 0, 2)  # LND -> NLD
+            x = self.ln_final(x)
+
+            # x.shape = [batch_size, n_ctx, transformer.width]
+            # take features from the eot embedding (eot_token is the highest number in each sequence)
+            x = self.text_projection(x[torch.arange(x.shape[0]), text.argmax(dim=-1)])
+        elif self.text_branch_type == "bert":
+            # text = self.list_of_dict_of_tensor2dict_of_tensor(text, device)
+            # text = BatchEncoding(text)
+            x = self.text_branch(
+                input_ids=text["input_ids"].to(device=device, non_blocking=True),
+                attention_mask=text["attention_mask"].to(
+                    device=device, non_blocking=True
+                ),
+                token_type_ids=text["token_type_ids"].to(
+                    device=device, non_blocking=True
+                ),
+            )["pooler_output"]
+            x = self.text_projection(x)
+        elif self.text_branch_type == "roberta":
+            x = self.text_branch(
+                input_ids=text["input_ids"].to(device=device, non_blocking=True),
+                attention_mask=text["attention_mask"].to(
+                    device=device, non_blocking=True
+                ),
+            )["pooler_output"]
+            x = self.text_projection(x)
+        elif self.text_branch_type == "bart":
+            x = torch.mean(
+                self.text_branch(
+                    input_ids=text["input_ids"].to(device=device, non_blocking=True),
+                    attention_mask=text["attention_mask"].to(
+                        device=device, non_blocking=True
+                    ),
+                )["encoder_last_hidden_state"],
+                axis=1,
+            )
+            x = self.text_projection(x)
+        else:
+            logging.error(f"Model type {self.text_branch_type} not found")
+            raise RuntimeError(f"Model type {self.text_branch_type} not found.")
+        return x
+
+    def forward(self, audio, text, device=None):
+        """Forward audio and text into the CLAP
+
+        Parameters
+        ----------
+        audio: torch.Tensor (batch_size, audio_length)
+            the time-domain audio input / the batch of mel_spec and longer list.
+        text: torch.Tensor () // need to add
+            the text token input
+        """
+        if device is None:
+            if audio is not None:
+                device = audio.device
+            elif text is not None:
+                device = text.device
+        if audio is None and text is None:
+            # a hack to get the logit scale
+            return self.logit_scale_a.exp(), self.logit_scale_t.exp()
+        elif audio is None:
+            return self.encode_text(text, device=device)
+        elif text is None:
+            return self.audio_projection(
+                self.encode_audio(audio, device=device)["embedding"]
+            )
+        audio_features = self.audio_projection(
+            self.encode_audio(audio, device=device)["embedding"]
+        )
+        audio_features = F.normalize(audio_features, dim=-1)
+
+        text_features = self.encode_text(text, device=device)
+        # print("text_features", text_features)
+        # print("text_features.shape", text_features.shape)
+        # print("text_features.type", type(text_features))
+        text_features = F.normalize(text_features, dim=-1)
+
+        audio_features_mlp = self.audio_transform(audio_features)
+        text_features_mlp = self.text_transform(text_features)
+        # Four outputs: audio features (basic & MLP), text features (basic & MLP)
+        return (
+            audio_features,
+            text_features,
+            audio_features_mlp,
+            text_features_mlp,
+            self.logit_scale_a.exp(),
+            self.logit_scale_t.exp(),
+        )
+
+    def get_logit_scale(self):
+        return self.logit_scale_a.exp(), self.logit_scale_t.exp()
+
+    def get_text_embedding(self, data):
+        """Get the text embedding from the model
+
+        Parameters
+        ----------
+        data: torch.Tensor
+            a tensor of text embedding
+
+        Returns
+        ----------
+        text_embed: torch.Tensor
+            a tensor of text_embeds (N, D)
+
+        """
+        device = next(self.parameters()).device
+        for k in data:
+            data[k] = data[k].to(device)
+            if len(data[k].size()) < 2:
+                data[k] = data[k].unsqueeze(0)
+        text_embeds = self.encode_text(data, device=device)
+        text_embeds = F.normalize(text_embeds, dim=-1)
+
+        return text_embeds
+
+    def get_audio_embedding(self, data):
+        """Get the audio embedding from the model
+
+        Parameters
+        ----------
+        data: a list of dict
+            the audio input dict list from 'get_audio_feature' method
+
+        Returns
+        ----------
+        audio_embed: torch.Tensor
+            a tensor of audio_embeds (N, D)
+
+        """
+        device = next(self.parameters()).device
+        input_dict = {}
+        keys = data[0].keys()
+        for k in keys:
+            input_dict[k] = torch.cat([d[k].unsqueeze(0) for d in data], dim=0).to(
+                device
+            )
+
+        audio_embeds = self.audio_projection(
+            self.encode_audio(input_dict, device=device)["embedding"]
+        )
+        audio_embeds = F.normalize(audio_embeds, dim=-1)
+
+        return audio_embeds
+
+    def audio_infer(self, audio, hopsize=None, device=None):
+        """Forward one audio and produce the audio embedding
+
+        Parameters
+        ----------
+        audio:  (audio_length)
+            the time-domain audio input, notice that it must be only one input
+        hopsize: int
+            the overlap hopsize as the sliding window
+
+        Returns
+        ----------
+        output_dict: {
+            key: [n, (embedding_shape)] if "HTS-AT"
+            or
+            key: [(embedding_shape)] if "PANN"
+        }
+            the list of key values of the audio branch
+
+        """
+
+        assert not self.training, "the inference mode must be run at eval stage"
+        output_dict = {}
+        # PANN
+        if self.audio_cfg.model_type == "PANN":
+            audio_input = audio.unsqueeze(dim=0)
+            output_dict[key] = self.encode_audio(audio_input, device=device)[
+                key
+            ].squeeze(dim=0)
+        elif self.audio_cfg.model_type == "HTSAT":
+            # repeat
+            audio_len = len(audio)
+            k = self.audio_cfg.clip_samples // audio_len
+            if k > 1:
+                audio = audio.repeat(k)
+                audio_len = len(audio)
+
+            if hopsize is None:
+                hopsize = min(hopsize, audio_len)
+
+            if audio_len > self.audio_cfg.clip_samples:
+                audio_input = [
+                    audio[pos : pos + self.audio_cfg.clip_samples].clone()
+                    for pos in range(
+                        0, audio_len - self.audio_cfg.clip_samples, hopsize
+                    )
+                ]
+                audio_input.append(audio[-self.audio_cfg.clip_samples :].clone())
+                audio_input = torch.stack(audio_input)
+                output_dict[key] = self.encode_audio(audio_input, device=device)[key]
+            else:
+                audio_input = audio.unsqueeze(dim=0)
+                output_dict[key] = self.encode_audio(audio_input, device=device)[
+                    key
+                ].squeeze(dim=0)
+
+        return output_dict
+
+
+def convert_weights_to_fp16(model: nn.Module):
+    """Convert applicable model parameters to fp16"""
+
+    def _convert_weights_to_fp16(l):
+        if isinstance(l, (nn.Conv1d, nn.Conv2d, nn.Linear)):
+            l.weight.data = l.weight.data.half()
+            if l.bias is not None:
+                l.bias.data = l.bias.data.half()
+
+        if isinstance(l, nn.MultiheadAttention):
+            for attr in [
+                *[f"{s}_proj_weight" for s in ["in", "q", "k", "v"]],
+                "in_proj_bias",
+                "bias_k",
+                "bias_v",
+            ]:
+                tensor = getattr(l, attr)
+                if tensor is not None:
+                    tensor.data = tensor.data.half()
+
+        for name in ["text_projection", "proj"]:
+            if hasattr(l, name):
+                attr = getattr(l, name)
+                if attr is not None:
+                    attr.data = attr.data.half()
+
+    model.apply(_convert_weights_to_fp16)
+
+
+# Ignore the state dict of the vision part
+def build_model_from_openai_state_dict(
+    state_dict: dict, model_cfg, enable_fusion: bool = False, fusion_type: str = "None"
+):
+
+    embed_dim = model_cfg["embed_dim"]
+    audio_cfg = model_cfg["audio_cfg"]
+    text_cfg = model_cfg["text_cfg"]
+    context_length = state_dict["positional_embedding"].shape[0]
+    vocab_size = state_dict["token_embedding.weight"].shape[0]
+    transformer_width = state_dict["ln_final.weight"].shape[0]
+    transformer_heads = transformer_width // 64
+    transformer_layers = len(
+        set(
+            k.split(".")[2]
+            for k in state_dict
+            if k.startswith(f"transformer.resblocks")
+        )
+    )
+
+    audio_cfg = CLAPAudioCfp(**audio_cfg)
+    text_cfg = CLAPTextCfg(**text_cfg)
+
+    model = CLAP(
+        embed_dim,
+        audio_cfg=audio_cfg,
+        text_cfg=text_cfg,
+        quick_gelu=True,  # OpenAI models were trained with QuickGELU
+        enable_fusion=enable_fusion,
+        fusion_type=fusion_type,
+    )
+    state_dict["logit_scale_a"] = state_dict["logit_scale"]
+    state_dict["logit_scale_t"] = state_dict["logit_scale"]
+    pop_keys = list(state_dict.keys())[::]
+    # pop the visual branch saved weights
+    for key in pop_keys:
+        if key.startswith("visual."):
+            state_dict.pop(key, None)
+
+    for key in ["logit_scale", "input_resolution", "context_length", "vocab_size"]:
+        state_dict.pop(key, None)
+
+    # not use fp16
+    # convert_weights_to_fp16(model)
+    model.load_state_dict(state_dict, strict=False)
+    return model.eval()
+
+
+def trace_model(model, batch_size=256, device=torch.device("cpu")):
+    model.eval()
+    audio_length = model.audio_cfg.audio_length
+    example_audio = torch.ones((batch_size, audio_length), device=device)
+    example_text = torch.zeros(
+        (batch_size, model.context_length), dtype=torch.int, device=device
+    )
+    model = torch.jit.trace_module(
+        model,
+        inputs=dict(
+            forward=(example_audio, example_text),
+            encode_text=(example_text,),
+            encode_image=(example_audio,),
+        ),
+    )
+    model.audio_cfg.audio_length = audio_length  # Question: what does this do?
+    return model
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-base.json b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-base.json
new file mode 100644
index 0000000000000000000000000000000000000000..6cef625a89daf4431f1c9f72e10bc9640eef2ba8
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-base.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 1024,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "HTSAT",
+        "model_name": "base"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-large.json b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-large.json
new file mode 100644
index 0000000000000000000000000000000000000000..699cdb1b16855582606551e4196b24aba2ffd871
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-large.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "HTSAT",
+        "model_name": "large"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny-win-1536.json b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny-win-1536.json
new file mode 100644
index 0000000000000000000000000000000000000000..73e42990fe8361a0df502e7f93d29f19f58c9ecb
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny-win-1536.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 768,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1536,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "HTSAT",
+        "model_name": "tiny"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny.json b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny.json
new file mode 100644
index 0000000000000000000000000000000000000000..a6e7821163d9afa81c27345a1e472475b92af169
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/HTSAT-tiny.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 768,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "HTSAT",
+        "model_name": "tiny"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-10.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-10.json
new file mode 100644
index 0000000000000000000000000000000000000000..954ddf62921aed7dde9c37ffffec98a2e96a4ee7
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-10.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 1024,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn10"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-18k.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-18k.json
new file mode 100644
index 0000000000000000000000000000000000000000..b7989bc0cd95d0d39049b7524eba508b3e386439
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-18k.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 18000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn14"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-8k-20s.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-8k-20s.json
new file mode 100644
index 0000000000000000000000000000000000000000..56bdb56bedc304ffa52d8bf5988cea2c1d82d14e
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-fmax-8k-20s.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 960000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 360,
+        "fmin": 50,
+        "fmax": 8000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn14"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-tiny-transformer.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-tiny-transformer.json
new file mode 100644
index 0000000000000000000000000000000000000000..5756e3bebc97cc985f512cb081930fee4e49bec1
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-tiny-transformer.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn14"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 4
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-win-1536.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-win-1536.json
new file mode 100644
index 0000000000000000000000000000000000000000..5a9e7e208b661619d5e26625e849da1adda8a475
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14-win-1536.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1536,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn14"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14.json
new file mode 100644
index 0000000000000000000000000000000000000000..39a5134cde1d8c50f4758377c952ef22f07bab41
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-14.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 2048,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn14"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/PANN-6.json b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-6.json
new file mode 100644
index 0000000000000000000000000000000000000000..21ebc344326de260c386ba77e0ad63cf9b04febf
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/PANN-6.json
@@ -0,0 +1,23 @@
+{
+    "embed_dim": 512,
+    "audio_cfg": {
+        "audio_length": 1024,
+        "clip_samples": 480000,
+        "mel_bins": 64,
+        "sample_rate": 48000,
+        "window_size": 1024,
+        "hop_size": 480,
+        "fmin": 50,
+        "fmax": 14000,
+        "class_num": 527,
+        "model_type": "PANN",
+        "model_name": "Cnn6"
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN101-quickgelu.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN101-quickgelu.json
new file mode 100644
index 0000000000000000000000000000000000000000..d0db2c161d13138788c4609d373b023b8454d624
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN101-quickgelu.json
@@ -0,0 +1,22 @@
+{
+    "embed_dim": 512,
+    "quick_gelu": true,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": [
+            3,
+            4,
+            23,
+            3
+        ],
+        "width": 64,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN101.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN101.json
new file mode 100644
index 0000000000000000000000000000000000000000..b88b4d3acbaa701c614ab0ea65fc88fcfe289c32
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN101.json
@@ -0,0 +1,21 @@
+{
+    "embed_dim": 512,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": [
+            3,
+            4,
+            23,
+            3
+        ],
+        "width": 64,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN50-quickgelu.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN50-quickgelu.json
new file mode 100644
index 0000000000000000000000000000000000000000..8c2f91260cdeb043434dc1e893cce81d4ce7f0d1
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN50-quickgelu.json
@@ -0,0 +1,22 @@
+{
+    "embed_dim": 1024,
+    "quick_gelu": true,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": [
+            3,
+            4,
+            6,
+            3
+        ],
+        "width": 64,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN50.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN50.json
new file mode 100644
index 0000000000000000000000000000000000000000..33aa884d54fee0076c33676831e49d5e1ffcb8f2
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN50.json
@@ -0,0 +1,21 @@
+{
+    "embed_dim": 1024,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": [
+            3,
+            4,
+            6,
+            3
+        ],
+        "width": 64,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN50x16.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN50x16.json
new file mode 100644
index 0000000000000000000000000000000000000000..3161e1a2c9a839161e652a4d729c2cdc971161db
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN50x16.json
@@ -0,0 +1,21 @@
+{
+    "embed_dim": 768,
+    "vision_cfg": {
+        "image_size": 384,
+        "layers": [
+            6,
+            8,
+            18,
+            8
+        ],
+        "width": 96,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 768,
+        "heads": 12,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/RN50x4.json b/picoaudio/audioldm/clap/open_clip/model_configs/RN50x4.json
new file mode 100644
index 0000000000000000000000000000000000000000..e155237f8ce1026aaaeecc80751eabe6f329f0bb
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/RN50x4.json
@@ -0,0 +1,21 @@
+{
+    "embed_dim": 640,
+    "vision_cfg": {
+        "image_size": 288,
+        "layers": [
+            4,
+            6,
+            10,
+            6
+        ],
+        "width": 80,
+        "patch_size": null
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 640,
+        "heads": 10,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-16.json b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-16.json
new file mode 100644
index 0000000000000000000000000000000000000000..395eea77ec3907c0611531aba63459b193e67b9c
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-16.json
@@ -0,0 +1,16 @@
+{
+    "embed_dim": 512,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": 12,
+        "width": 768,
+        "patch_size": 16
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32-quickgelu.json b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32-quickgelu.json
new file mode 100644
index 0000000000000000000000000000000000000000..ce6bd923593293ed50dfcfb28b73ca7403bcf3c5
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32-quickgelu.json
@@ -0,0 +1,17 @@
+{
+    "embed_dim": 512,
+    "quick_gelu": true,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": 12,
+        "width": 768,
+        "patch_size": 32
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32.json b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32.json
new file mode 100644
index 0000000000000000000000000000000000000000..07c8e28eb06fa1813ba932fe4eec668262d1c47f
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-B-32.json
@@ -0,0 +1,16 @@
+{
+    "embed_dim": 512,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": 12,
+        "width": 768,
+        "patch_size": 32
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 512,
+        "heads": 8,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/model_configs/ViT-L-14.json b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-L-14.json
new file mode 100644
index 0000000000000000000000000000000000000000..d4a4bbb1dd4ed4edb317d3ace4f3ad13b211c241
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/model_configs/ViT-L-14.json
@@ -0,0 +1,16 @@
+{
+    "embed_dim": 768,
+    "vision_cfg": {
+        "image_size": 224,
+        "layers": 24,
+        "width": 1024,
+        "patch_size": 14
+    },
+    "text_cfg": {
+        "context_length": 77,
+        "vocab_size": 49408,
+        "width": 768,
+        "heads": 12,
+        "layers": 12
+    }
+}
\ No newline at end of file
diff --git a/picoaudio/audioldm/clap/open_clip/openai.py b/picoaudio/audioldm/clap/open_clip/openai.py
new file mode 100644
index 0000000000000000000000000000000000000000..fcb624f54a8b9d2c4b11e3adb50c53c3261716d4
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/openai.py
@@ -0,0 +1,159 @@
+""" OpenAI pretrained model functions
+
+Adapted from https://github.com/openai/CLIP. Originally MIT License, Copyright (c) 2021 OpenAI.
+"""
+
+import os
+import warnings
+from typing import Union, List
+
+import torch
+
+from .model import build_model_from_openai_state_dict
+from .pretrained import (
+    get_pretrained_url,
+    list_pretrained_tag_models,
+    download_pretrained,
+)
+
+__all__ = ["list_openai_models", "load_openai_model"]
+
+CACHE_DIR = os.getenv("AUDIOLDM_CACHE_DIR", "~/.cache")
+
+
+
+def list_openai_models() -> List[str]:
+    """Returns the names of available CLIP models"""
+    return list_pretrained_tag_models("openai")
+
+
+def load_openai_model(
+    name: str,
+    model_cfg,
+    device: Union[str, torch.device] = "cuda" if torch.cuda.is_available() else "cpu",
+    jit=True,
+    cache_dir=os.path.expanduser(f"{CACHE_DIR}/clip"),
+    enable_fusion: bool = False,
+    fusion_type: str = "None",
+):
+    """Load a CLIP model, preserve its text pretrained part, and set in the CLAP model
+
+    Parameters
+    ----------
+    name : str
+        A model name listed by `clip.available_models()`, or the path to a model checkpoint containing the state_dict
+    device : Union[str, torch.device]
+        The device to put the loaded model
+    jit : bool
+        Whether to load the optimized JIT model (default) or more hackable non-JIT model.
+
+    Returns
+    -------
+    model : torch.nn.Module
+        The CLAP model
+    preprocess : Callable[[PIL.Image], torch.Tensor]
+        A torchvision transform that converts a PIL image into a tensor that the returned model can take as its input
+    """
+    if get_pretrained_url(name, "openai"):
+        model_path = download_pretrained(
+            get_pretrained_url(name, "openai"), root=cache_dir
+        )
+    elif os.path.isfile(name):
+        model_path = name
+    else:
+        raise RuntimeError(
+            f"Model {name} not found; available models = {list_openai_models()}"
+        )
+
+    try:
+        # loading JIT archive
+        model = torch.jit.load(model_path, map_location=device if jit else "cpu").eval()
+        state_dict = None
+    except RuntimeError:
+        # loading saved state dict
+        if jit:
+            warnings.warn(
+                f"File {model_path} is not a JIT archive. Loading as a state dict instead"
+            )
+            jit = False
+        state_dict = torch.load(model_path, map_location="cpu")
+
+    if not jit:
+        try:
+            model = build_model_from_openai_state_dict(
+                state_dict or model.state_dict(), model_cfg, enable_fusion, fusion_type
+            ).to(device)
+        except KeyError:
+            sd = {k[7:]: v for k, v in state_dict["state_dict"].items()}
+            model = build_model_from_openai_state_dict(
+                sd, model_cfg, enable_fusion, fusion_type
+            ).to(device)
+
+        if str(device) == "cpu":
+            model.float()
+        return model
+
+    # patch the device names
+    device_holder = torch.jit.trace(
+        lambda: torch.ones([]).to(torch.device(device)), example_inputs=[]
+    )
+    device_node = [
+        n
+        for n in device_holder.graph.findAllNodes("prim::Constant")
+        if "Device" in repr(n)
+    ][-1]
+
+    def patch_device(module):
+        try:
+            graphs = [module.graph] if hasattr(module, "graph") else []
+        except RuntimeError:
+            graphs = []
+
+        if hasattr(module, "forward1"):
+            graphs.append(module.forward1.graph)
+
+        for graph in graphs:
+            for node in graph.findAllNodes("prim::Constant"):
+                if "value" in node.attributeNames() and str(node["value"]).startswith(
+                    "cuda"
+                ):
+                    node.copyAttributes(device_node)
+
+    model.apply(patch_device)
+    patch_device(model.encode_audio)
+    patch_device(model.encode_text)
+
+    # patch dtype to float32 on CPU
+    if str(device) == "cpu":
+        float_holder = torch.jit.trace(
+            lambda: torch.ones([]).float(), example_inputs=[]
+        )
+        float_input = list(float_holder.graph.findNode("aten::to").inputs())[1]
+        float_node = float_input.node()
+
+        def patch_float(module):
+            try:
+                graphs = [module.graph] if hasattr(module, "graph") else []
+            except RuntimeError:
+                graphs = []
+
+            if hasattr(module, "forward1"):
+                graphs.append(module.forward1.graph)
+
+            for graph in graphs:
+                for node in graph.findAllNodes("aten::to"):
+                    inputs = list(node.inputs())
+                    for i in [
+                        1,
+                        2,
+                    ]:  # dtype can be the second or third argument to aten::to()
+                        if inputs[i].node()["value"] == 5:
+                            inputs[i].node().copyAttributes(float_node)
+
+        model.apply(patch_float)
+        patch_float(model.encode_audio)
+        patch_float(model.encode_text)
+        model.float()
+
+    model.audio_branch.audio_length = model.audio_cfg.audio_length
+    return model
diff --git a/picoaudio/audioldm/clap/open_clip/pann_model.py b/picoaudio/audioldm/clap/open_clip/pann_model.py
new file mode 100644
index 0000000000000000000000000000000000000000..0d9a8eb0bf897ad6ec04923361b01e5de433b2ef
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/pann_model.py
@@ -0,0 +1,704 @@
+# PANNs: Large-Scale Pretrained Audio Neural Networks for Audio Pattern Recognition
+# Reference from https://github.com/qiuqiangkong/audioset_tagging_cnn
+# Some layers are re-designed for CLAP
+import os
+
+os.environ["NUMBA_CACHE_DIR"] = "/tmp/"
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torchlibrosa.stft import Spectrogram, LogmelFilterBank
+from torchlibrosa.augmentation import SpecAugmentation
+
+from .utils import do_mixup, interpolate, pad_framewise_output
+from .feature_fusion import iAFF, AFF, DAF
+
+
+def init_layer(layer):
+    """Initialize a Linear or Convolutional layer."""
+    nn.init.xavier_uniform_(layer.weight)
+
+    if hasattr(layer, "bias"):
+        if layer.bias is not None:
+            layer.bias.data.fill_(0.0)
+
+
+def init_bn(bn):
+    """Initialize a Batchnorm layer."""
+    bn.bias.data.fill_(0.0)
+    bn.weight.data.fill_(1.0)
+
+
+class ConvBlock(nn.Module):
+    def __init__(self, in_channels, out_channels):
+
+        super(ConvBlock, self).__init__()
+
+        self.conv1 = nn.Conv2d(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=(3, 3),
+            stride=(1, 1),
+            padding=(1, 1),
+            bias=False,
+        )
+
+        self.conv2 = nn.Conv2d(
+            in_channels=out_channels,
+            out_channels=out_channels,
+            kernel_size=(3, 3),
+            stride=(1, 1),
+            padding=(1, 1),
+            bias=False,
+        )
+
+        self.bn1 = nn.BatchNorm2d(out_channels)
+        self.bn2 = nn.BatchNorm2d(out_channels)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_layer(self.conv1)
+        init_layer(self.conv2)
+        init_bn(self.bn1)
+        init_bn(self.bn2)
+
+    def forward(self, input, pool_size=(2, 2), pool_type="avg"):
+
+        x = input
+        x = F.relu_(self.bn1(self.conv1(x)))
+        x = F.relu_(self.bn2(self.conv2(x)))
+        if pool_type == "max":
+            x = F.max_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg":
+            x = F.avg_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg+max":
+            x1 = F.avg_pool2d(x, kernel_size=pool_size)
+            x2 = F.max_pool2d(x, kernel_size=pool_size)
+            x = x1 + x2
+        else:
+            raise Exception("Incorrect argument!")
+
+        return x
+
+
+class ConvBlock5x5(nn.Module):
+    def __init__(self, in_channels, out_channels):
+
+        super(ConvBlock5x5, self).__init__()
+
+        self.conv1 = nn.Conv2d(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            kernel_size=(5, 5),
+            stride=(1, 1),
+            padding=(2, 2),
+            bias=False,
+        )
+
+        self.bn1 = nn.BatchNorm2d(out_channels)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_layer(self.conv1)
+        init_bn(self.bn1)
+
+    def forward(self, input, pool_size=(2, 2), pool_type="avg"):
+
+        x = input
+        x = F.relu_(self.bn1(self.conv1(x)))
+        if pool_type == "max":
+            x = F.max_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg":
+            x = F.avg_pool2d(x, kernel_size=pool_size)
+        elif pool_type == "avg+max":
+            x1 = F.avg_pool2d(x, kernel_size=pool_size)
+            x2 = F.max_pool2d(x, kernel_size=pool_size)
+            x = x1 + x2
+        else:
+            raise Exception("Incorrect argument!")
+
+        return x
+
+
+class AttBlock(nn.Module):
+    def __init__(self, n_in, n_out, activation="linear", temperature=1.0):
+        super(AttBlock, self).__init__()
+
+        self.activation = activation
+        self.temperature = temperature
+        self.att = nn.Conv1d(
+            in_channels=n_in,
+            out_channels=n_out,
+            kernel_size=1,
+            stride=1,
+            padding=0,
+            bias=True,
+        )
+        self.cla = nn.Conv1d(
+            in_channels=n_in,
+            out_channels=n_out,
+            kernel_size=1,
+            stride=1,
+            padding=0,
+            bias=True,
+        )
+
+        self.bn_att = nn.BatchNorm1d(n_out)
+        self.init_weights()
+
+    def init_weights(self):
+        init_layer(self.att)
+        init_layer(self.cla)
+        init_bn(self.bn_att)
+
+    def forward(self, x):
+        # x: (n_samples, n_in, n_time)
+        norm_att = torch.softmax(torch.clamp(self.att(x), -10, 10), dim=-1)
+        cla = self.nonlinear_transform(self.cla(x))
+        x = torch.sum(norm_att * cla, dim=2)
+        return x, norm_att, cla
+
+    def nonlinear_transform(self, x):
+        if self.activation == "linear":
+            return x
+        elif self.activation == "sigmoid":
+            return torch.sigmoid(x)
+
+
+class Cnn14(nn.Module):
+    def __init__(
+        self,
+        sample_rate,
+        window_size,
+        hop_size,
+        mel_bins,
+        fmin,
+        fmax,
+        classes_num,
+        enable_fusion=False,
+        fusion_type="None",
+    ):
+
+        super(Cnn14, self).__init__()
+
+        window = "hann"
+        center = True
+        pad_mode = "reflect"
+        ref = 1.0
+        amin = 1e-10
+        top_db = None
+
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+
+        # Spectrogram extractor
+        self.spectrogram_extractor = Spectrogram(
+            n_fft=window_size,
+            hop_length=hop_size,
+            win_length=window_size,
+            window=window,
+            center=center,
+            pad_mode=pad_mode,
+            freeze_parameters=True,
+        )
+
+        # Logmel feature extractor
+        self.logmel_extractor = LogmelFilterBank(
+            sr=sample_rate,
+            n_fft=window_size,
+            n_mels=mel_bins,
+            fmin=fmin,
+            fmax=fmax,
+            ref=ref,
+            amin=amin,
+            top_db=top_db,
+            freeze_parameters=True,
+        )
+
+        # Spec augmenter
+        self.spec_augmenter = SpecAugmentation(
+            time_drop_width=64,
+            time_stripes_num=2,
+            freq_drop_width=8,
+            freq_stripes_num=2,
+        )
+
+        self.bn0 = nn.BatchNorm2d(64)
+
+        if (self.enable_fusion) and (self.fusion_type == "channel_map"):
+            self.conv_block1 = ConvBlock(in_channels=4, out_channels=64)
+        else:
+            self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
+        self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
+        self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
+        self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
+        self.conv_block5 = ConvBlock(in_channels=512, out_channels=1024)
+        self.conv_block6 = ConvBlock(in_channels=1024, out_channels=2048)
+
+        self.fc1 = nn.Linear(2048, 2048, bias=True)
+        self.fc_audioset = nn.Linear(2048, classes_num, bias=True)
+
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_1d", "aff_1d", "iaff_1d"]
+        ):
+            self.mel_conv1d = nn.Sequential(
+                nn.Conv1d(64, 64, kernel_size=5, stride=3, padding=2),
+                nn.BatchNorm1d(64),  # No Relu
+            )
+            if self.fusion_type == "daf_1d":
+                self.fusion_model = DAF()
+            elif self.fusion_type == "aff_1d":
+                self.fusion_model = AFF(channels=64, type="1D")
+            elif self.fusion_type == "iaff_1d":
+                self.fusion_model = iAFF(channels=64, type="1D")
+
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d"]
+        ):
+            self.mel_conv2d = nn.Sequential(
+                nn.Conv2d(1, 64, kernel_size=(5, 5), stride=(6, 2), padding=(2, 2)),
+                nn.BatchNorm2d(64),
+                nn.ReLU(inplace=True),
+            )
+
+            if self.fusion_type == "daf_2d":
+                self.fusion_model = DAF()
+            elif self.fusion_type == "aff_2d":
+                self.fusion_model = AFF(channels=64, type="2D")
+            elif self.fusion_type == "iaff_2d":
+                self.fusion_model = iAFF(channels=64, type="2D")
+        self.init_weight()
+
+    def init_weight(self):
+        init_bn(self.bn0)
+        init_layer(self.fc1)
+        init_layer(self.fc_audioset)
+
+    def forward(self, input, mixup_lambda=None, device=None):
+        """
+        Input: (batch_size, data_length)"""
+
+        if self.enable_fusion and input["longer"].sum() == 0:
+            # if no audio is longer than 10s, then randomly select one audio to be longer
+            input["longer"][torch.randint(0, input["longer"].shape[0], (1,))] = True
+
+        if not self.enable_fusion:
+            x = self.spectrogram_extractor(
+                input["waveform"].to(device=device, non_blocking=True)
+            )  # (batch_size, 1, time_steps, freq_bins)
+            x = self.logmel_extractor(x)  # (batch_size, 1, time_steps, mel_bins)
+
+            x = x.transpose(1, 3)
+            x = self.bn0(x)
+            x = x.transpose(1, 3)
+        else:
+            longer_list = input["longer"].to(device=device, non_blocking=True)
+            x = input["mel_fusion"].to(device=device, non_blocking=True)
+            longer_list_idx = torch.where(longer_list)[0]
+            x = x.transpose(1, 3)
+            x = self.bn0(x)
+            x = x.transpose(1, 3)
+            if self.fusion_type in ["daf_1d", "aff_1d", "iaff_1d"]:
+                new_x = x[:, 0:1, :, :].clone().contiguous()
+                # local processing
+                if len(longer_list_idx) > 0:
+                    fusion_x_local = x[longer_list_idx, 1:, :, :].clone().contiguous()
+                    FB, FC, FT, FF = fusion_x_local.size()
+                    fusion_x_local = fusion_x_local.view(FB * FC, FT, FF)
+                    fusion_x_local = torch.permute(
+                        fusion_x_local, (0, 2, 1)
+                    ).contiguous()
+                    fusion_x_local = self.mel_conv1d(fusion_x_local)
+                    fusion_x_local = fusion_x_local.view(
+                        FB, FC, FF, fusion_x_local.size(-1)
+                    )
+                    fusion_x_local = (
+                        torch.permute(fusion_x_local, (0, 2, 1, 3))
+                        .contiguous()
+                        .flatten(2)
+                    )
+                    if fusion_x_local.size(-1) < FT:
+                        fusion_x_local = torch.cat(
+                            [
+                                fusion_x_local,
+                                torch.zeros(
+                                    (FB, FF, FT - fusion_x_local.size(-1)),
+                                    device=device,
+                                ),
+                            ],
+                            dim=-1,
+                        )
+                    else:
+                        fusion_x_local = fusion_x_local[:, :, :FT]
+                    # 1D fusion
+                    new_x = new_x.squeeze(1).permute((0, 2, 1)).contiguous()
+                    new_x[longer_list_idx] = self.fusion_model(
+                        new_x[longer_list_idx], fusion_x_local
+                    )
+                    x = new_x.permute((0, 2, 1)).contiguous()[:, None, :, :]
+                else:
+                    x = new_x
+            elif self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d", "channel_map"]:
+                x = x  # no change
+
+        if self.training:
+            x = self.spec_augmenter(x)
+        # Mixup on spectrogram
+        if self.training and mixup_lambda is not None:
+            x = do_mixup(x, mixup_lambda)
+        if (self.enable_fusion) and (
+            self.fusion_type in ["daf_2d", "aff_2d", "iaff_2d"]
+        ):
+            global_x = x[:, 0:1, :, :]
+
+            # global processing
+            B, C, H, W = global_x.shape
+            global_x = self.conv_block1(global_x, pool_size=(2, 2), pool_type="avg")
+            if len(longer_list_idx) > 0:
+                local_x = x[longer_list_idx, 1:, :, :].contiguous()
+                TH = global_x.size(-2)
+                # local processing
+                B, C, H, W = local_x.shape
+                local_x = local_x.view(B * C, 1, H, W)
+                local_x = self.mel_conv2d(local_x)
+                local_x = local_x.view(
+                    B, C, local_x.size(1), local_x.size(2), local_x.size(3)
+                )
+                local_x = local_x.permute((0, 2, 1, 3, 4)).contiguous().flatten(2, 3)
+                TB, TC, _, TW = local_x.size()
+                if local_x.size(-2) < TH:
+                    local_x = torch.cat(
+                        [
+                            local_x,
+                            torch.zeros(
+                                (TB, TC, TH - local_x.size(-2), TW),
+                                device=global_x.device,
+                            ),
+                        ],
+                        dim=-2,
+                    )
+                else:
+                    local_x = local_x[:, :, :TH, :]
+
+                global_x[longer_list_idx] = self.fusion_model(
+                    global_x[longer_list_idx], local_x
+                )
+            x = global_x
+        else:
+            x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
+
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block6(x, pool_size=(1, 1), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = torch.mean(x, dim=3)
+
+        latent_x1 = F.max_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x2 = F.avg_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x = latent_x1 + latent_x2
+        latent_x = latent_x.transpose(1, 2)
+        latent_x = F.relu_(self.fc1(latent_x))
+        latent_output = interpolate(latent_x, 32)
+
+        (x1, _) = torch.max(x, dim=2)
+        x2 = torch.mean(x, dim=2)
+        x = x1 + x2
+        x = F.dropout(x, p=0.5, training=self.training)
+        x = F.relu_(self.fc1(x))
+        embedding = F.dropout(x, p=0.5, training=self.training)
+        clipwise_output = torch.sigmoid(self.fc_audioset(x))
+
+        output_dict = {
+            "clipwise_output": clipwise_output,
+            "embedding": embedding,
+            "fine_grained_embedding": latent_output,
+        }
+        return output_dict
+
+
+class Cnn6(nn.Module):
+    def __init__(
+        self,
+        sample_rate,
+        window_size,
+        hop_size,
+        mel_bins,
+        fmin,
+        fmax,
+        classes_num,
+        enable_fusion=False,
+        fusion_type="None",
+    ):
+
+        super(Cnn6, self).__init__()
+
+        window = "hann"
+        center = True
+        pad_mode = "reflect"
+        ref = 1.0
+        amin = 1e-10
+        top_db = None
+
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+
+        # Spectrogram extractor
+        self.spectrogram_extractor = Spectrogram(
+            n_fft=window_size,
+            hop_length=hop_size,
+            win_length=window_size,
+            window=window,
+            center=center,
+            pad_mode=pad_mode,
+            freeze_parameters=True,
+        )
+
+        # Logmel feature extractor
+        self.logmel_extractor = LogmelFilterBank(
+            sr=sample_rate,
+            n_fft=window_size,
+            n_mels=mel_bins,
+            fmin=fmin,
+            fmax=fmax,
+            ref=ref,
+            amin=amin,
+            top_db=top_db,
+            freeze_parameters=True,
+        )
+
+        # Spec augmenter
+        self.spec_augmenter = SpecAugmentation(
+            time_drop_width=64,
+            time_stripes_num=2,
+            freq_drop_width=8,
+            freq_stripes_num=2,
+        )
+
+        self.bn0 = nn.BatchNorm2d(64)
+
+        self.conv_block1 = ConvBlock5x5(in_channels=1, out_channels=64)
+        self.conv_block2 = ConvBlock5x5(in_channels=64, out_channels=128)
+        self.conv_block3 = ConvBlock5x5(in_channels=128, out_channels=256)
+        self.conv_block4 = ConvBlock5x5(in_channels=256, out_channels=512)
+
+        self.fc1 = nn.Linear(512, 512, bias=True)
+        self.fc_audioset = nn.Linear(512, classes_num, bias=True)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_bn(self.bn0)
+        init_layer(self.fc1)
+        init_layer(self.fc_audioset)
+
+    def forward(self, input, mixup_lambda=None, device=None):
+        """
+        Input: (batch_size, data_length)"""
+
+        x = self.spectrogram_extractor(input)  # (batch_size, 1, time_steps, freq_bins)
+        x = self.logmel_extractor(x)  # (batch_size, 1, time_steps, mel_bins)
+
+        x = x.transpose(1, 3)
+        x = self.bn0(x)
+        x = x.transpose(1, 3)
+
+        if self.training:
+            x = self.spec_augmenter(x)
+
+        # Mixup on spectrogram
+        if self.training and mixup_lambda is not None:
+            x = do_mixup(x, mixup_lambda)
+
+        x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = torch.mean(x, dim=3)
+
+        latent_x1 = F.max_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x2 = F.avg_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x = latent_x1 + latent_x2
+        latent_x = latent_x.transpose(1, 2)
+        latent_x = F.relu_(self.fc1(latent_x))
+        latent_output = interpolate(latent_x, 16)
+
+        (x1, _) = torch.max(x, dim=2)
+        x2 = torch.mean(x, dim=2)
+        x = x1 + x2
+        x = F.dropout(x, p=0.5, training=self.training)
+        x = F.relu_(self.fc1(x))
+        embedding = F.dropout(x, p=0.5, training=self.training)
+        clipwise_output = torch.sigmoid(self.fc_audioset(x))
+
+        output_dict = {
+            "clipwise_output": clipwise_output,
+            "embedding": embedding,
+            "fine_grained_embedding": latent_output,
+        }
+
+        return output_dict
+
+
+class Cnn10(nn.Module):
+    def __init__(
+        self,
+        sample_rate,
+        window_size,
+        hop_size,
+        mel_bins,
+        fmin,
+        fmax,
+        classes_num,
+        enable_fusion=False,
+        fusion_type="None",
+    ):
+
+        super(Cnn10, self).__init__()
+
+        window = "hann"
+        center = True
+        pad_mode = "reflect"
+        ref = 1.0
+        amin = 1e-10
+        top_db = None
+
+        self.enable_fusion = enable_fusion
+        self.fusion_type = fusion_type
+
+        # Spectrogram extractor
+        self.spectrogram_extractor = Spectrogram(
+            n_fft=window_size,
+            hop_length=hop_size,
+            win_length=window_size,
+            window=window,
+            center=center,
+            pad_mode=pad_mode,
+            freeze_parameters=True,
+        )
+
+        # Logmel feature extractor
+        self.logmel_extractor = LogmelFilterBank(
+            sr=sample_rate,
+            n_fft=window_size,
+            n_mels=mel_bins,
+            fmin=fmin,
+            fmax=fmax,
+            ref=ref,
+            amin=amin,
+            top_db=top_db,
+            freeze_parameters=True,
+        )
+
+        # Spec augmenter
+        self.spec_augmenter = SpecAugmentation(
+            time_drop_width=64,
+            time_stripes_num=2,
+            freq_drop_width=8,
+            freq_stripes_num=2,
+        )
+
+        self.bn0 = nn.BatchNorm2d(64)
+
+        self.conv_block1 = ConvBlock(in_channels=1, out_channels=64)
+        self.conv_block2 = ConvBlock(in_channels=64, out_channels=128)
+        self.conv_block3 = ConvBlock(in_channels=128, out_channels=256)
+        self.conv_block4 = ConvBlock(in_channels=256, out_channels=512)
+        self.conv_block5 = ConvBlock(in_channels=512, out_channels=1024)
+
+        self.fc1 = nn.Linear(1024, 1024, bias=True)
+        self.fc_audioset = nn.Linear(1024, classes_num, bias=True)
+
+        self.init_weight()
+
+    def init_weight(self):
+        init_bn(self.bn0)
+        init_layer(self.fc1)
+        init_layer(self.fc_audioset)
+
+    def forward(self, input, mixup_lambda=None, device=None):
+        """
+        Input: (batch_size, data_length)"""
+
+        x = self.spectrogram_extractor(input)  # (batch_size, 1, time_steps, freq_bins)
+        x = self.logmel_extractor(x)  # (batch_size, 1, time_steps, mel_bins)
+
+        x = x.transpose(1, 3)
+        x = self.bn0(x)
+        x = x.transpose(1, 3)
+
+        if self.training:
+            x = self.spec_augmenter(x)
+
+        # Mixup on spectrogram
+        if self.training and mixup_lambda is not None:
+            x = do_mixup(x, mixup_lambda)
+
+        x = self.conv_block1(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block2(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block3(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block4(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = self.conv_block5(x, pool_size=(2, 2), pool_type="avg")
+        x = F.dropout(x, p=0.2, training=self.training)
+        x = torch.mean(x, dim=3)
+
+        latent_x1 = F.max_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x2 = F.avg_pool1d(x, kernel_size=3, stride=1, padding=1)
+        latent_x = latent_x1 + latent_x2
+        latent_x = latent_x.transpose(1, 2)
+        latent_x = F.relu_(self.fc1(latent_x))
+        latent_output = interpolate(latent_x, 32)
+
+        (x1, _) = torch.max(x, dim=2)
+        x2 = torch.mean(x, dim=2)
+        x = x1 + x2
+        x = F.dropout(x, p=0.5, training=self.training)
+        x = F.relu_(self.fc1(x))
+        embedding = F.dropout(x, p=0.5, training=self.training)
+        clipwise_output = torch.sigmoid(self.fc_audioset(x))
+
+        output_dict = {
+            "clipwise_output": clipwise_output,
+            "embedding": embedding,
+            "fine_grained_embedding": latent_output,
+        }
+
+        return output_dict
+
+
+def create_pann_model(audio_cfg, enable_fusion=False, fusion_type="None"):
+    try:
+        ModelProto = eval(audio_cfg.model_name)
+        model = ModelProto(
+            sample_rate=audio_cfg.sample_rate,
+            window_size=audio_cfg.window_size,
+            hop_size=audio_cfg.hop_size,
+            mel_bins=audio_cfg.mel_bins,
+            fmin=audio_cfg.fmin,
+            fmax=audio_cfg.fmax,
+            classes_num=audio_cfg.class_num,
+            enable_fusion=enable_fusion,
+            fusion_type=fusion_type,
+        )
+        return model
+    except:
+        raise RuntimeError(
+            f"Import Model for {audio_cfg.model_name} not found, or the audio cfg parameters are not enough."
+        )
diff --git a/picoaudio/audioldm/clap/open_clip/pretrained.py b/picoaudio/audioldm/clap/open_clip/pretrained.py
new file mode 100644
index 0000000000000000000000000000000000000000..8ed2ae1732a28c4e98d1f3412157ef27054e41dc
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/pretrained.py
@@ -0,0 +1,169 @@
+import hashlib
+import os
+import urllib
+import warnings
+
+from tqdm import tqdm
+
+CACHE_DIR = os.getenv("AUDIOLDM_CACHE_DIR", "~/.cache")
+
+_RN50 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/afeb0e10f9e5a86da6080e35cf09123aca3b358a0c3e3b6c78a7b63bc04b6762/RN50.pt",
+    yfcc15m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn50-quickgelu-yfcc15m-455df137.pt",
+    cc12m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn50-quickgelu-cc12m-f000538c.pt",
+)
+
+_RN50_quickgelu = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/afeb0e10f9e5a86da6080e35cf09123aca3b358a0c3e3b6c78a7b63bc04b6762/RN50.pt",
+    yfcc15m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn50-quickgelu-yfcc15m-455df137.pt",
+    cc12m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn50-quickgelu-cc12m-f000538c.pt",
+)
+
+_RN101 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/8fa8567bab74a42d41c5915025a8e4538c3bdbe8804a470a72f30b0d94fab599/RN101.pt",
+    yfcc15m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn101-quickgelu-yfcc15m-3e04b30e.pt",
+)
+
+_RN101_quickgelu = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/8fa8567bab74a42d41c5915025a8e4538c3bdbe8804a470a72f30b0d94fab599/RN101.pt",
+    yfcc15m="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/rn101-quickgelu-yfcc15m-3e04b30e.pt",
+)
+
+_RN50x4 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/7e526bd135e493cef0776de27d5f42653e6b4c8bf9e0f653bb11773263205fdd/RN50x4.pt",
+)
+
+_RN50x16 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/52378b407f34354e150460fe41077663dd5b39c54cd0bfd2b27167a4a06ec9aa/RN50x16.pt",
+)
+
+_RN50x64 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/be1cfb55d75a9666199fb2206c106743da0f6468c9d327f3e0d0a543a9919d9c/RN50x64.pt",
+)
+
+_VITB32 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/40d365715913c9da98579312b702a82c18be219cc2a73407c4526f58eba950af/ViT-B-32.pt",
+    laion400m_e31="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_e31-d867053b.pt",
+    laion400m_e32="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_e32-46683a32.pt",
+    laion400m_avg="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_avg-8a00ab3c.pt",
+)
+
+_VITB32_quickgelu = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/40d365715913c9da98579312b702a82c18be219cc2a73407c4526f58eba950af/ViT-B-32.pt",
+    laion400m_e31="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_e31-d867053b.pt",
+    laion400m_e32="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_e32-46683a32.pt",
+    laion400m_avg="https://github.com/mlfoundations/open_clip/releases/download/v0.2-weights/vit_b_32-quickgelu-laion400m_avg-8a00ab3c.pt",
+)
+
+_VITB16 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/5806e77cd80f8b59890b7e101eabd078d9fb84e6937f9e85e4ecb61988df416f/ViT-B-16.pt",
+)
+
+_VITL14 = dict(
+    openai="https://openaipublic.azureedge.net/clip/models/b8cca3fd41ae0c99ba7e8951adf17d267cdb84cd88be6f7c2e0eca1737a03836/ViT-L-14.pt",
+)
+
+_PRETRAINED = {
+    "RN50": _RN50,
+    "RN50-quickgelu": _RN50_quickgelu,
+    "RN101": _RN101,
+    "RN101-quickgelu": _RN101_quickgelu,
+    "RN50x4": _RN50x4,
+    "RN50x16": _RN50x16,
+    "ViT-B-32": _VITB32,
+    "ViT-B-32-quickgelu": _VITB32_quickgelu,
+    "ViT-B-16": _VITB16,
+    "ViT-L-14": _VITL14,
+}
+
+
+def list_pretrained(as_str: bool = False):
+    """returns list of pretrained models
+    Returns a tuple (model_name, pretrain_tag) by default or 'name:tag' if as_str == True
+    """
+    return [
+        ":".join([k, t]) if as_str else (k, t)
+        for k in _PRETRAINED.keys()
+        for t in _PRETRAINED[k].keys()
+    ]
+
+
+def list_pretrained_tag_models(tag: str):
+    """return all models having the specified pretrain tag"""
+    models = []
+    for k in _PRETRAINED.keys():
+        if tag in _PRETRAINED[k]:
+            models.append(k)
+    return models
+
+
+def list_pretrained_model_tags(model: str):
+    """return all pretrain tags for the specified model architecture"""
+    tags = []
+    if model in _PRETRAINED:
+        tags.extend(_PRETRAINED[model].keys())
+    return tags
+
+
+def get_pretrained_url(model: str, tag: str):
+    if model not in _PRETRAINED:
+        return ""
+    model_pretrained = _PRETRAINED[model]
+    if tag not in model_pretrained:
+        return ""
+    return model_pretrained[tag]
+
+
+def download_pretrained(url: str, root: str = os.path.expanduser(f"{CACHE_DIR}/clip")):
+    os.makedirs(root, exist_ok=True)
+    filename = os.path.basename(url)
+
+    if "openaipublic" in url:
+        expected_sha256 = url.split("/")[-2]
+    else:
+        expected_sha256 = ""
+
+    download_target = os.path.join(root, filename)
+
+    if os.path.exists(download_target) and not os.path.isfile(download_target):
+        raise RuntimeError(f"{download_target} exists and is not a regular file")
+
+    if os.path.isfile(download_target):
+        if expected_sha256:
+            if (
+                hashlib.sha256(open(download_target, "rb").read()).hexdigest()
+                == expected_sha256
+            ):
+                return download_target
+            else:
+                warnings.warn(
+                    f"{download_target} exists, but the SHA256 checksum does not match; re-downloading the file"
+                )
+        else:
+            return download_target
+
+    with urllib.request.urlopen(url) as source, open(download_target, "wb") as output:
+        with tqdm(
+            total=int(source.info().get("Content-Length")),
+            ncols=80,
+            unit="iB",
+            unit_scale=True,
+        ) as loop:
+            while True:
+                buffer = source.read(8192)
+                if not buffer:
+                    break
+
+                output.write(buffer)
+                loop.update(len(buffer))
+
+    if (
+        expected_sha256
+        and hashlib.sha256(open(download_target, "rb").read()).hexdigest()
+        != expected_sha256
+    ):
+        raise RuntimeError(
+            f"Model has been downloaded but the SHA256 checksum does not not match"
+        )
+
+    return download_target
diff --git a/picoaudio/audioldm/clap/open_clip/timm_model.py b/picoaudio/audioldm/clap/open_clip/timm_model.py
new file mode 100644
index 0000000000000000000000000000000000000000..c9d1ab4666b5bab5038d44b90c9ddca5087de460
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/timm_model.py
@@ -0,0 +1,112 @@
+""" timm model adapter
+
+Wraps timm (https://github.com/rwightman/pytorch-image-models) models for use as a vision tower in CLIP model.
+"""
+from collections import OrderedDict
+
+import torch.nn as nn
+
+try:
+    import timm
+    from timm.models.layers import Mlp, to_2tuple
+    from timm.models.layers.attention_pool2d import RotAttentionPool2d
+    from timm.models.layers.attention_pool2d import (
+        AttentionPool2d as AbsAttentionPool2d,
+    )
+except ImportError as e:
+    timm = None
+
+from .utils import freeze_batch_norm_2d
+
+
+class TimmModel(nn.Module):
+    """timm model adapter
+    # FIXME this adapter is a work in progress, may change in ways that break weight compat
+    """
+
+    def __init__(
+        self,
+        model_name,
+        embed_dim,
+        image_size=224,
+        pool="avg",
+        proj="linear",
+        drop=0.0,
+        pretrained=False,
+    ):
+        super().__init__()
+        if timm is None:
+            raise RuntimeError("Please `pip install timm` to use timm models.")
+
+        self.image_size = to_2tuple(image_size)
+        self.trunk = timm.create_model(model_name, pretrained=pretrained)
+        feat_size = self.trunk.default_cfg.get("pool_size", None)
+        feature_ndim = 1 if not feat_size else 2
+        if pool in ("abs_attn", "rot_attn"):
+            assert feature_ndim == 2
+            # if attn pooling used, remove both classifier and default pool
+            self.trunk.reset_classifier(0, global_pool="")
+        else:
+            # reset global pool if pool config set, otherwise leave as network default
+            reset_kwargs = dict(global_pool=pool) if pool else {}
+            self.trunk.reset_classifier(0, **reset_kwargs)
+        prev_chs = self.trunk.num_features
+
+        head_layers = OrderedDict()
+        if pool == "abs_attn":
+            head_layers["pool"] = AbsAttentionPool2d(
+                prev_chs, feat_size=feat_size, out_features=embed_dim
+            )
+            prev_chs = embed_dim
+        elif pool == "rot_attn":
+            head_layers["pool"] = RotAttentionPool2d(prev_chs, out_features=embed_dim)
+            prev_chs = embed_dim
+        else:
+            assert proj, "projection layer needed if non-attention pooling is used."
+
+        # NOTE attention pool ends with a projection layer, so proj should usually be set to '' if such pooling is used
+        if proj == "linear":
+            head_layers["drop"] = nn.Dropout(drop)
+            head_layers["proj"] = nn.Linear(prev_chs, embed_dim)
+        elif proj == "mlp":
+            head_layers["mlp"] = Mlp(prev_chs, 2 * embed_dim, embed_dim, drop=drop)
+
+        self.head = nn.Sequential(head_layers)
+
+    def lock(self, unlocked_groups=0, freeze_bn_stats=False):
+        """lock modules
+        Args:
+            unlocked_groups (int): leave last n layer groups unlocked (default: 0)
+        """
+        if not unlocked_groups:
+            # lock full model
+            for param in self.trunk.parameters():
+                param.requires_grad = False
+            if freeze_bn_stats:
+                freeze_batch_norm_2d(self.trunk)
+        else:
+            # NOTE: partial freeze requires latest timm (master) branch and is subject to change
+            try:
+                # FIXME import here until API stable and in an official release
+                from timm.models.helpers import group_parameters, group_modules
+            except ImportError:
+                raise RuntimeError(
+                    "Please install latest timm `pip install git+https://github.com/rwightman/pytorch-image-models`"
+                )
+            matcher = self.trunk.group_matcher()
+            gparams = group_parameters(self.trunk, matcher)
+            max_layer_id = max(gparams.keys())
+            max_layer_id = max_layer_id - unlocked_groups
+            for group_idx in range(max_layer_id + 1):
+                group = gparams[group_idx]
+                for param in group:
+                    self.trunk.get_parameter(param).requires_grad = False
+            if freeze_bn_stats:
+                gmodules = group_modules(self.trunk, matcher, reverse=True)
+                gmodules = {k for k, v in gmodules.items() if v <= max_layer_id}
+                freeze_batch_norm_2d(self.trunk, gmodules)
+
+    def forward(self, x):
+        x = self.trunk(x)
+        x = self.head(x)
+        return x
diff --git a/picoaudio/audioldm/clap/open_clip/tokenizer.py b/picoaudio/audioldm/clap/open_clip/tokenizer.py
new file mode 100644
index 0000000000000000000000000000000000000000..ee4d28450ec5dd12a79daf38cf3088e9e73c2cd5
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/tokenizer.py
@@ -0,0 +1,197 @@
+""" CLIP tokenizer
+
+Copied from https://github.com/openai/CLIP. Originally MIT License, Copyright (c) 2021 OpenAI.
+"""
+import gzip
+import html
+import os
+from functools import lru_cache
+from typing import Union, List
+
+import ftfy
+import regex as re
+import torch
+
+
+@lru_cache()
+def default_bpe():
+    return os.path.join(
+        os.path.dirname(os.path.abspath(__file__)), "bpe_simple_vocab_16e6.txt.gz"
+    )
+
+
+@lru_cache()
+def bytes_to_unicode():
+    """
+    Returns list of utf-8 byte and a corresponding list of unicode strings.
+    The reversible bpe codes work on unicode strings.
+    This means you need a large # of unicode characters in your vocab if you want to avoid UNKs.
+    When you're at something like a 10B token dataset you end up needing around 5K for decent coverage.
+    This is a signficant percentage of your normal, say, 32K bpe vocab.
+    To avoid that, we want lookup tables between utf-8 bytes and unicode strings.
+    And avoids mapping to whitespace/control characters the bpe code barfs on.
+    """
+    bs = (
+        list(range(ord("!"), ord("~") + 1))
+        + list(range(ord("¡"), ord("¬") + 1))
+        + list(range(ord("®"), ord("ÿ") + 1))
+    )
+    cs = bs[:]
+    n = 0
+    for b in range(2**8):
+        if b not in bs:
+            bs.append(b)
+            cs.append(2**8 + n)
+            n += 1
+    cs = [chr(n) for n in cs]
+    return dict(zip(bs, cs))
+
+
+def get_pairs(word):
+    """Return set of symbol pairs in a word.
+    Word is represented as tuple of symbols (symbols being variable-length strings).
+    """
+    pairs = set()
+    prev_char = word[0]
+    for char in word[1:]:
+        pairs.add((prev_char, char))
+        prev_char = char
+    return pairs
+
+
+def basic_clean(text):
+    text = ftfy.fix_text(text)
+    text = html.unescape(html.unescape(text))
+    return text.strip()
+
+
+def whitespace_clean(text):
+    text = re.sub(r"\s+", " ", text)
+    text = text.strip()
+    return text
+
+
+class SimpleTokenizer(object):
+    def __init__(self, bpe_path: str = default_bpe(), special_tokens=None):
+        self.byte_encoder = bytes_to_unicode()
+        self.byte_decoder = {v: k for k, v in self.byte_encoder.items()}
+        merges = gzip.open(bpe_path).read().decode("utf-8").split("\n")
+        merges = merges[1 : 49152 - 256 - 2 + 1]
+        merges = [tuple(merge.split()) for merge in merges]
+        vocab = list(bytes_to_unicode().values())
+        vocab = vocab + [v + "</w>" for v in vocab]
+        for merge in merges:
+            vocab.append("".join(merge))
+        if not special_tokens:
+            special_tokens = ["<start_of_text>", "<end_of_text>"]
+        else:
+            special_tokens = ["<start_of_text>", "<end_of_text>"] + special_tokens
+        vocab.extend(special_tokens)
+        self.encoder = dict(zip(vocab, range(len(vocab))))
+        self.decoder = {v: k for k, v in self.encoder.items()}
+        self.bpe_ranks = dict(zip(merges, range(len(merges))))
+        self.cache = {t: t for t in special_tokens}
+        special = "|".join(special_tokens)
+        self.pat = re.compile(
+            special + r"""|'s|'t|'re|'ve|'m|'ll|'d|[\p{L}]+|[\p{N}]|[^\s\p{L}\p{N}]+""",
+            re.IGNORECASE,
+        )
+
+        self.vocab_size = len(self.encoder)
+        self.all_special_ids = [self.encoder[t] for t in special_tokens]
+
+    def bpe(self, token):
+        if token in self.cache:
+            return self.cache[token]
+        word = tuple(token[:-1]) + (token[-1] + "</w>",)
+        pairs = get_pairs(word)
+
+        if not pairs:
+            return token + "</w>"
+
+        while True:
+            bigram = min(pairs, key=lambda pair: self.bpe_ranks.get(pair, float("inf")))
+            if bigram not in self.bpe_ranks:
+                break
+            first, second = bigram
+            new_word = []
+            i = 0
+            while i < len(word):
+                try:
+                    j = word.index(first, i)
+                    new_word.extend(word[i:j])
+                    i = j
+                except:
+                    new_word.extend(word[i:])
+                    break
+
+                if word[i] == first and i < len(word) - 1 and word[i + 1] == second:
+                    new_word.append(first + second)
+                    i += 2
+                else:
+                    new_word.append(word[i])
+                    i += 1
+            new_word = tuple(new_word)
+            word = new_word
+            if len(word) == 1:
+                break
+            else:
+                pairs = get_pairs(word)
+        word = " ".join(word)
+        self.cache[token] = word
+        return word
+
+    def encode(self, text):
+        bpe_tokens = []
+        text = whitespace_clean(basic_clean(text)).lower()
+        for token in re.findall(self.pat, text):
+            token = "".join(self.byte_encoder[b] for b in token.encode("utf-8"))
+            bpe_tokens.extend(
+                self.encoder[bpe_token] for bpe_token in self.bpe(token).split(" ")
+            )
+        return bpe_tokens
+
+    def decode(self, tokens):
+        text = "".join([self.decoder[token] for token in tokens])
+        text = (
+            bytearray([self.byte_decoder[c] for c in text])
+            .decode("utf-8", errors="replace")
+            .replace("</w>", " ")
+        )
+        return text
+
+
+_tokenizer = SimpleTokenizer()
+
+
+def tokenize(
+    texts: Union[str, List[str]], context_length: int = 77
+) -> torch.LongTensor:
+    """
+    Returns the tokenized representation of given input string(s)
+
+    Parameters
+    ----------
+    texts : Union[str, List[str]]
+        An input string or a list of input strings to tokenize
+    context_length : int
+        The context length to use; all CLIP models use 77 as the context length
+
+    Returns
+    -------
+    A two-dimensional tensor containing the resulting tokens, shape = [number of input strings, context_length]
+    """
+    if isinstance(texts, str):
+        texts = [texts]
+
+    sot_token = _tokenizer.encoder["<start_of_text>"]
+    eot_token = _tokenizer.encoder["<end_of_text>"]
+    all_tokens = [[sot_token] + _tokenizer.encode(text) + [eot_token] for text in texts]
+    result = torch.zeros(len(all_tokens), context_length, dtype=torch.long)
+
+    for i, tokens in enumerate(all_tokens):
+        if len(tokens) > context_length:
+            tokens = tokens[:context_length]  # Truncate
+        result[i, : len(tokens)] = torch.tensor(tokens)
+
+    return result
diff --git a/picoaudio/audioldm/clap/open_clip/transform.py b/picoaudio/audioldm/clap/open_clip/transform.py
new file mode 100644
index 0000000000000000000000000000000000000000..77aaa722c4a5544ac50de6df35d3e922f63b111d
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/transform.py
@@ -0,0 +1,45 @@
+from torchvision.transforms import (
+    Normalize,
+    Compose,
+    RandomResizedCrop,
+    InterpolationMode,
+    ToTensor,
+    Resize,
+    CenterCrop,
+)
+
+
+def _convert_to_rgb(image):
+    return image.convert("RGB")
+
+
+def image_transform(
+    image_size: int,
+    is_train: bool,
+    mean=(0.48145466, 0.4578275, 0.40821073),
+    std=(0.26862954, 0.26130258, 0.27577711),
+):
+    normalize = Normalize(mean=mean, std=std)
+    if is_train:
+        return Compose(
+            [
+                RandomResizedCrop(
+                    image_size,
+                    scale=(0.9, 1.0),
+                    interpolation=InterpolationMode.BICUBIC,
+                ),
+                _convert_to_rgb,
+                ToTensor(),
+                normalize,
+            ]
+        )
+    else:
+        return Compose(
+            [
+                Resize(image_size, interpolation=InterpolationMode.BICUBIC),
+                CenterCrop(image_size),
+                _convert_to_rgb,
+                ToTensor(),
+                normalize,
+            ]
+        )
diff --git a/picoaudio/audioldm/clap/open_clip/utils.py b/picoaudio/audioldm/clap/open_clip/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..34ecbced4cb7e6b6f92154a666e2c7efc7c922c6
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/utils.py
@@ -0,0 +1,362 @@
+import numpy as np
+import torch
+from torch import nn as nn
+from torchvision.ops.misc import FrozenBatchNorm2d
+import logging
+
+# import h5py
+from tqdm import tqdm
+import random
+import json
+import os
+import pathlib
+
+# TODO: (yusong) this not a good place to store those information and does not scale. Need to be fixed later.
+dataset_split = {
+    "audiocaps": ["train", "valid", "test"],
+    "audioset": ["balanced_train", "unbalanced_train", "eval"],
+    "BBCSoundEffects": ["train", "test"],
+    "Clotho": ["train", "test", "valid"],
+    "free_to_use_sounds": ["train", "test"],
+    "paramount_motion": ["train", "test"],
+    "sonniss_game_effects": ["train", "test"],
+    "wesoundeffects": ["train", "test"],
+    "MACS": ["train", "test"],
+    "freesound": ["train", "test"],
+    "FSD50K": ["train", "test", "valid"],
+    "fsd50k_class_label": ["train", "test", "valid"],
+    "esc50": ["train", "test"],
+    "audiostock": ["train", "test"],
+    "freesound_no_overlap_noesc50": ["train", "test"],
+    "epidemic_sound_effects": ["train", "test"],
+    "VGGSound": ["train", "test"],
+    "urbansound8k_class_label": ["train", "test"],
+    "audioset_t5": ["balanced_train", "unbalanced_train", "eval"],
+    "epidemic_sound_effects_t5": ["train", "test"],
+    "WavText5K": ["train", "test"],
+    "esc50_no_overlap": ["train", "test"],
+    "usd8k_no_overlap": ["train", "test"],
+    "fsd50k_200_class_label": ["train", "test", "valid"],
+}
+
+
+def freeze_batch_norm_2d(module, module_match={}, name=""):
+    """
+    Converts all `BatchNorm2d` and `SyncBatchNorm` layers of provided module into `FrozenBatchNorm2d`. If `module` is
+    itself an instance of either `BatchNorm2d` or `SyncBatchNorm`, it is converted into `FrozenBatchNorm2d` and
+    returned. Otherwise, the module is walked recursively and submodules are converted in place.
+
+    Args:
+        module (torch.nn.Module): Any PyTorch module.
+        module_match (dict): Dictionary of full module names to freeze (all if empty)
+        name (str): Full module name (prefix)
+
+    Returns:
+        torch.nn.Module: Resulting module
+
+    Inspired by https://github.com/pytorch/pytorch/blob/a5895f85be0f10212791145bfedc0261d364f103/torch/nn/modules/batchnorm.py#L762
+    """
+    res = module
+    is_match = True
+    if module_match:
+        is_match = name in module_match
+    if is_match and isinstance(
+        module, (nn.modules.batchnorm.BatchNorm2d, nn.modules.batchnorm.SyncBatchNorm)
+    ):
+        res = FrozenBatchNorm2d(module.num_features)
+        res.num_features = module.num_features
+        res.affine = module.affine
+        if module.affine:
+            res.weight.data = module.weight.data.clone().detach()
+            res.bias.data = module.bias.data.clone().detach()
+        res.running_mean.data = module.running_mean.data
+        res.running_var.data = module.running_var.data
+        res.eps = module.eps
+    else:
+        for child_name, child in module.named_children():
+            full_child_name = ".".join([name, child_name]) if name else child_name
+            new_child = freeze_batch_norm_2d(child, module_match, full_child_name)
+            if new_child is not child:
+                res.add_module(child_name, new_child)
+    return res
+
+
+def exist(dataset_name, dataset_type):
+    """
+    Check if dataset exists
+    """
+    if dataset_type in dataset_split[dataset_name]:
+        return True
+    else:
+        return False
+
+
+def get_tar_path_from_dataset_name(
+    dataset_names, dataset_types, islocal, dataset_path, proportion=1, full_dataset=None
+):
+    """
+    Get tar path from dataset name and type
+    """
+    output = []
+    for n in dataset_names:
+        if full_dataset is not None and n in full_dataset:
+            current_dataset_types = dataset_split[n]
+        else:
+            current_dataset_types = dataset_types
+        for s in current_dataset_types:
+            tmp = []
+            if islocal:
+                sizefilepath_ = f"{dataset_path}/{n}/{s}/sizes.json"
+                if not os.path.exists(sizefilepath_):
+                    sizefilepath_ = f"./json_files/{n}/{s}/sizes.json"
+            else:
+                sizefilepath_ = f"./json_files/{n}/{s}/sizes.json"
+            if not os.path.exists(sizefilepath_):
+                continue
+            sizes = json.load(open(sizefilepath_, "r"))
+            for k in sizes.keys():
+                if islocal:
+                    tmp.append(f"{dataset_path}/{n}/{s}/{k}")
+                else:
+                    tmp.append(
+                        f"pipe:aws s3 --cli-connect-timeout 0 cp s3://s-laion-audio/webdataset_tar/{n}/{s}/{k} -"
+                    )
+            if proportion != 1:
+                tmp = random.sample(tmp, int(proportion * len(tmp)))
+            output.append(tmp)
+    return sum(output, [])
+
+
+def get_tar_path_from_txts(txt_path, islocal, proportion=1):
+    """
+    Get tar path from txt path
+    """
+    if isinstance(txt_path, (list, tuple)):
+        return sum(
+            [
+                get_tar_path_from_txts(
+                    txt_path[i], islocal=islocal, proportion=proportion
+                )
+                for i in range(len(txt_path))
+            ],
+            [],
+        )
+    if isinstance(txt_path, str):
+        with open(txt_path) as f:
+            lines = f.readlines()
+        if islocal:
+            lines = [
+                lines[i]
+                .split("\n")[0]
+                .replace("pipe:aws s3 cp s3://s-laion-audio/", "/mnt/audio_clip/")
+                for i in range(len(lines))
+            ]
+        else:
+            lines = [
+                lines[i].split("\n")[0].replace(".tar", ".tar -")
+                for i in range(len(lines))
+            ]
+        if proportion != 1:
+            print("Sampling tars with proportion of {}".format(proportion))
+            lines = random.sample(lines, int(proportion * len(lines)))
+        return lines
+
+
+def get_mix_lambda(mixup_alpha, batch_size):
+    mixup_lambdas = [
+        np.random.beta(mixup_alpha, mixup_alpha, 1)[0] for _ in range(batch_size)
+    ]
+    return np.array(mixup_lambdas).astype(np.float32)
+
+
+def do_mixup(x, mixup_lambda):
+    """
+    Args:
+      x: (batch_size , ...)
+      mixup_lambda: (batch_size,)
+    Returns:
+      out: (batch_size, ...)
+    """
+    out = (
+        x.transpose(0, -1) * mixup_lambda
+        + torch.flip(x, dims=[0]).transpose(0, -1) * (1 - mixup_lambda)
+    ).transpose(0, -1)
+    return out
+
+
+def interpolate(x, ratio):
+    """Interpolate data in time domain. This is used to compensate the
+    resolution reduction in downsampling of a CNN.
+
+    Args:
+      x: (batch_size, time_steps, classes_num)
+      ratio: int, ratio to interpolate
+    Returns:
+      upsampled: (batch_size, time_steps * ratio, classes_num)
+    """
+    (batch_size, time_steps, classes_num) = x.shape
+    upsampled = x[:, :, None, :].repeat(1, 1, ratio, 1)
+    upsampled = upsampled.reshape(batch_size, time_steps * ratio, classes_num)
+    return upsampled
+
+
+def pad_framewise_output(framewise_output, frames_num):
+    """Pad framewise_output to the same length as input frames. The pad value
+    is the same as the value of the last frame.
+    Args:
+      framewise_output: (batch_size, frames_num, classes_num)
+      frames_num: int, number of frames to pad
+    Outputs:
+      output: (batch_size, frames_num, classes_num)
+    """
+    pad = framewise_output[:, -1:, :].repeat(
+        1, frames_num - framewise_output.shape[1], 1
+    )
+    """tensor for padding"""
+
+    output = torch.cat((framewise_output, pad), dim=1)
+    """(batch_size, frames_num, classes_num)"""
+
+
+# def process_ipc(index_path, classes_num, filename):
+#     # load data
+#     logging.info("Load Data...............")
+#     ipc = [[] for _ in range(classes_num)]
+#     with h5py.File(index_path, "r") as f:
+#         for i in tqdm(range(len(f["target"]))):
+#             t_class = np.where(f["target"][i])[0]
+#             for t in t_class:
+#                 ipc[t].append(i)
+#     print(ipc)
+#     np.save(filename, ipc)
+#     logging.info("Load Data Succeed...............")
+
+
+def save_to_dict(s, o_={}):
+    sp = s.split(": ")
+    o_.update({sp[0]: float(sp[1])})
+    return o_
+
+
+def get_data_from_log(txt_path):
+    """
+    Output dictionary from out.txt log file
+    """
+    with open(txt_path) as f:
+        lines = f.readlines()
+    val_data = {}
+    train_data = {}
+    train_losses = []
+    train_losses_epoch = []
+    for i in range(len(lines)):
+        if "| INFO |" in lines[i]:
+            if "Eval Epoch" in lines[i]:
+                if "val_loss" in lines[i]:
+                    # float(regex.sub("", lines[310].split("	")[-1]).replace(" ", ""))
+                    line = lines[i].split("Eval Epoch: ")[-1]
+                    num_epoch = int(line.split("	")[0].split(" ")[0])
+                    d = {
+                        line.split("	")[0]
+                        .split(" ")[1]
+                        .replace(":", ""): float(line.split("	")[0].split(" ")[-1])
+                    }
+                    for i in range(1, len(line.split("	"))):
+                        d = save_to_dict(line.split("	")[i], d)
+                    val_data[num_epoch] = d
+            elif "Train Epoch" in lines[i]:
+                num_epoch = int(lines[i].split("Train Epoch: ")[1][0])
+                loss = float(lines[i].split("Loss: ")[-1].split(" (")[0])
+                train_losses.append(loss)
+                train_losses_epoch.append(num_epoch)
+    for i in range(len(train_losses)):
+        train_data[i] = {
+            "num_epoch": train_losses_epoch[i],
+            "train_loss": train_losses[i],
+        }
+    return train_data, val_data
+
+
+def save_p(obj, filename):
+    import pickle
+
+    try:
+        from deepdiff import DeepDiff
+    except:
+        os.system("pip install deepdiff")
+        from deepdiff import DeepDiff
+    with open(filename, "wb") as file:
+        pickle.dump(obj, file, protocol=pickle.HIGHEST_PROTOCOL)  # highest protocol
+    with open(filename, "rb") as file:
+        z = pickle.load(file)
+    assert (
+        DeepDiff(obj, z, ignore_string_case=True) == {}
+    ), "there is something wrong with the saving process"
+    return
+
+
+def load_p(filename):
+    import pickle
+
+    with open(filename, "rb") as file:
+        z = pickle.load(file)
+    return z
+
+
+def save_json(data, name="data.json"):
+    import json
+
+    with open(name, "w") as fp:
+        json.dump(data, fp)
+    return
+
+
+def load_json(name):
+    import json
+
+    with open(name, "r") as fp:
+        data = json.load(fp)
+    return data
+
+
+from multiprocessing import Process, Manager
+from multiprocessing import Process, Value, Array
+from ctypes import c_wchar
+
+
+def load_class_label(path):
+    # https://stackoverflow.com/questions/48004243/how-to-share-large-read-only-dictionary-list-across-processes-in-multiprocessing
+    # https://stackoverflow.com/questions/45693949/storing-strings-in-a-multiprocessing-sharedctypes-array
+    out = None
+    if path is not None:
+        if pathlib.Path(path).suffix in [".pkl", ".pickle"]:
+            out = load_p(path)
+        elif pathlib.Path(path).suffix in [".json", ".txt"]:
+            out = load_json(path)
+        elif pathlib.Path(path).suffix in [".npy", ".npz"]:
+            out = np.load(path)
+        elif pathlib.Path(path).suffix in [".csv"]:
+            import pandas as pd
+
+            out = pd.read_csv(path)
+    return out
+    # if out is None:
+    #     return None
+    # else:
+    #     key = Array(c_wchar, '\n'.join(list(out.keys())), lock=False)
+    #     val = Array('i', out.values(), lock=False)
+    #     return (key, val)
+
+
+from torch import optim
+
+
+def get_optimizer(params, lr, betas, eps, momentum, optimizer_name):
+    if optimizer_name.lower() == "adamw":
+        optimizer = optim.AdamW(params, lr=lr, betas=betas, eps=eps)
+    elif optimizer_name.lower() == "sgd":
+        optimizer = optim.SGD(params, lr=lr, momentum=momentum)
+    elif optimizer_name.lower() == "adam":
+        optimizer = optim.Adam(params, lr=lr, betas=betas, eps=eps)
+    else:
+        raise ValueError("optimizer name is not correct")
+    return optimizer
diff --git a/picoaudio/audioldm/clap/open_clip/version.py b/picoaudio/audioldm/clap/open_clip/version.py
new file mode 100644
index 0000000000000000000000000000000000000000..3ced3581bb601ae91b1e1da4b8f4f520855a065e
--- /dev/null
+++ b/picoaudio/audioldm/clap/open_clip/version.py
@@ -0,0 +1 @@
+__version__ = "0.2.1"
diff --git a/picoaudio/audioldm/clap/training/__init__.py b/picoaudio/audioldm/clap/training/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/picoaudio/audioldm/clap/training/audioset_textmap.npy b/picoaudio/audioldm/clap/training/audioset_textmap.npy
new file mode 100644
index 0000000000000000000000000000000000000000..3da4c92d3819aaec11e5f576464a9973a6df811b
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/audioset_textmap.npy
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bada103070d92f9eadd33e1b4f45ec8583f59080ef218c966b43294bd4c86d5b
+size 84448
diff --git a/picoaudio/audioldm/clap/training/data.py b/picoaudio/audioldm/clap/training/data.py
new file mode 100644
index 0000000000000000000000000000000000000000..a005fee2f51e577446839b8cffd117d9ae93abc9
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/data.py
@@ -0,0 +1,981 @@
+import ast
+import json
+import logging
+import math
+import os
+import random
+
+# import h5py
+from dataclasses import dataclass
+from audioldm.clap.training.params import parse_args
+
+# import braceexpand
+import numpy as np
+import pandas as pd
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+import torchvision.datasets as datasets
+import torchvision.transforms
+
+# import webdataset as wds
+from PIL import Image
+from torch.utils.data import Dataset, DataLoader, SubsetRandomSampler
+from torch.utils.data.distributed import DistributedSampler
+from functools import partial
+import soundfile as sf
+import io
+from pathlib import Path
+
+# import wget
+
+from audioldm.clap.open_clip.utils import (
+    get_tar_path_from_dataset_name,
+    dataset_split,
+)
+from audioldm.clap.open_clip.utils import load_p, load_class_label
+import copy
+
+try:
+    import horovod.torch as hvd
+except ImportError:
+    hvd = None
+
+try:
+    import torchaudio
+except ImportError:
+    torchaudio = None
+
+from audioldm.clap.open_clip import tokenize
+
+
+def tokenizer(text):
+    return tokenize(text).squeeze(0)
+
+
+from transformers import RobertaTokenizer
+
+tokenize = RobertaTokenizer.from_pretrained("roberta-base")
+
+
+def tokenizer(text):
+    result = tokenize(
+        text,
+        padding="max_length",
+        truncation=True,
+        max_length=77,
+        return_tensors="pt",
+    )
+    return {k: v.squeeze(0) for k, v in result.items()}
+
+
+# initizlied the audioset map
+_AUDIOSET_MAP_PATH = os.path.join(Path(__file__).parent, "audioset_textmap.npy")
+_AUDIOSET_MAP = np.load(_AUDIOSET_MAP_PATH, allow_pickle=True)
+
+
+def int16_to_float32(x):
+    return (x / 32767.0).astype(np.float32)
+
+
+def float32_to_int16(x):
+    x = np.clip(x, a_min=-1.0, a_max=1.0)
+    return (x * 32767.0).astype(np.int16)
+
+
+# For Toy Dataset
+# class ToyDataset(Dataset):
+#     def __init__(self, index_path, ipc, config, eval_mode=False):
+#         """Toy Dataset for testing the audioset input with text labels
+#         Parameters
+#         ----------
+#             index_path: str
+#                 the link to the h5 file of each audio
+#             idc: str
+#                 the link to the npy file, the number of samples in each class
+#             config: dict
+#                 the audio cfg file
+#            eval_model (bool): to indicate if the dataset is a testing dataset
+#         """
+#         self.audio_cfg = config["audio_cfg"]
+#         self.text_cfg = config["text_cfg"]
+#         self.fp = h5py.File(index_path, "r")
+#         self.ipc = np.load(ipc, allow_pickle=True)
+#         self.total_size = len(self.fp["audio_name"])
+#         self.classes_num = self.audio_cfg["class_num"]
+#         self.eval_mode = eval_mode
+
+#         if not eval_mode:
+#             self.generate_queue()
+#         else:
+#             self.queue = []
+#             for i in range(self.total_size):
+#                 target = self.fp["target"][i]
+#                 if np.sum(target) > 0:
+#                     self.queue.append(i)
+#             self.total_size = len(self.queue)
+#         logging.info("total dataset size: %d" % (self.total_size))
+#         logging.info("class num: %d" % (self.classes_num))
+
+#     def time_shifting(self, x):
+#         frame_num = len(x)
+#         shift_len = random.randint(0, frame_num - 1)
+#         new_sample = np.concatenate([x[shift_len:], x[:shift_len]], axis=0)
+#         return new_sample
+
+#     def generate_queue(self):
+#         self.queue = []
+#         while len(self.queue) < self.total_size:
+#             class_set = [*range(self.classes_num)]
+#             random.shuffle(class_set)
+#             self.queue += [
+#                 self.ipc[d][random.randint(0, len(self.ipc[d]) - 1)] for d in class_set
+#             ]
+#         self.queue = self.queue[: self.total_size]
+
+#         logging.info("queue regenerated:%s" % (self.queue[-5:]))
+
+#     def crop_wav(self, x):
+#         crop_size = self.audio_cfg["crop_size"]
+#         crop_pos = random.randint(0, len(x) - crop_size - 1)
+#         return x[crop_pos : crop_pos + crop_size]
+
+#     def prompt_text(self, target):
+#         events = _AUDIOSET_MAP[np.where(target > 0)]
+#         event_text = "The sounds of " + ", ".join(events[:-1]) + " and " + events[-1]
+#         text = tokenize(event_text)[0]
+#         return text
+
+#     def __getitem__(self, index):
+#         """Load waveform, text, and target of an audio clip
+
+#         Parameters
+#         ----------
+#             index: int
+#                 the index number
+#         Return
+#         ------
+#             output: dict {
+#                 "hdf5_path": str,
+#                 "index_in_hdf5": int,
+#                 "audio_name": str,
+#                 "waveform": list (audio_length,),
+#                 "target": list (class_num, ),
+#                 "text": torch.tensor (context_length,)
+#             }
+#                 the output dictionary
+#         """
+#         s_index = self.queue[index]
+
+#         audio_name = self.fp["audio_name"][s_index].decode()
+#         # Hardcode here CHANGE
+#         hdf5_path = (
+#             self.fp["hdf5_path"][s_index]
+#             .decode()
+#             .replace(
+#                 "../workspace",
+#                 "/home/la/kechen/Research/ke_zsasp/workspace",
+#             )
+#         )
+#         r_idx = self.fp["index_in_hdf5"][s_index]
+#         target = self.fp["target"][s_index].astype(np.float32)
+#         text = self.prompt_text(target)
+#         with h5py.File(hdf5_path, "r") as f:
+#             waveform = int16_to_float32(f["waveform"][r_idx])[
+#                 : self.audio_cfg["clip_samples"]
+#             ]
+#         assert (
+#             len(waveform) == self.audio_cfg["clip_samples"]
+#         ), "The sample length is not match"
+#         # Time shift
+#         # if (self.config.enable_time_shift) and (not self.eval_mode):
+#         #     waveform = self.time_shifting(waveform)
+#         # # Label Enhance
+#         # if (self.config.crop_size is not None) and (not self.eval_mode):
+#         #     waveform = self.crop_wav(waveform)
+#         # # the label enhance rate is fixed 0.5
+#         # if (self.config.enable_label_enhance) and (not self.eval_mode) and random.random() < 0.5:
+#         #     kidx = np.where(target)[0]
+#         #     for k in kidx:
+#         #         for add_key in self.class_map[k][1]:
+#         #             target[add_key] = 1.0
+#         #         if len(self.class_map[k][2]) > 0:
+#         #             add_key = random.choice(self.class_map[k][2])
+#         #             target[add_key] = 1.0
+
+#         # missing the text input
+#         mel_spec = get_mel(torch.from_numpy(waveform), self.audio_cfg)[None, :, :]
+#         mel_spec = (
+#             torch.cat(
+#                 [mel_spec, mel_spec.clone(), mel_spec.clone(), mel_spec.clone()], dim=0
+#             )
+#             .cpu()
+#             .numpy()
+#         )
+#         longer = random.choice([True, False])
+#         if longer == False:
+#             mel_spec[1:, :, :] = 0.0
+#         data_dict = {
+#             "hdf5_path": hdf5_path,
+#             "index_in_hdf5": r_idx,
+#             "audio_name": audio_name,
+#             "waveform": waveform,
+#             "class_label": target,
+#             "text": text,
+#             "longer": longer,
+#             "mel_fusion": mel_spec,
+#         }
+#         return data_dict
+
+#     def __len__(self):
+#         return self.total_size
+
+
+class CsvDataset(Dataset):
+    def __init__(self, input_filename, transforms, img_key, caption_key, sep="\t"):
+        logging.debug(f"Loading csv data from {input_filename}.")
+        df = pd.read_csv(input_filename, sep=sep)
+
+        self.images = df[img_key].tolist()
+        self.captions = df[caption_key].tolist()
+        self.transforms = transforms
+        logging.debug("Done loading data.")
+
+    def __len__(self):
+        return len(self.captions)
+
+    def __getitem__(self, idx):
+        images = self.transforms(Image.open(str(self.images[idx])))
+        texts = tokenize([str(self.captions[idx])])[0]
+        return images, texts
+
+
+@dataclass
+class DataInfo:
+    dataloader: DataLoader
+    sampler: DistributedSampler
+
+
+def preprocess_txt(text):
+    return tokenize([str(text)])[0]
+
+
+def get_dataset_size(shards, sizefilepath_=None, is_local=True):
+    if isinstance(shards, list):
+        size_list = []
+        for s in shards:
+            size_list.append(
+                get_dataset_size(s, sizefilepath_=sizefilepath_, is_local=is_local)[0]
+            )
+    else:
+        if not is_local:
+            for n in dataset_split.keys():
+                if n in shards.split("/"):
+                    break
+            for s in dataset_split[n]:
+                if s in shards.split("/"):
+                    break
+            sizefilepath_ = f"./json_files/{n}/{s}/sizes.json"
+        shards_list = list(braceexpand.braceexpand(shards))
+        dir_path = os.path.dirname(shards)
+        if sizefilepath_ is not None:
+            sizes = json.load(open(sizefilepath_, "r"))
+            total_size = sum(
+                [
+                    int(sizes[os.path.basename(shard.replace(".tar -", ".tar"))])
+                    for shard in shards_list
+                ]
+            )
+        else:
+            sizes_filename = os.path.join(dir_path, "sizes.json")
+            len_filename = os.path.join(dir_path, "__len__")
+            if os.path.exists(sizes_filename):
+                sizes = json.load(open(sizes_filename, "r"))
+                total_size = sum(
+                    [int(sizes[os.path.basename(shard)]) for shard in shards_list]
+                )
+            elif os.path.exists(len_filename):
+                # FIXME this used to be eval(open(...)) but that seemed rather unsafe
+                total_size = ast.literal_eval(open(len_filename, "r").read())
+            else:
+                raise Exception(
+                    "Cannot find sizes file for dataset. Please specify the path to the file."
+                )
+                # total_size = None  # num samples undefined
+                # some common dataset sizes (at time of authors last download)
+                # cc3m-train: 2905954
+                # cc12m: 10968539
+                # LAION-400m: 407332084
+        num_shards = len(shards_list)
+    if isinstance(shards, list):
+        return sum(size_list), len(shards)
+    else:
+        return total_size, num_shards
+
+
+def get_imagenet(args, preprocess_fns, split):
+    assert split in ["train", "val", "v2"]
+    is_train = split == "train"
+    preprocess_train, preprocess_val = preprocess_fns
+
+    if split == "v2":
+        from imagenetv2_pytorch import ImageNetV2Dataset
+
+        dataset = ImageNetV2Dataset(location=args.imagenet_v2, transform=preprocess_val)
+    else:
+        if is_train:
+            data_path = args.imagenet_train
+            preprocess_fn = preprocess_train
+        else:
+            data_path = args.imagenet_val
+            preprocess_fn = preprocess_val
+        assert data_path
+
+        dataset = datasets.ImageFolder(data_path, transform=preprocess_fn)
+
+    if is_train:
+        idxs = np.zeros(len(dataset.targets))
+        target_array = np.array(dataset.targets)
+        k = 50
+        for c in range(1000):
+            m = target_array == c
+            n = len(idxs[m])
+            arr = np.zeros(n)
+            arr[:k] = 1
+            np.random.shuffle(arr)
+            idxs[m] = arr
+
+        idxs = idxs.astype("int")
+        sampler = SubsetRandomSampler(np.where(idxs)[0])
+    else:
+        sampler = None
+
+    dataloader = torch.utils.data.DataLoader(
+        dataset,
+        batch_size=args.batch_size,
+        num_workers=args.workers,
+        sampler=sampler,
+    )
+
+    return DataInfo(dataloader, sampler)
+
+
+def count_samples(dataloader):
+    os.environ["WDS_EPOCH"] = "0"
+    n_elements, n_batches = 0, 0
+    for images, texts in dataloader:
+        n_batches += 1
+        n_elements += len(images)
+        assert len(images) == len(texts)
+    return n_elements, n_batches
+
+
+def filter_no_caption(sample):
+    return "txt" in sample
+
+
+def log_and_continue(exn):
+    """Call in an exception handler to ignore any exception, isssue a warning, and continue."""
+    logging.warning(f"Handling webdataset error ({repr(exn)}). Ignoring.")
+    return True
+
+
+_SHARD_SHUFFLE_SIZE = 2000
+_SHARD_SHUFFLE_INITIAL = 500
+_SAMPLE_SHUFFLE_SIZE = 5000
+_SAMPLE_SHUFFLE_INITIAL = 1000
+
+
+def sample_prop(sizefile, inputs, proportion, is_local=True):
+    """
+    Sample a proportion of the data.
+    """
+    file_path_dict = {
+        os.path.split(inputs[i])[1]: os.path.split(inputs[i])[0]
+        for i in range(len(inputs))
+    }
+    sampled_filepath_dict = {}
+    sampled_size_dict = {}
+    if not is_local:
+        if os.path.exists("sizes.json"):
+            os.remove("sizes.json")
+        wget.download(sizefile, "sizes.json")
+        sizefile = "sizes.json"
+    with open(sizefile, "r", encoding="UTF-8") as f:
+        load_dict = json.load(f)
+    L = int(len(file_path_dict) * proportion)
+    subkeys = random.sample(file_path_dict.keys(), L)
+    for k in subkeys:
+        sampled_size_dict[k] = load_dict[k]
+        sampled_filepath_dict[k] = file_path_dict[k]
+    return (
+        sum(sampled_size_dict.values()),
+        L,
+        [os.path.join(v, k) for k, v in sampled_filepath_dict.items()],
+        sampled_size_dict,
+    )
+
+
+def get_mel(audio_data, audio_cfg):
+    # mel shape: (n_mels, T)
+    mel = torchaudio.transforms.MelSpectrogram(
+        sample_rate=audio_cfg["sample_rate"],
+        n_fft=audio_cfg["window_size"],
+        win_length=audio_cfg["window_size"],
+        hop_length=audio_cfg["hop_size"],
+        center=True,
+        pad_mode="reflect",
+        power=2.0,
+        norm=None,
+        onesided=True,
+        n_mels=64,
+        f_min=audio_cfg["fmin"],
+        f_max=audio_cfg["fmax"],
+    ).to(audio_data.device)
+    mel = mel(audio_data)
+    # Align to librosa:
+    # librosa_melspec = librosa.feature.melspectrogram(
+    #     waveform,
+    #     sr=audio_cfg['sample_rate'],
+    #     n_fft=audio_cfg['window_size'],
+    #     hop_length=audio_cfg['hop_size'],
+    #     win_length=audio_cfg['window_size'],
+    #     center=True,
+    #     pad_mode="reflect",
+    #     power=2.0,
+    #     n_mels=64,
+    #     norm=None,
+    #     htk=True,
+    #     f_min=audio_cfg['fmin'],
+    #     f_max=audio_cfg['fmax']
+    # )
+    # we use log mel spectrogram as input
+    mel = torchaudio.transforms.AmplitudeToDB(top_db=None)(mel)
+    return mel.T  # (T, n_mels)
+
+
+def get_audio_features(
+    sample, audio_data, max_len, data_truncating, data_filling, audio_cfg
+):
+    """
+    Calculate and add audio features to sample.
+    Sample: a dict containing all the data of current sample.
+    audio_data: a tensor of shape (T) containing audio data.
+    max_len: the maximum length of audio data.
+    data_truncating: the method of truncating data.
+    data_filling: the method of filling data.
+    audio_cfg: a dict containing audio configuration. Comes from model_cfg['audio_cfg'].
+    """
+    with torch.no_grad():
+        if len(audio_data) > max_len:
+            if data_truncating == "rand_trunc":
+                longer = torch.tensor([True])
+            elif data_truncating == "fusion":
+                # fusion
+                mel = get_mel(audio_data, audio_cfg)
+                # split to three parts
+                chunk_frames = (
+                    max_len // audio_cfg["hop_size"] + 1
+                )  # the +1 related to how the spectrogram is computed
+                total_frames = mel.shape[0]
+                if chunk_frames == total_frames:
+                    # there is a corner case where the audio length is
+                    # larger than max_len but smaller than max_len+hop_size.
+                    # In this case, we just use the whole audio.
+                    mel_fusion = torch.stack([mel, mel, mel, mel], dim=0)
+                    sample["mel_fusion"] = mel_fusion
+                    longer = torch.tensor([False])
+                else:
+                    ranges = np.array_split(
+                        list(range(0, total_frames - chunk_frames + 1)), 3
+                    )
+                    # print('total_frames-chunk_frames:', total_frames-chunk_frames,
+                    #       'len(audio_data):', len(audio_data),
+                    #       'chunk_frames:', chunk_frames,
+                    #       'total_frames:', total_frames)
+                    if len(ranges[1]) == 0:
+                        # if the audio is too short, we just use the first chunk
+                        ranges[1] = [0]
+                    if len(ranges[2]) == 0:
+                        # if the audio is too short, we just use the first chunk
+                        ranges[2] = [0]
+                    # randomly choose index for each part
+                    idx_front = np.random.choice(ranges[0])
+                    idx_middle = np.random.choice(ranges[1])
+                    idx_back = np.random.choice(ranges[2])
+                    # select mel
+                    mel_chunk_front = mel[idx_front : idx_front + chunk_frames, :]
+                    mel_chunk_middle = mel[idx_middle : idx_middle + chunk_frames, :]
+                    mel_chunk_back = mel[idx_back : idx_back + chunk_frames, :]
+
+                    # shrink the mel
+                    mel_shrink = torchvision.transforms.Resize(size=[chunk_frames, 64])(
+                        mel[None]
+                    )[0]
+                    # logging.info(f"mel_shrink.shape: {mel_shrink.shape}")
+
+                    # stack
+                    mel_fusion = torch.stack(
+                        [mel_chunk_front, mel_chunk_middle, mel_chunk_back, mel_shrink],
+                        dim=0,
+                    )
+                    sample["mel_fusion"] = mel_fusion
+                    longer = torch.tensor([True])
+            else:
+                raise NotImplementedError(
+                    f"data_truncating {data_truncating} not implemented"
+                )
+            # random crop to max_len (for compatibility)
+            overflow = len(audio_data) - max_len
+            idx = np.random.randint(0, overflow + 1)
+            audio_data = audio_data[idx : idx + max_len]
+
+        else:  # padding if too short
+            if len(audio_data) < max_len:  # do nothing if equal
+                if data_filling == "repeatpad":
+                    n_repeat = int(max_len / len(audio_data))
+                    audio_data = audio_data.repeat(n_repeat)
+                    # audio_data = audio_data.unsqueeze(0).unsqueeze(0).unsqueeze(0)
+                    # audio_data = F.interpolate(audio_data,size=max_len,mode="bicubic")[0,0,0]
+                    audio_data = F.pad(
+                        audio_data,
+                        (0, max_len - len(audio_data)),
+                        mode="constant",
+                        value=0,
+                    )
+                elif data_filling == "pad":
+                    audio_data = F.pad(
+                        audio_data,
+                        (0, max_len - len(audio_data)),
+                        mode="constant",
+                        value=0,
+                    )
+                elif data_filling == "repeat":
+                    n_repeat = int(max_len / len(audio_data))
+                    audio_data = audio_data.repeat(n_repeat + 1)[:max_len]
+                else:
+                    raise NotImplementedError(
+                        f"data_filling {data_filling} not implemented"
+                    )
+            if data_truncating == "fusion":
+                mel = get_mel(audio_data, audio_cfg)
+                mel_fusion = torch.stack([mel, mel, mel, mel], dim=0)
+                sample["mel_fusion"] = mel_fusion
+            longer = torch.tensor([False])
+
+    sample["longer"] = longer
+    sample["waveform"] = audio_data
+
+    return sample
+
+
+def preprocess(
+    sample,
+    audio_ext,
+    text_ext,
+    max_len,
+    audio_cfg,
+    class_index_dict=None,
+    data_filling="pad",
+    data_truncating="rand_trunc",
+    text_augment_selection=None,
+):
+    """
+    Preprocess a single sample for wdsdataloader.
+    """
+    audio_data, orig_sr = sf.read(io.BytesIO(sample[audio_ext]))
+    audio_data = int16_to_float32(float32_to_int16(audio_data))
+    audio_data = torch.tensor(audio_data).float()
+
+    # TODO: (yusong) to be include in the future
+    # # if torchaudio not installed, use soundfile to load audio
+    # if torchaudio is None:
+    #     audio_data, orig_sr = sf.read(io.BytesIO(sample[audio_ext]))
+    #     audio_data = torch.tensor(audio_data).float()
+    # else:
+    #     # https://github.com/webdataset/webdataset/blob/main/webdataset/autodecode.py
+    #     with tempfile.TemporaryDirectory() as dirname:
+    #         os.makedirs(dirname, exist_ok=True)
+    #         fname = os.path.join(dirname, f"file.flac")
+    #         with open(fname, "wb") as stream:
+    #             stream.write(sample[audio_ext])
+    #         audio_data, orig_sr = torchaudio.load(fname)
+    #         audio_data = audio_data[0, :].float()
+
+    sample = get_audio_features(
+        sample, audio_data, max_len, data_truncating, data_filling, audio_cfg
+    )
+    del sample[audio_ext]
+
+    try:
+        json_dict_raw = json.loads(sample[text_ext].decode("utf-8"))
+    except:
+        print("sample[__url__]:", sample["__url__"])
+
+    # For selecting augmented text from dataset
+    if text_augment_selection is None or text_augment_selection == "none":
+        texts = json_dict_raw["text"]
+    elif text_augment_selection == "all":
+        if "text_augment_all" in json_dict_raw.keys():
+            texts = json_dict_raw["text_augment_all"]
+        else:
+            texts = json_dict_raw["text"]
+    elif text_augment_selection == "augment_only":
+        if "text_augment_all" in json_dict_raw.keys():
+            if json_dict_raw["text_augment_t5"] is None:
+                texts = json_dict_raw["text"]
+            else:
+                texts = json_dict_raw["text_augment_t5"]
+        else:
+            texts = json_dict_raw["text"]
+    else:
+        raise NotImplementedError(
+            f"text_augment_selection {text_augment_selection} not implemented"
+        )
+    sample["full_text"] = texts
+
+    if isinstance(texts, list) and isinstance(texts[0], str) and len(texts) > 1:
+        texts = random.choice(texts)
+    sample["raw_text"] = texts
+    sample["text"] = tokenizer(texts)  # text shape: [num_token]
+    if class_index_dict is not None:
+        # https://stackoverflow.com/questions/48004243/how-to-share-large-read-only-dictionary-list-across-processes-in-multiprocessing
+        # https://stackoverflow.com/questions/45693949/storing-strings-in-a-multiprocessing-sharedctypes-array
+        # key, val = class_index_dict
+        # key = key[:].split('\n')
+        # _dict = {k: v for k, v in zip(key, val)}
+        sample["class_label"] = np.zeros(len(class_index_dict.keys()))
+        for x in json_dict_raw["tag"]:
+            sample["class_label"][class_index_dict[x]] = 1
+        sample["class_label"] = torch.tensor(sample["class_label"]).float()
+    del sample[text_ext]
+    sample["audio_name"] = sample["__key__"].split("/")[-1] + "." + audio_ext
+    sample["text_name"] = sample["__key__"].split("/")[-1] + "." + text_ext
+    sample["audio_orig_sr"] = orig_sr
+    return sample
+
+
+def collate_fn(batch):
+    """
+    Collate function for wdsdataloader.
+    batch: a list of dict, each dict is a sample
+    """
+    # concatenate values in each dictionary. if it is a tensor, concatenate. if it is a list, extend.
+    batch_dict = {}
+    for k in batch[0].keys():
+        if isinstance(batch[0][k], dict):  # dealwith bert tokenizer output
+            batch_dict[k] = {}
+            for kk in batch[0][k].keys():
+                tmp = []
+                for i in range(len(batch)):
+                    tmp.append(batch[i][k][kk])
+                batch_dict[k][kk] = torch.vstack(tmp)
+        elif isinstance(batch[0][k], torch.Tensor):
+            batch_dict[k] = torch.stack([sample[k] for sample in batch])
+        elif isinstance(batch[0][k], np.ndarray):
+            batch_dict[k] = torch.tensor(np.stack([sample[k] for sample in batch]))
+        else:
+            batch_dict[k] = [sample[k] for sample in batch]
+    return batch_dict
+
+
+def get_wds_dataset(
+    args,
+    model_cfg,
+    is_train,
+    audio_ext="flac",
+    text_ext="json",
+    max_len=480000,
+    proportion=1.0,
+    sizefilepath_=None,
+    is_local=None,
+):
+    """
+    Get a dataset for wdsdataloader.
+    """
+    if is_local is None and (not args.remotedata is None):
+        is_local = not args.remotedata
+
+    input_shards = args.train_data if is_train else args.val_data
+    assert input_shards is not None
+
+    if not sizefilepath_ is None:
+        sizefilepath = sizefilepath_
+    else:
+        sizefilepath = os.path.join(os.path.dirname(input_shards[0]), "sizes.json")
+
+    if proportion != 1.0:
+        num_samples, num_shards, input_shards, _ = sample_prop(
+            sizefilepath, input_shards, proportion, is_local=is_local
+        )
+    else:
+        num_samples, num_shards = get_dataset_size(
+            input_shards, sizefilepath_=sizefilepath_, is_local=is_local
+        )
+
+    if not num_samples:
+        if is_train:
+            num_samples = args.train_num_samples
+            if not num_samples:
+                raise RuntimeError(
+                    "Currently, number of dataset samples must be specified for training dataset. "
+                    "Please specify via `--train-num-samples` if no dataset length info present."
+                )
+        else:
+            num_samples = (
+                args.val_num_samples or 0
+            )  # eval will just exhaust the iterator if not specified
+
+    pipeline = [wds.SimpleShardList(input_shards)]
+    # at this point we have an iterator over all the shards
+    # TODO: (yusong): add a if statement of distributed. If not, we don't need to split_by_node
+    if is_train or args.parallel_eval:
+        pipeline.extend(
+            [
+                wds.detshuffle(
+                    bufsize=_SHARD_SHUFFLE_SIZE,
+                    initial=_SHARD_SHUFFLE_INITIAL,
+                    seed=args.seed,
+                ),
+                wds.split_by_node,
+                wds.split_by_worker,
+                # at this point, we have an iterator over the shards assigned to each worker at each node
+                wds.tarfile_to_samples(handler=log_and_continue),
+                wds.shuffle(
+                    bufsize=_SAMPLE_SHUFFLE_SIZE,
+                    initial=_SAMPLE_SHUFFLE_INITIAL,
+                    rng=random.Random(args.seed),
+                ),
+                # wds.repeatedly,  # FIXME determine if this is beneficial
+            ]
+        )
+    else:
+        pipeline.extend(
+            [
+                wds.split_by_worker,
+                # at this point, we have an iterator over the shards assigned to each worker
+                wds.tarfile_to_samples(handler=log_and_continue),
+            ]
+        )
+    pipeline.append(
+        wds.map(
+            partial(
+                preprocess,
+                audio_ext=audio_ext,
+                text_ext=text_ext,
+                max_len=max_len,
+                audio_cfg=model_cfg["audio_cfg"],
+                class_index_dict=copy.deepcopy(args.class_index_dict),
+                data_filling=args.data_filling,
+                data_truncating=args.data_truncating,
+                text_augment_selection=args.text_augment_selection,
+            )
+        ),
+    )
+
+    pipeline.append(
+        wds.batched(
+            args.batch_size,
+            partial=not (is_train or args.parallel_eval),
+            collation_fn=collate_fn,
+        )
+    )
+
+    dataset = wds.DataPipeline(*pipeline)
+    if is_train or args.parallel_eval:
+        # (yusong): Currently parallel evaluation will be not precise as we are repeat the last few samples.
+        # (yusong): See comments below.
+        # roll over and repeat a few samples to get same number of full batches on each node
+        global_batch_size = args.batch_size * args.world_size
+        num_batches = math.ceil(num_samples / global_batch_size)
+        num_workers = max(1, args.workers)
+        num_worker_batches = math.ceil(
+            num_batches / num_workers
+        )  # per dataloader worker
+        num_batches = num_worker_batches * num_workers
+        num_samples = num_batches * global_batch_size
+        dataset = dataset.with_epoch(
+            num_worker_batches
+        )  # each worker is iterating over this
+    else:
+        # last batches are partial, eval is done on single (master) node
+        num_batches = math.ceil(num_samples / args.batch_size)
+
+    kwargs = {}
+    if args.horovod:  # multi-node training on summit
+        kwargs["multiprocessing_context"] = "forkserver"
+
+    dataloader = wds.WebLoader(
+        dataset, batch_size=None, shuffle=False, num_workers=args.workers, **kwargs
+    )
+
+    # FIXME not clear which approach is better, with_epoch before vs after dataloader?
+    # hoping to resolve via https://github.com/webdataset/webdataset/issues/169
+    # if is_train:
+    #     # roll over and repeat a few samples to get same number of full batches on each node
+    #     global_batch_size = args.batch_size * args.world_size
+    #     num_batches = math.ceil(num_samples / global_batch_size)
+    #     num_workers = max(1, args.workers)
+    #     num_batches = math.ceil(num_batches / num_workers) * num_workers
+    #     num_samples = num_batches * global_batch_size
+    #     dataloader = dataloader.with_epoch(num_batches)
+    # else:
+    #     # last batches are partial, eval is done on single (master) node
+    #     num_batches = math.ceil(num_samples / args.batch_size)
+
+    # add meta-data to dataloader instance for convenience
+    dataloader.num_batches = num_batches
+    dataloader.num_samples = num_samples
+
+    return DataInfo(dataloader, None)
+
+
+def wds_batch_list2dict(
+    batch,
+    keys=[
+        "__url__",
+        "__key__",
+        "waveform",
+        "text",
+        "raw_text",
+        "audio_name",
+        "text_name",
+        "audio_orig_sr",
+    ],
+):
+    """
+    Return a dictionary of the batch, with keys as the names of the fields.
+    """
+    assert len(keys) == len(
+        batch
+    ), "batch must have same number of keys as keys argument"
+    return {keys[i]: batch[i] for i in range(len(batch))}
+
+
+def get_csv_dataset(args, preprocess_fn, is_train):
+    input_filename = args.train_data if is_train else args.val_data
+    assert input_filename
+    dataset = CsvDataset(
+        input_filename,
+        preprocess_fn,
+        img_key=args.csv_img_key,
+        caption_key=args.csv_caption_key,
+        sep=args.csv_separator,
+    )
+    num_samples = len(dataset)
+    sampler = DistributedSampler(dataset) if args.distributed and is_train else None
+    shuffle = is_train and sampler is None
+
+    dataloader = DataLoader(
+        dataset,
+        batch_size=args.batch_size,
+        shuffle=shuffle,
+        num_workers=args.workers,
+        pin_memory=True,
+        sampler=sampler,
+        drop_last=is_train,
+    )
+    dataloader.num_samples = num_samples
+    dataloader.num_batches = len(dataloader)
+
+    return DataInfo(dataloader, sampler)
+
+
+def get_toy_dataset(args, model_cfg, is_train):
+    index_path = args.train_data if is_train else args.val_data
+    ipc_path = args.train_ipc if is_train else args.val_ipc
+    assert index_path and ipc_path
+    eval_mode = not is_train
+    dataset = ToyDataset(index_path, ipc_path, model_cfg, eval_mode=eval_mode)
+
+    num_samples = len(dataset)
+    sampler = (
+        DistributedSampler(dataset, shuffle=False)
+        if args.distributed and is_train
+        else None
+    )
+
+    dataloader = DataLoader(
+        dataset,
+        batch_size=args.batch_size,
+        shuffle=False,
+        num_workers=args.workers,
+        sampler=sampler,
+        drop_last=is_train,
+    )
+    dataloader.num_samples = num_samples
+    dataloader.num_batches = len(dataloader)
+
+    return DataInfo(dataloader, sampler)
+
+
+def get_dataset_fn(data_path, dataset_type):
+    if dataset_type == "webdataset":
+        return get_wds_dataset
+    elif dataset_type == "csv":
+        return get_csv_dataset
+    elif dataset_type == "auto":
+        ext = data_path.split(".")[-1]
+        if ext in ["csv", "tsv"]:
+            return get_csv_dataset
+        elif ext in ["tar"]:
+            return get_wds_dataset
+        else:
+            raise ValueError(
+                f"Tried to figure out dataset type, but failed for extention {ext}."
+            )
+    elif dataset_type == "toy":
+        return get_toy_dataset
+    else:
+        raise ValueError(f"Unsupported dataset type: {dataset_type}")
+
+
+def get_data(args, model_cfg):
+    data = {}
+
+    args.class_index_dict = load_class_label(args.class_label_path)
+
+    if args.datasetinfos is None:
+        args.datasetinfos = ["train", "unbalanced_train", "balanced_train"]
+    if args.dataset_type == "webdataset":
+        args.train_data = get_tar_path_from_dataset_name(
+            args.datasetnames,
+            args.datasetinfos,
+            islocal=not args.remotedata,
+            proportion=args.dataset_proportion,
+            dataset_path=args.datasetpath,
+            full_dataset=args.full_train_dataset,
+        )
+
+        if args.full_train_dataset is None:
+            args.full_train_dataset = []
+        if args.exclude_eval_dataset is None:
+            args.exclude_eval_dataset = []
+        excluded_eval_datasets = args.full_train_dataset + args.exclude_eval_dataset
+
+        val_dataset_names = (
+            [n for n in args.datasetnames if n not in excluded_eval_datasets]
+            if excluded_eval_datasets
+            else args.datasetnames
+        )
+        args.val_dataset_names = val_dataset_names
+        args.val_data = get_tar_path_from_dataset_name(
+            val_dataset_names,
+            ["valid", "test", "eval"],
+            islocal=not args.remotedata,
+            proportion=1,
+            dataset_path=args.datasetpath,
+            full_dataset=None,
+        )
+
+    if args.train_data:
+        data["train"] = get_dataset_fn(args.train_data, args.dataset_type)(
+            args, model_cfg, is_train=True
+        )
+
+    if args.val_data:
+        data["val"] = get_dataset_fn(args.val_data, args.dataset_type)(
+            args, model_cfg, is_train=False
+        )
+
+    return data
diff --git a/picoaudio/audioldm/clap/training/distributed.py b/picoaudio/audioldm/clap/training/distributed.py
new file mode 100644
index 0000000000000000000000000000000000000000..2fa61f76c5cc3ab9f6a9643042afa8e1f2e1cb7f
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/distributed.py
@@ -0,0 +1,150 @@
+import os
+
+import torch
+import socket
+
+try:
+    import horovod.torch as hvd
+except ImportError:
+    hvd = None
+
+
+def is_global_master(args):
+    return args.rank == 0
+
+
+def is_local_master(args):
+    return args.local_rank == 0
+
+
+def is_master(args, local=False):
+    return is_local_master(args) if local else is_global_master(args)
+
+
+def is_using_horovod():
+    # NOTE w/ horovod run, OMPI vars should be set, but w/ SLURM PMI vars will be set
+    # Differentiating between horovod and DDP use via SLURM may not be possible, so horovod arg still required...
+    ompi_vars = ["OMPI_COMM_WORLD_RANK", "OMPI_COMM_WORLD_SIZE"]
+    pmi_vars = ["PMI_RANK", "PMI_SIZE"]
+    if all([var in os.environ for var in ompi_vars]) or all(
+        [var in os.environ for var in pmi_vars]
+    ):
+        return True
+    else:
+        return False
+
+
+def is_using_distributed():
+    if "WORLD_SIZE" in os.environ:
+        return int(os.environ["WORLD_SIZE"]) > 1
+    if "SLURM_NTASKS" in os.environ:
+        return int(os.environ["SLURM_NTASKS"]) > 1
+    return False
+
+
+def world_info_from_env():
+    local_rank = 0
+    for v in (
+        "SLURM_LOCALID",
+        "MPI_LOCALRANKID",
+        "OMPI_COMM_WORLD_LOCAL_RANK",
+        "LOCAL_RANK",
+    ):
+        if v in os.environ:
+            local_rank = int(os.environ[v])
+            break
+    global_rank = 0
+    for v in ("SLURM_PROCID", "PMI_RANK", "OMPI_COMM_WORLD_RANK", "RANK"):
+        if v in os.environ:
+            global_rank = int(os.environ[v])
+            break
+    world_size = 1
+    for v in ("SLURM_NTASKS", "PMI_SIZE", "OMPI_COMM_WORLD_SIZE", "WORLD_SIZE"):
+        if v in os.environ:
+            world_size = int(os.environ[v])
+            break
+
+    return local_rank, global_rank, world_size
+
+
+def init_distributed_device(args):
+    # Distributed training = training on more than one GPU.
+    # Works in both single and multi-node scenarios.
+    args.distributed = False
+    args.world_size = 1
+    args.rank = 0  # global rank
+    args.local_rank = 0
+    if args.horovod:
+        assert hvd is not None, "Horovod is not installed"
+        hvd.init()
+        world_size = int(os.environ["OMPI_COMM_WORLD_SIZE"])
+        world_rank = int(os.environ["OMPI_COMM_WORLD_RANK"])
+        local_rank = int(os.environ["OMPI_COMM_WORLD_LOCAL_RANK"])
+        args.local_rank = local_rank
+        args.rank = world_rank
+        args.world_size = world_size
+        # args.local_rank = int(hvd.local_rank())
+        # args.rank = hvd.rank()
+        # args.world_size = hvd.size()
+        args.distributed = True
+        os.environ["LOCAL_RANK"] = str(args.local_rank)
+        os.environ["RANK"] = str(args.rank)
+        os.environ["WORLD_SIZE"] = str(args.world_size)
+        print(
+            f"Distributed training: local_rank={args.local_rank}, "
+            f"rank={args.rank}, world_size={args.world_size}, "
+            f"hostname={socket.gethostname()}, pid={os.getpid()}"
+        )
+    elif is_using_distributed():
+        if "SLURM_PROCID" in os.environ:
+            # DDP via SLURM
+            args.local_rank, args.rank, args.world_size = world_info_from_env()
+            # SLURM var -> torch.distributed vars in case needed
+            os.environ["LOCAL_RANK"] = str(args.local_rank)
+            os.environ["RANK"] = str(args.rank)
+            os.environ["WORLD_SIZE"] = str(args.world_size)
+            torch.distributed.init_process_group(
+                backend=args.dist_backend,
+                init_method=args.dist_url,
+                world_size=args.world_size,
+                rank=args.rank,
+            )
+        elif "OMPI_COMM_WORLD_SIZE" in os.environ:  # using Summit cluster
+            world_size = int(os.environ["OMPI_COMM_WORLD_SIZE"])
+            world_rank = int(os.environ["OMPI_COMM_WORLD_RANK"])
+            local_rank = int(os.environ["OMPI_COMM_WORLD_LOCAL_RANK"])
+            args.local_rank = local_rank
+            args.rank = world_rank
+            args.world_size = world_size
+            torch.distributed.init_process_group(
+                backend=args.dist_backend,
+                init_method=args.dist_url,
+                world_size=args.world_size,
+                rank=args.rank,
+            )
+        else:
+            # DDP via torchrun, torch.distributed.launch
+            args.local_rank, _, _ = world_info_from_env()
+            torch.distributed.init_process_group(
+                backend=args.dist_backend, init_method=args.dist_url
+            )
+            args.world_size = torch.distributed.get_world_size()
+            args.rank = torch.distributed.get_rank()
+        args.distributed = True
+        print(
+            f"Distributed training: local_rank={args.local_rank}, "
+            f"rank={args.rank}, world_size={args.world_size}, "
+            f"hostname={socket.gethostname()}, pid={os.getpid()}"
+        )
+
+    if torch.cuda.is_available():
+        if args.distributed and not args.no_set_device_rank:
+            device = "cuda:%d" % args.local_rank
+        else:
+            device = "cuda:0"
+        torch.cuda.set_device(device)
+    else:
+        device = "cpu"
+    args.device = device
+    device = torch.device(device)
+    return device
diff --git a/picoaudio/audioldm/clap/training/imagenet_zeroshot_data.py b/picoaudio/audioldm/clap/training/imagenet_zeroshot_data.py
new file mode 100644
index 0000000000000000000000000000000000000000..d32e55328d6799ccb8d61625f43abb80a33d6c17
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/imagenet_zeroshot_data.py
@@ -0,0 +1,1088 @@
+# NOTE: This script is currently not supported for CLAP.
+
+imagenet_classnames = [
+    "tench",
+    "goldfish",
+    "great white shark",
+    "tiger shark",
+    "hammerhead shark",
+    "electric ray",
+    "stingray",
+    "rooster",
+    "hen",
+    "ostrich",
+    "brambling",
+    "goldfinch",
+    "house finch",
+    "junco",
+    "indigo bunting",
+    "American robin",
+    "bulbul",
+    "jay",
+    "magpie",
+    "chickadee",
+    "American dipper",
+    "kite (bird of prey)",
+    "bald eagle",
+    "vulture",
+    "great grey owl",
+    "fire salamander",
+    "smooth newt",
+    "newt",
+    "spotted salamander",
+    "axolotl",
+    "American bullfrog",
+    "tree frog",
+    "tailed frog",
+    "loggerhead sea turtle",
+    "leatherback sea turtle",
+    "mud turtle",
+    "terrapin",
+    "box turtle",
+    "banded gecko",
+    "green iguana",
+    "Carolina anole",
+    "desert grassland whiptail lizard",
+    "agama",
+    "frilled-necked lizard",
+    "alligator lizard",
+    "Gila monster",
+    "European green lizard",
+    "chameleon",
+    "Komodo dragon",
+    "Nile crocodile",
+    "American alligator",
+    "triceratops",
+    "worm snake",
+    "ring-necked snake",
+    "eastern hog-nosed snake",
+    "smooth green snake",
+    "kingsnake",
+    "garter snake",
+    "water snake",
+    "vine snake",
+    "night snake",
+    "boa constrictor",
+    "African rock python",
+    "Indian cobra",
+    "green mamba",
+    "sea snake",
+    "Saharan horned viper",
+    "eastern diamondback rattlesnake",
+    "sidewinder rattlesnake",
+    "trilobite",
+    "harvestman",
+    "scorpion",
+    "yellow garden spider",
+    "barn spider",
+    "European garden spider",
+    "southern black widow",
+    "tarantula",
+    "wolf spider",
+    "tick",
+    "centipede",
+    "black grouse",
+    "ptarmigan",
+    "ruffed grouse",
+    "prairie grouse",
+    "peafowl",
+    "quail",
+    "partridge",
+    "african grey parrot",
+    "macaw",
+    "sulphur-crested cockatoo",
+    "lorikeet",
+    "coucal",
+    "bee eater",
+    "hornbill",
+    "hummingbird",
+    "jacamar",
+    "toucan",
+    "duck",
+    "red-breasted merganser",
+    "goose",
+    "black swan",
+    "tusker",
+    "echidna",
+    "platypus",
+    "wallaby",
+    "koala",
+    "wombat",
+    "jellyfish",
+    "sea anemone",
+    "brain coral",
+    "flatworm",
+    "nematode",
+    "conch",
+    "snail",
+    "slug",
+    "sea slug",
+    "chiton",
+    "chambered nautilus",
+    "Dungeness crab",
+    "rock crab",
+    "fiddler crab",
+    "red king crab",
+    "American lobster",
+    "spiny lobster",
+    "crayfish",
+    "hermit crab",
+    "isopod",
+    "white stork",
+    "black stork",
+    "spoonbill",
+    "flamingo",
+    "little blue heron",
+    "great egret",
+    "bittern bird",
+    "crane bird",
+    "limpkin",
+    "common gallinule",
+    "American coot",
+    "bustard",
+    "ruddy turnstone",
+    "dunlin",
+    "common redshank",
+    "dowitcher",
+    "oystercatcher",
+    "pelican",
+    "king penguin",
+    "albatross",
+    "grey whale",
+    "killer whale",
+    "dugong",
+    "sea lion",
+    "Chihuahua",
+    "Japanese Chin",
+    "Maltese",
+    "Pekingese",
+    "Shih Tzu",
+    "King Charles Spaniel",
+    "Papillon",
+    "toy terrier",
+    "Rhodesian Ridgeback",
+    "Afghan Hound",
+    "Basset Hound",
+    "Beagle",
+    "Bloodhound",
+    "Bluetick Coonhound",
+    "Black and Tan Coonhound",
+    "Treeing Walker Coonhound",
+    "English foxhound",
+    "Redbone Coonhound",
+    "borzoi",
+    "Irish Wolfhound",
+    "Italian Greyhound",
+    "Whippet",
+    "Ibizan Hound",
+    "Norwegian Elkhound",
+    "Otterhound",
+    "Saluki",
+    "Scottish Deerhound",
+    "Weimaraner",
+    "Staffordshire Bull Terrier",
+    "American Staffordshire Terrier",
+    "Bedlington Terrier",
+    "Border Terrier",
+    "Kerry Blue Terrier",
+    "Irish Terrier",
+    "Norfolk Terrier",
+    "Norwich Terrier",
+    "Yorkshire Terrier",
+    "Wire Fox Terrier",
+    "Lakeland Terrier",
+    "Sealyham Terrier",
+    "Airedale Terrier",
+    "Cairn Terrier",
+    "Australian Terrier",
+    "Dandie Dinmont Terrier",
+    "Boston Terrier",
+    "Miniature Schnauzer",
+    "Giant Schnauzer",
+    "Standard Schnauzer",
+    "Scottish Terrier",
+    "Tibetan Terrier",
+    "Australian Silky Terrier",
+    "Soft-coated Wheaten Terrier",
+    "West Highland White Terrier",
+    "Lhasa Apso",
+    "Flat-Coated Retriever",
+    "Curly-coated Retriever",
+    "Golden Retriever",
+    "Labrador Retriever",
+    "Chesapeake Bay Retriever",
+    "German Shorthaired Pointer",
+    "Vizsla",
+    "English Setter",
+    "Irish Setter",
+    "Gordon Setter",
+    "Brittany dog",
+    "Clumber Spaniel",
+    "English Springer Spaniel",
+    "Welsh Springer Spaniel",
+    "Cocker Spaniel",
+    "Sussex Spaniel",
+    "Irish Water Spaniel",
+    "Kuvasz",
+    "Schipperke",
+    "Groenendael dog",
+    "Malinois",
+    "Briard",
+    "Australian Kelpie",
+    "Komondor",
+    "Old English Sheepdog",
+    "Shetland Sheepdog",
+    "collie",
+    "Border Collie",
+    "Bouvier des Flandres dog",
+    "Rottweiler",
+    "German Shepherd Dog",
+    "Dobermann",
+    "Miniature Pinscher",
+    "Greater Swiss Mountain Dog",
+    "Bernese Mountain Dog",
+    "Appenzeller Sennenhund",
+    "Entlebucher Sennenhund",
+    "Boxer",
+    "Bullmastiff",
+    "Tibetan Mastiff",
+    "French Bulldog",
+    "Great Dane",
+    "St. Bernard",
+    "husky",
+    "Alaskan Malamute",
+    "Siberian Husky",
+    "Dalmatian",
+    "Affenpinscher",
+    "Basenji",
+    "pug",
+    "Leonberger",
+    "Newfoundland dog",
+    "Great Pyrenees dog",
+    "Samoyed",
+    "Pomeranian",
+    "Chow Chow",
+    "Keeshond",
+    "brussels griffon",
+    "Pembroke Welsh Corgi",
+    "Cardigan Welsh Corgi",
+    "Toy Poodle",
+    "Miniature Poodle",
+    "Standard Poodle",
+    "Mexican hairless dog (xoloitzcuintli)",
+    "grey wolf",
+    "Alaskan tundra wolf",
+    "red wolf or maned wolf",
+    "coyote",
+    "dingo",
+    "dhole",
+    "African wild dog",
+    "hyena",
+    "red fox",
+    "kit fox",
+    "Arctic fox",
+    "grey fox",
+    "tabby cat",
+    "tiger cat",
+    "Persian cat",
+    "Siamese cat",
+    "Egyptian Mau",
+    "cougar",
+    "lynx",
+    "leopard",
+    "snow leopard",
+    "jaguar",
+    "lion",
+    "tiger",
+    "cheetah",
+    "brown bear",
+    "American black bear",
+    "polar bear",
+    "sloth bear",
+    "mongoose",
+    "meerkat",
+    "tiger beetle",
+    "ladybug",
+    "ground beetle",
+    "longhorn beetle",
+    "leaf beetle",
+    "dung beetle",
+    "rhinoceros beetle",
+    "weevil",
+    "fly",
+    "bee",
+    "ant",
+    "grasshopper",
+    "cricket insect",
+    "stick insect",
+    "cockroach",
+    "praying mantis",
+    "cicada",
+    "leafhopper",
+    "lacewing",
+    "dragonfly",
+    "damselfly",
+    "red admiral butterfly",
+    "ringlet butterfly",
+    "monarch butterfly",
+    "small white butterfly",
+    "sulphur butterfly",
+    "gossamer-winged butterfly",
+    "starfish",
+    "sea urchin",
+    "sea cucumber",
+    "cottontail rabbit",
+    "hare",
+    "Angora rabbit",
+    "hamster",
+    "porcupine",
+    "fox squirrel",
+    "marmot",
+    "beaver",
+    "guinea pig",
+    "common sorrel horse",
+    "zebra",
+    "pig",
+    "wild boar",
+    "warthog",
+    "hippopotamus",
+    "ox",
+    "water buffalo",
+    "bison",
+    "ram (adult male sheep)",
+    "bighorn sheep",
+    "Alpine ibex",
+    "hartebeest",
+    "impala (antelope)",
+    "gazelle",
+    "arabian camel",
+    "llama",
+    "weasel",
+    "mink",
+    "European polecat",
+    "black-footed ferret",
+    "otter",
+    "skunk",
+    "badger",
+    "armadillo",
+    "three-toed sloth",
+    "orangutan",
+    "gorilla",
+    "chimpanzee",
+    "gibbon",
+    "siamang",
+    "guenon",
+    "patas monkey",
+    "baboon",
+    "macaque",
+    "langur",
+    "black-and-white colobus",
+    "proboscis monkey",
+    "marmoset",
+    "white-headed capuchin",
+    "howler monkey",
+    "titi monkey",
+    "Geoffroy's spider monkey",
+    "common squirrel monkey",
+    "ring-tailed lemur",
+    "indri",
+    "Asian elephant",
+    "African bush elephant",
+    "red panda",
+    "giant panda",
+    "snoek fish",
+    "eel",
+    "silver salmon",
+    "rock beauty fish",
+    "clownfish",
+    "sturgeon",
+    "gar fish",
+    "lionfish",
+    "pufferfish",
+    "abacus",
+    "abaya",
+    "academic gown",
+    "accordion",
+    "acoustic guitar",
+    "aircraft carrier",
+    "airliner",
+    "airship",
+    "altar",
+    "ambulance",
+    "amphibious vehicle",
+    "analog clock",
+    "apiary",
+    "apron",
+    "trash can",
+    "assault rifle",
+    "backpack",
+    "bakery",
+    "balance beam",
+    "balloon",
+    "ballpoint pen",
+    "Band-Aid",
+    "banjo",
+    "baluster / handrail",
+    "barbell",
+    "barber chair",
+    "barbershop",
+    "barn",
+    "barometer",
+    "barrel",
+    "wheelbarrow",
+    "baseball",
+    "basketball",
+    "bassinet",
+    "bassoon",
+    "swimming cap",
+    "bath towel",
+    "bathtub",
+    "station wagon",
+    "lighthouse",
+    "beaker",
+    "military hat (bearskin or shako)",
+    "beer bottle",
+    "beer glass",
+    "bell tower",
+    "baby bib",
+    "tandem bicycle",
+    "bikini",
+    "ring binder",
+    "binoculars",
+    "birdhouse",
+    "boathouse",
+    "bobsleigh",
+    "bolo tie",
+    "poke bonnet",
+    "bookcase",
+    "bookstore",
+    "bottle cap",
+    "hunting bow",
+    "bow tie",
+    "brass memorial plaque",
+    "bra",
+    "breakwater",
+    "breastplate",
+    "broom",
+    "bucket",
+    "buckle",
+    "bulletproof vest",
+    "high-speed train",
+    "butcher shop",
+    "taxicab",
+    "cauldron",
+    "candle",
+    "cannon",
+    "canoe",
+    "can opener",
+    "cardigan",
+    "car mirror",
+    "carousel",
+    "tool kit",
+    "cardboard box / carton",
+    "car wheel",
+    "automated teller machine",
+    "cassette",
+    "cassette player",
+    "castle",
+    "catamaran",
+    "CD player",
+    "cello",
+    "mobile phone",
+    "chain",
+    "chain-link fence",
+    "chain mail",
+    "chainsaw",
+    "storage chest",
+    "chiffonier",
+    "bell or wind chime",
+    "china cabinet",
+    "Christmas stocking",
+    "church",
+    "movie theater",
+    "cleaver",
+    "cliff dwelling",
+    "cloak",
+    "clogs",
+    "cocktail shaker",
+    "coffee mug",
+    "coffeemaker",
+    "spiral or coil",
+    "combination lock",
+    "computer keyboard",
+    "candy store",
+    "container ship",
+    "convertible",
+    "corkscrew",
+    "cornet",
+    "cowboy boot",
+    "cowboy hat",
+    "cradle",
+    "construction crane",
+    "crash helmet",
+    "crate",
+    "infant bed",
+    "Crock Pot",
+    "croquet ball",
+    "crutch",
+    "cuirass",
+    "dam",
+    "desk",
+    "desktop computer",
+    "rotary dial telephone",
+    "diaper",
+    "digital clock",
+    "digital watch",
+    "dining table",
+    "dishcloth",
+    "dishwasher",
+    "disc brake",
+    "dock",
+    "dog sled",
+    "dome",
+    "doormat",
+    "drilling rig",
+    "drum",
+    "drumstick",
+    "dumbbell",
+    "Dutch oven",
+    "electric fan",
+    "electric guitar",
+    "electric locomotive",
+    "entertainment center",
+    "envelope",
+    "espresso machine",
+    "face powder",
+    "feather boa",
+    "filing cabinet",
+    "fireboat",
+    "fire truck",
+    "fire screen",
+    "flagpole",
+    "flute",
+    "folding chair",
+    "football helmet",
+    "forklift",
+    "fountain",
+    "fountain pen",
+    "four-poster bed",
+    "freight car",
+    "French horn",
+    "frying pan",
+    "fur coat",
+    "garbage truck",
+    "gas mask or respirator",
+    "gas pump",
+    "goblet",
+    "go-kart",
+    "golf ball",
+    "golf cart",
+    "gondola",
+    "gong",
+    "gown",
+    "grand piano",
+    "greenhouse",
+    "radiator grille",
+    "grocery store",
+    "guillotine",
+    "hair clip",
+    "hair spray",
+    "half-track",
+    "hammer",
+    "hamper",
+    "hair dryer",
+    "hand-held computer",
+    "handkerchief",
+    "hard disk drive",
+    "harmonica",
+    "harp",
+    "combine harvester",
+    "hatchet",
+    "holster",
+    "home theater",
+    "honeycomb",
+    "hook",
+    "hoop skirt",
+    "gymnastic horizontal bar",
+    "horse-drawn vehicle",
+    "hourglass",
+    "iPod",
+    "clothes iron",
+    "carved pumpkin",
+    "jeans",
+    "jeep",
+    "T-shirt",
+    "jigsaw puzzle",
+    "rickshaw",
+    "joystick",
+    "kimono",
+    "knee pad",
+    "knot",
+    "lab coat",
+    "ladle",
+    "lampshade",
+    "laptop computer",
+    "lawn mower",
+    "lens cap",
+    "letter opener",
+    "library",
+    "lifeboat",
+    "lighter",
+    "limousine",
+    "ocean liner",
+    "lipstick",
+    "slip-on shoe",
+    "lotion",
+    "music speaker",
+    "loupe magnifying glass",
+    "sawmill",
+    "magnetic compass",
+    "messenger bag",
+    "mailbox",
+    "tights",
+    "one-piece bathing suit",
+    "manhole cover",
+    "maraca",
+    "marimba",
+    "mask",
+    "matchstick",
+    "maypole",
+    "maze",
+    "measuring cup",
+    "medicine cabinet",
+    "megalith",
+    "microphone",
+    "microwave oven",
+    "military uniform",
+    "milk can",
+    "minibus",
+    "miniskirt",
+    "minivan",
+    "missile",
+    "mitten",
+    "mixing bowl",
+    "mobile home",
+    "ford model t",
+    "modem",
+    "monastery",
+    "monitor",
+    "moped",
+    "mortar and pestle",
+    "graduation cap",
+    "mosque",
+    "mosquito net",
+    "vespa",
+    "mountain bike",
+    "tent",
+    "computer mouse",
+    "mousetrap",
+    "moving van",
+    "muzzle",
+    "metal nail",
+    "neck brace",
+    "necklace",
+    "baby pacifier",
+    "notebook computer",
+    "obelisk",
+    "oboe",
+    "ocarina",
+    "odometer",
+    "oil filter",
+    "pipe organ",
+    "oscilloscope",
+    "overskirt",
+    "bullock cart",
+    "oxygen mask",
+    "product packet / packaging",
+    "paddle",
+    "paddle wheel",
+    "padlock",
+    "paintbrush",
+    "pajamas",
+    "palace",
+    "pan flute",
+    "paper towel",
+    "parachute",
+    "parallel bars",
+    "park bench",
+    "parking meter",
+    "railroad car",
+    "patio",
+    "payphone",
+    "pedestal",
+    "pencil case",
+    "pencil sharpener",
+    "perfume",
+    "Petri dish",
+    "photocopier",
+    "plectrum",
+    "Pickelhaube",
+    "picket fence",
+    "pickup truck",
+    "pier",
+    "piggy bank",
+    "pill bottle",
+    "pillow",
+    "ping-pong ball",
+    "pinwheel",
+    "pirate ship",
+    "drink pitcher",
+    "block plane",
+    "planetarium",
+    "plastic bag",
+    "plate rack",
+    "farm plow",
+    "plunger",
+    "Polaroid camera",
+    "pole",
+    "police van",
+    "poncho",
+    "pool table",
+    "soda bottle",
+    "plant pot",
+    "potter's wheel",
+    "power drill",
+    "prayer rug",
+    "printer",
+    "prison",
+    "missile",
+    "projector",
+    "hockey puck",
+    "punching bag",
+    "purse",
+    "quill",
+    "quilt",
+    "race car",
+    "racket",
+    "radiator",
+    "radio",
+    "radio telescope",
+    "rain barrel",
+    "recreational vehicle",
+    "fishing casting reel",
+    "reflex camera",
+    "refrigerator",
+    "remote control",
+    "restaurant",
+    "revolver",
+    "rifle",
+    "rocking chair",
+    "rotisserie",
+    "eraser",
+    "rugby ball",
+    "ruler measuring stick",
+    "sneaker",
+    "safe",
+    "safety pin",
+    "salt shaker",
+    "sandal",
+    "sarong",
+    "saxophone",
+    "scabbard",
+    "weighing scale",
+    "school bus",
+    "schooner",
+    "scoreboard",
+    "CRT monitor",
+    "screw",
+    "screwdriver",
+    "seat belt",
+    "sewing machine",
+    "shield",
+    "shoe store",
+    "shoji screen / room divider",
+    "shopping basket",
+    "shopping cart",
+    "shovel",
+    "shower cap",
+    "shower curtain",
+    "ski",
+    "balaclava ski mask",
+    "sleeping bag",
+    "slide rule",
+    "sliding door",
+    "slot machine",
+    "snorkel",
+    "snowmobile",
+    "snowplow",
+    "soap dispenser",
+    "soccer ball",
+    "sock",
+    "solar thermal collector",
+    "sombrero",
+    "soup bowl",
+    "keyboard space bar",
+    "space heater",
+    "space shuttle",
+    "spatula",
+    "motorboat",
+    "spider web",
+    "spindle",
+    "sports car",
+    "spotlight",
+    "stage",
+    "steam locomotive",
+    "through arch bridge",
+    "steel drum",
+    "stethoscope",
+    "scarf",
+    "stone wall",
+    "stopwatch",
+    "stove",
+    "strainer",
+    "tram",
+    "stretcher",
+    "couch",
+    "stupa",
+    "submarine",
+    "suit",
+    "sundial",
+    "sunglasses",
+    "sunglasses",
+    "sunscreen",
+    "suspension bridge",
+    "mop",
+    "sweatshirt",
+    "swim trunks / shorts",
+    "swing",
+    "electrical switch",
+    "syringe",
+    "table lamp",
+    "tank",
+    "tape player",
+    "teapot",
+    "teddy bear",
+    "television",
+    "tennis ball",
+    "thatched roof",
+    "front curtain",
+    "thimble",
+    "threshing machine",
+    "throne",
+    "tile roof",
+    "toaster",
+    "tobacco shop",
+    "toilet seat",
+    "torch",
+    "totem pole",
+    "tow truck",
+    "toy store",
+    "tractor",
+    "semi-trailer truck",
+    "tray",
+    "trench coat",
+    "tricycle",
+    "trimaran",
+    "tripod",
+    "triumphal arch",
+    "trolleybus",
+    "trombone",
+    "hot tub",
+    "turnstile",
+    "typewriter keyboard",
+    "umbrella",
+    "unicycle",
+    "upright piano",
+    "vacuum cleaner",
+    "vase",
+    "vaulted or arched ceiling",
+    "velvet fabric",
+    "vending machine",
+    "vestment",
+    "viaduct",
+    "violin",
+    "volleyball",
+    "waffle iron",
+    "wall clock",
+    "wallet",
+    "wardrobe",
+    "military aircraft",
+    "sink",
+    "washing machine",
+    "water bottle",
+    "water jug",
+    "water tower",
+    "whiskey jug",
+    "whistle",
+    "hair wig",
+    "window screen",
+    "window shade",
+    "Windsor tie",
+    "wine bottle",
+    "airplane wing",
+    "wok",
+    "wooden spoon",
+    "wool",
+    "split-rail fence",
+    "shipwreck",
+    "sailboat",
+    "yurt",
+    "website",
+    "comic book",
+    "crossword",
+    "traffic or street sign",
+    "traffic light",
+    "dust jacket",
+    "menu",
+    "plate",
+    "guacamole",
+    "consomme",
+    "hot pot",
+    "trifle",
+    "ice cream",
+    "popsicle",
+    "baguette",
+    "bagel",
+    "pretzel",
+    "cheeseburger",
+    "hot dog",
+    "mashed potatoes",
+    "cabbage",
+    "broccoli",
+    "cauliflower",
+    "zucchini",
+    "spaghetti squash",
+    "acorn squash",
+    "butternut squash",
+    "cucumber",
+    "artichoke",
+    "bell pepper",
+    "cardoon",
+    "mushroom",
+    "Granny Smith apple",
+    "strawberry",
+    "orange",
+    "lemon",
+    "fig",
+    "pineapple",
+    "banana",
+    "jackfruit",
+    "cherimoya (custard apple)",
+    "pomegranate",
+    "hay",
+    "carbonara",
+    "chocolate syrup",
+    "dough",
+    "meatloaf",
+    "pizza",
+    "pot pie",
+    "burrito",
+    "red wine",
+    "espresso",
+    "tea cup",
+    "eggnog",
+    "mountain",
+    "bubble",
+    "cliff",
+    "coral reef",
+    "geyser",
+    "lakeshore",
+    "promontory",
+    "sandbar",
+    "beach",
+    "valley",
+    "volcano",
+    "baseball player",
+    "bridegroom",
+    "scuba diver",
+    "rapeseed",
+    "daisy",
+    "yellow lady's slipper",
+    "corn",
+    "acorn",
+    "rose hip",
+    "horse chestnut seed",
+    "coral fungus",
+    "agaric",
+    "gyromitra",
+    "stinkhorn mushroom",
+    "earth star fungus",
+    "hen of the woods mushroom",
+    "bolete",
+    "corn cob",
+    "toilet paper",
+]
+
+
+openai_imagenet_template = [
+    lambda c: f"a bad photo of a {c}.",
+    lambda c: f"a photo of many {c}.",
+    lambda c: f"a sculpture of a {c}.",
+    lambda c: f"a photo of the hard to see {c}.",
+    lambda c: f"a low resolution photo of the {c}.",
+    lambda c: f"a rendering of a {c}.",
+    lambda c: f"graffiti of a {c}.",
+    lambda c: f"a bad photo of the {c}.",
+    lambda c: f"a cropped photo of the {c}.",
+    lambda c: f"a tattoo of a {c}.",
+    lambda c: f"the embroidered {c}.",
+    lambda c: f"a photo of a hard to see {c}.",
+    lambda c: f"a bright photo of a {c}.",
+    lambda c: f"a photo of a clean {c}.",
+    lambda c: f"a photo of a dirty {c}.",
+    lambda c: f"a dark photo of the {c}.",
+    lambda c: f"a drawing of a {c}.",
+    lambda c: f"a photo of my {c}.",
+    lambda c: f"the plastic {c}.",
+    lambda c: f"a photo of the cool {c}.",
+    lambda c: f"a close-up photo of a {c}.",
+    lambda c: f"a black and white photo of the {c}.",
+    lambda c: f"a painting of the {c}.",
+    lambda c: f"a painting of a {c}.",
+    lambda c: f"a pixelated photo of the {c}.",
+    lambda c: f"a sculpture of the {c}.",
+    lambda c: f"a bright photo of the {c}.",
+    lambda c: f"a cropped photo of a {c}.",
+    lambda c: f"a plastic {c}.",
+    lambda c: f"a photo of the dirty {c}.",
+    lambda c: f"a jpeg corrupted photo of a {c}.",
+    lambda c: f"a blurry photo of the {c}.",
+    lambda c: f"a photo of the {c}.",
+    lambda c: f"a good photo of the {c}.",
+    lambda c: f"a rendering of the {c}.",
+    lambda c: f"a {c} in a video game.",
+    lambda c: f"a photo of one {c}.",
+    lambda c: f"a doodle of a {c}.",
+    lambda c: f"a close-up photo of the {c}.",
+    lambda c: f"a photo of a {c}.",
+    lambda c: f"the origami {c}.",
+    lambda c: f"the {c} in a video game.",
+    lambda c: f"a sketch of a {c}.",
+    lambda c: f"a doodle of the {c}.",
+    lambda c: f"a origami {c}.",
+    lambda c: f"a low resolution photo of a {c}.",
+    lambda c: f"the toy {c}.",
+    lambda c: f"a rendition of the {c}.",
+    lambda c: f"a photo of the clean {c}.",
+    lambda c: f"a photo of a large {c}.",
+    lambda c: f"a rendition of a {c}.",
+    lambda c: f"a photo of a nice {c}.",
+    lambda c: f"a photo of a weird {c}.",
+    lambda c: f"a blurry photo of a {c}.",
+    lambda c: f"a cartoon {c}.",
+    lambda c: f"art of a {c}.",
+    lambda c: f"a sketch of the {c}.",
+    lambda c: f"a embroidered {c}.",
+    lambda c: f"a pixelated photo of a {c}.",
+    lambda c: f"itap of the {c}.",
+    lambda c: f"a jpeg corrupted photo of the {c}.",
+    lambda c: f"a good photo of a {c}.",
+    lambda c: f"a plushie {c}.",
+    lambda c: f"a photo of the nice {c}.",
+    lambda c: f"a photo of the small {c}.",
+    lambda c: f"a photo of the weird {c}.",
+    lambda c: f"the cartoon {c}.",
+    lambda c: f"art of the {c}.",
+    lambda c: f"a drawing of the {c}.",
+    lambda c: f"a photo of the large {c}.",
+    lambda c: f"a black and white photo of a {c}.",
+    lambda c: f"the plushie {c}.",
+    lambda c: f"a dark photo of a {c}.",
+    lambda c: f"itap of a {c}.",
+    lambda c: f"graffiti of the {c}.",
+    lambda c: f"a toy {c}.",
+    lambda c: f"itap of my {c}.",
+    lambda c: f"a photo of a cool {c}.",
+    lambda c: f"a photo of a small {c}.",
+    lambda c: f"a tattoo of the {c}.",
+]
diff --git a/picoaudio/audioldm/clap/training/infer_demo.py b/picoaudio/audioldm/clap/training/infer_demo.py
new file mode 100644
index 0000000000000000000000000000000000000000..7d1f4784898dbfeb69affefb6f624711adc8cb42
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/infer_demo.py
@@ -0,0 +1,105 @@
+import sys
+
+import os
+import torch
+import librosa
+from open_clip import create_model
+from training.data import get_audio_features
+from training.data import int16_to_float32, float32_to_int16
+from transformers import RobertaTokenizer
+
+tokenize = RobertaTokenizer.from_pretrained("roberta-base")
+
+
+def tokenizer(text):
+    result = tokenize(
+        text,
+        padding="max_length",
+        truncation=True,
+        max_length=77,
+        return_tensors="pt",
+    )
+    return {k: v.squeeze(0) for k, v in result.items()}
+
+
+PRETRAINED_PATH = "/mnt/fast/nobackup/users/hl01486/projects/contrastive_pretraining/CLAP/assets/checkpoints/epoch_top_0_audioset_no_fusion.pt"
+WAVE_48k_PATH = "/mnt/fast/nobackup/users/hl01486/projects/contrastive_pretraining/CLAP/assets/audio/machine.wav"
+
+
+def infer_text():
+    device = "cuda:0" if torch.cuda.is_available() else "cpu"
+    precision = "fp32"
+    amodel = "HTSAT-tiny"  # or 'PANN-14'
+    tmodel = "roberta"  # the best text encoder in our training
+    enable_fusion = False  # False if you do not want to use the fusion model
+    fusion_type = "aff_2d"
+    pretrained = PRETRAINED_PATH
+
+    model, model_cfg = create_model(
+        amodel,
+        tmodel,
+        pretrained,
+        precision=precision,
+        device=device,
+        enable_fusion=enable_fusion,
+        fusion_type=fusion_type,
+    )
+    # load the text, can be a list (i.e. batch size)
+    text_data = ["I love the contrastive learning", "I love the pretrain model"]
+    # tokenize for roberta, if you want to tokenize for another text encoder, please refer to data.py#L43-90
+    text_data = tokenizer(text_data)
+
+    text_embed = model.get_text_embedding(text_data)
+    print(text_embed.size())
+
+
+def infer_audio():
+
+    device = "cuda:0" if torch.cuda.is_available() else "cpu"
+    precision = "fp32"
+    amodel = "HTSAT-tiny"  # or 'PANN-14'
+    tmodel = "roberta"  # the best text encoder in our training
+    enable_fusion = False  # False if you do not want to use the fusion model
+    fusion_type = "aff_2d"
+    pretrained = PRETRAINED_PATH
+
+    model, model_cfg = create_model(
+        amodel,
+        tmodel,
+        pretrained,
+        precision=precision,
+        device=device,
+        enable_fusion=enable_fusion,
+        fusion_type=fusion_type,
+    )
+
+    # load the waveform of the shape (T,), should resample to 48000
+    audio_waveform, sr = librosa.load(WAVE_48k_PATH, sr=48000)
+    # quantize
+    audio_waveform = int16_to_float32(float32_to_int16(audio_waveform))
+    audio_waveform = torch.from_numpy(audio_waveform).float()
+    audio_dict = {}
+
+    # the 'fusion' truncate mode can be changed to 'rand_trunc' if run in unfusion mode
+    import ipdb
+
+    ipdb.set_trace()
+    audio_dict = get_audio_features(
+        audio_dict,
+        audio_waveform,
+        480000,
+        data_truncating="fusion",
+        data_filling="repeatpad",
+        audio_cfg=model_cfg["audio_cfg"],
+    )
+    # can send a list to the model, to process many audio tracks in one time (i.e. batch size)
+    audio_embed = model.get_audio_embedding([audio_dict])
+    print(audio_embed.size())
+    import ipdb
+
+    ipdb.set_trace()
+
+
+if __name__ == "__main__":
+    infer_text()
+    infer_audio()
diff --git a/picoaudio/audioldm/clap/training/logger.py b/picoaudio/audioldm/clap/training/logger.py
new file mode 100644
index 0000000000000000000000000000000000000000..ac4634970fae6aacde2b7b808355dbd50c90ce73
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/logger.py
@@ -0,0 +1,30 @@
+import logging
+
+
+def setup_logging(log_file, level, include_host=False):
+    if include_host:
+        import socket
+
+        hostname = socket.gethostname()
+        formatter = logging.Formatter(
+            f"%(asctime)s |  {hostname} | %(levelname)s | %(message)s",
+            datefmt="%Y-%m-%d,%H:%M:%S",
+        )
+    else:
+        formatter = logging.Formatter(
+            "%(asctime)s | %(levelname)s | %(message)s", datefmt="%Y-%m-%d,%H:%M:%S"
+        )
+
+    logging.root.setLevel(level)
+    loggers = [logging.getLogger(name) for name in logging.root.manager.loggerDict]
+    for logger in loggers:
+        logger.setLevel(level)
+
+    stream_handler = logging.StreamHandler()
+    stream_handler.setFormatter(formatter)
+    logging.root.addHandler(stream_handler)
+
+    if log_file:
+        file_handler = logging.FileHandler(filename=log_file)
+        file_handler.setFormatter(formatter)
+        logging.root.addHandler(file_handler)
diff --git a/picoaudio/audioldm/clap/training/lp_main.py b/picoaudio/audioldm/clap/training/lp_main.py
new file mode 100644
index 0000000000000000000000000000000000000000..c2d4e8c85aaa3c8e4221963ef56a815cc14f354f
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/lp_main.py
@@ -0,0 +1,670 @@
+from cmath import cos
+from inspect import getargs
+import logging
+import os
+import random
+from datetime import datetime
+import bisect
+import copy
+from sched import scheduler
+import numpy as np
+import torch
+import torch.backends.cudnn as cudnn
+from torch import optim
+from torch.cuda.amp import GradScaler
+import faulthandler
+import pathlib
+import argparse
+import time
+
+try:
+    import wandb
+except ImportError:
+    wandb = None
+
+try:
+    import torch.utils.tensorboard as tensorboard
+except ImportError:
+    tensorboard = None
+
+try:
+    import horovod.torch as hvd
+except ImportError:
+    hvd = None
+
+from open_clip import create_model_and_transforms, trace_model, create_model
+from training.data import get_data
+from training.params import parse_args
+from training.distributed import is_master, init_distributed_device, world_info_from_env
+from training.logger import setup_logging
+from training.scheduler import cosine_lr
+from training.lp_train import train_one_epoch, evaluate
+from open_clip.utils import get_tar_path_from_dataset_name, dataset_split, get_optimizer
+from open_clip.utils import load_p, load_class_label
+from open_clip.linear_probe import LinearProbe
+
+
+def maintain_ckpts(args, startidx, all_idx_len):
+    for i in reversed(range(startidx, all_idx_len)):
+        if os.path.exists(os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt")):
+            os.rename(
+                os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt"),
+                os.path.join(args.checkpoint_path, f"epoch_top_{i+1}.pt"),
+            )
+    if os.path.exists(
+        os.path.join(args.checkpoint_path, f"epoch_top_{all_idx_len}.pt")
+    ):
+        os.remove(os.path.join(args.checkpoint_path, f"epoch_top_{all_idx_len}.pt"))
+    return
+
+
+def update_top_k_performance(
+    new_metrics_inputs, current_top_k_ckpt_metrics, args, ckpt, bignumbetter=True
+):
+    """
+    Record the top-k performance of the current epoch.
+    current_top_k_metrics is a dictionary of the form: {1: top_1_ckpt_measure, 2: top_2_ckpt_measure, ...}
+    """
+    if isinstance(new_metrics_inputs, (list, tuple)):
+        new_metrics_inputs = np.mean(new_metrics_inputs)
+        return update_top_k_performance(
+            new_metrics_inputs,
+            current_top_k_ckpt_metrics,
+            args=args,
+            ckpt=ckpt,
+            bignumbetter=bignumbetter,
+        )
+    elif isinstance(new_metrics_inputs, dict):
+        new_metrics_inputs = np.mean(list(new_metrics_inputs.values()))
+        return update_top_k_performance(
+            new_metrics_inputs,
+            current_top_k_ckpt_metrics,
+            args=args,
+            ckpt=ckpt,
+            bignumbetter=bignumbetter,
+        )
+    elif isinstance(new_metrics_inputs, (float, int)):
+        update_flag = {k: False for k in current_top_k_ckpt_metrics.keys()}
+        sorted_keys = sorted(current_top_k_ckpt_metrics.keys())
+        sorted_values = sorted(
+            current_top_k_ckpt_metrics.values(), reverse=bignumbetter
+        )
+        sorted_values_ = copy.deepcopy(sorted_values)
+        sorted_values.append(new_metrics_inputs)
+        sorted_values = sorted(sorted_values, reverse=bignumbetter)
+        sorted_values = sorted_values[:-1]
+
+        if sorted_values == sorted_values_:
+            return current_top_k_ckpt_metrics, new_metrics_inputs
+        else:
+            for i in range(len(sorted_keys)):
+                if current_top_k_ckpt_metrics[sorted_keys[i]] != sorted_values[i]:
+                    current_top_k_ckpt_metrics[sorted_keys[i]] = sorted_values[i]
+                    update_flag[sorted_keys[i]] = True
+            for i in range(len(update_flag)):
+                if update_flag[i]:
+                    maintain_ckpts(args, i, len(sorted_keys))
+                    torch.save(
+                        ckpt,
+                        os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt"),
+                    )
+                    break
+            return current_top_k_ckpt_metrics, new_metrics_inputs
+
+
+# def updateifNone(a, b):
+#     a = b if None else a
+#     return a
+
+
+def is_pretrained_params(n):
+    return (
+        n.startswith("clap_model.transformer")
+        or n in ["clap_model.positional_embedding", "clap_model.text_projection"]
+        or n.startswith("clap_model.token_embedding")
+        or n.startswith("clap_model.ln_final")
+        or n.startswith("clap_model.logit_scale_t")
+    )
+
+
+def random_seed(seed=42, rank=0):
+    torch.manual_seed(seed + rank)
+    np.random.seed(seed + rank)
+    random.seed(seed + rank)
+
+
+def config_lp_optimizer(model, data, args):
+    # set wd-related params to 0 if use adam optimizer
+    if args.optimizer == "adam":
+        args.wd = 0
+        args.wd_pretrained = 0
+        args.wd_new = 0
+
+    in_clap = lambda n, p: n.startswith("clap_model")
+
+    named_parameters = list(model.named_parameters())
+
+    optimizer = {}
+    scheduler = {}
+
+    # freeze text encoder
+    text_freeze_parameters = [
+        p
+        for n, p in named_parameters
+        if n.startswith("clap_model.transformer")
+        or n in ["clap_model.positional_embedding", "clap_model.text_projection"]
+        or n.startswith("clap_model.token_embedding")
+        or n.startswith("clap_model.ln_final")
+    ]
+
+    if args.freeze_text:
+        logging.info("Freeze Text!!!!")
+        for k in text_freeze_parameters:
+            k.requires_grad = False
+
+    if not args.lp_freeze:
+        exclude = (
+            lambda n, p: p.ndim < 2
+            or "bn" in n
+            or "ln" in n
+            or "bias" in n
+            or "logit_scale" in n
+        )
+        include = lambda n, p: not exclude(n, p)
+
+        # (yusong): we do not split the learning rate anymore
+        # p for n, p in named_parameters if in_clap(n,p) and exclude(n, p) and p.requires_grad
+        gain_or_bias_params = [
+            p for n, p in named_parameters if exclude(n, p) and p.requires_grad
+        ]
+        # rest_params = [p for n, p in named_parameters if in_clap(n,p) and include(n, p) and p.requires_grad]
+        rest_params = [
+            p for n, p in named_parameters if include(n, p) and p.requires_grad
+        ]
+
+        if args.train_data is None:
+            optimizer = None
+            scheduler = None
+        else:
+            total_steps = data["train"].dataloader.num_batches * args.epochs
+
+            if args.split_opt:
+                for x in ["lr", "beta1", "beta2", "eps", "wd"]:
+                    for y in ["_new", "_pretrained"]:
+                        if getattr(args, x + y) is None:
+                            setattr(args, x + y, getattr(args, x))
+
+                gain_or_bias_pretrained_params = [
+                    p
+                    for n, p in named_parameters
+                    if (exclude(n, p) and p.requires_grad) and is_pretrained_params(n)
+                ]
+                rest_pretrained_params = [
+                    p
+                    for n, p in named_parameters
+                    if (include(n, p) and p.requires_grad) and is_pretrained_params(n)
+                ]
+                gain_or_bias_new_params = [
+                    p
+                    for n, p in named_parameters
+                    if (exclude(n, p) and p.requires_grad)
+                    and (not is_pretrained_params(n))
+                ]
+                rest_new_params = [
+                    p
+                    for n, p in named_parameters
+                    if (include(n, p) and p.requires_grad)
+                    and (not is_pretrained_params(n))
+                ]
+
+                pretrained_params_optimizer = get_optimizer(
+                    [
+                        {"params": gain_or_bias_pretrained_params, "weight_decay": 0.0},
+                        {
+                            "params": rest_pretrained_params,
+                            "weight_decay": args.wd_pretrained,
+                        },
+                    ],
+                    lr=args.lr_pretrained,
+                    betas=(args.beta1_pretrained, args.beta2_pretrained),
+                    eps=args.eps_pretrained,
+                    momentum=args.momentum_pretrained,
+                    optimizer_name=args.optimizer,
+                )
+                pretrained_params_scheduler = cosine_lr(
+                    pretrained_params_optimizer,
+                    args.lr_pretrained,
+                    args.warmup,
+                    total_steps,
+                )
+
+                new_params_optimizer = get_optimizer(
+                    [
+                        {"params": gain_or_bias_new_params, "weight_decay": 0.0},
+                        {"params": rest_new_params, "weight_decay": args.wd_new},
+                    ],
+                    lr=args.lr_new,
+                    betas=(args.beta1_new, args.beta2_new),
+                    eps=args.eps_new,
+                    momentum=args.momentum_new,
+                    optimizer_name=args.optimizer,
+                )
+                new_params_scheduler = cosine_lr(
+                    new_params_optimizer, args.lr_new, args.warmup, total_steps
+                )
+
+                optimizer["text"] = pretrained_params_optimizer
+                optimizer["audio"] = new_params_optimizer
+                scheduler["text"] = pretrained_params_scheduler
+                scheduler["audio"] = new_params_scheduler
+
+                if args.horovod:
+                    pretrained_params_optimizer = hvd.DistributedOptimizer(
+                        pretrained_params_optimizer,
+                        named_parameters=model.named_parameters(),
+                    )
+                    new_params_optimizer = hvd.DistributedOptimizer(
+                        new_params_optimizer, named_parameters=model.named_parameters()
+                    )
+                    hvd.broadcast_parameters(model.state_dict(), root_rank=0)
+                    hvd.broadcast_optimizer_state(
+                        pretrained_params_optimizer, root_rank=0
+                    )
+                    hvd.broadcast_optimizer_state(new_params_optimizer, root_rank=0)
+            else:
+
+                optimizer["clap"] = get_optimizer(
+                    [
+                        {"params": gain_or_bias_params, "weight_decay": 0.0},
+                        {"params": rest_params, "weight_decay": args.wd},
+                    ],
+                    lr=args.lr,
+                    betas=(args.beta1, args.beta2),
+                    eps=args.eps,
+                    momentum=args.momentum,
+                    optimizer_name=args.optimizer,
+                )
+                scheduler["clap"] = cosine_lr(
+                    optimizer["clap"], args.lr, args.warmup, total_steps
+                )
+
+                if args.horovod:
+                    optimizer["clap"] = hvd.DistributedOptimizer(
+                        optimizer["clap"], named_parameters=model.named_parameters()
+                    )
+                    hvd.broadcast_parameters(model.state_dict(), root_rank=0)
+                    hvd.broadcast_optimizer_state(optimizer["clap"], root_rank=0)
+
+    # linear probe optimizer
+    else:
+        lp_params = [
+            p for n, p in named_parameters if (not in_clap(n, p)) and p.requires_grad
+        ]
+        lp_optim = get_optimizer(
+            lp_params,
+            lr=args.lp_lr,
+            betas=(args.beta1, args.beta2),
+            eps=args.eps,
+            momentum=0.9,
+            optimizer_name=args.optimizer,
+        )
+        optimizer["lp"] = lp_optim
+
+    return optimizer, scheduler, text_freeze_parameters
+
+
+def main():
+    args = parse_args()
+
+    time.sleep(args.sleep)
+
+    # sanitize model name for filesystem / uri use, easier if we don't use / in name as a rule?
+    args.amodel = args.amodel.replace("/", "-")
+    # download sizes.json file
+
+    # (yusong): the below two lines are for debug
+    # print("setting up faulthandler")
+    # faulthandler.register(10)
+
+    random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.cuda.manual_seed(args.seed)
+    torch.cuda.manual_seed_all(args.seed)
+    np.random.seed(args.seed)
+    args.class_index_dict = load_class_label(args.class_label_path)
+
+    # get the name of the experiments
+    if args.name is None:
+        args.name = "-".join(
+            [
+                datetime.now().strftime("%Y_%m_%d-%H_%M_%S"),
+                f"linear_probe" f"model_{args.amodel}",
+                f"lr_{args.lr}",
+                f"b_{args.batch_size}",
+                f"j_{args.workers}",
+                f"p_{args.precision}",
+            ]
+        )
+
+    # discover initial world args early so we can log properly
+    args.distributed = False
+    args.local_rank, args.rank, args.world_size = world_info_from_env()
+
+    if args.remotedata and is_master(args):
+        for dataset_name in args.datasetnames:
+            for split in dataset_split[dataset_name]:
+                if not os.path.exists(f"./json_files/{dataset_name}/{split}"):
+                    os.makedirs(f"./json_files/{dataset_name}/{split}")
+                os.system(
+                    f"aws s3 cp s3://s-laion-audio/webdataset_tar/{dataset_name}/{split}/sizes.json ./json_files/{dataset_name}/{split}/sizes.json"
+                )
+
+    args.log_path = None
+    if is_master(args, local=args.log_local):
+        log_base_path = os.path.join(args.logs, args.name)
+        os.makedirs(log_base_path, exist_ok=True)
+        log_filename = f"out-{args.rank}" if args.log_local else "out.log"
+        args.log_path = os.path.join(log_base_path, log_filename)
+
+        # avoid log dir in same name:
+        postfix = 0
+        while os.path.exists(args.log_path):
+            postfix += 1
+            log_base_path_new = log_base_path + "-" + str(postfix)
+            os.makedirs(log_base_path_new, exist_ok=True)
+            log_filename = f"out-{args.rank}" if args.log_local else "out.log"
+            args.log_path = os.path.join(log_base_path_new, log_filename)
+            # print(
+            #     "Error. Experiment already exists. Use --name {} to specify a new experiment."
+            # )
+            # return -1
+
+    # Set logger
+    args.log_level = logging.DEBUG if args.debug else logging.INFO
+    setup_logging(args.log_path, args.log_level)
+
+    # fully initialize distributed device environment
+    device = init_distributed_device(args)
+
+    args.wandb = "wandb" in args.report_to or "all" in args.report_to
+    args.tensorboard = "tensorboard" in args.report_to or "all" in args.report_to
+    if is_master(args):
+        args.tensorboard_path = (
+            os.path.join(args.logs, args.name, "tensorboard")
+            if args.tensorboard
+            else ""
+        )
+        args.checkpoint_path = os.path.join(args.logs, args.name, "checkpoints")
+        for dirname in [args.tensorboard_path, args.checkpoint_path]:
+            if dirname:
+                os.makedirs(dirname, exist_ok=True)
+    else:
+        args.tensorboard_path = ""
+        args.checkpoint_path = ""
+
+    if args.copy_codebase:
+        copy_codebase(args)
+
+    assert args.precision in ["amp", "fp16", "fp32"]
+    if args.precision == "fp16":
+        logging.warning(
+            "It is recommended to use AMP mixed-precision instead of FP16. "
+            "FP16 support needs further verification and tuning, especially for train."
+        )
+
+    if args.horovod:
+        logging.info(
+            f"Running in horovod mode with multiple processes / nodes. Device: {args.device}."
+            f"Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}."
+        )
+    elif args.distributed:
+        logging.info(
+            f"Running in distributed mode with multiple processes. Device: {args.device}."
+            f"Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}."
+        )
+    else:
+        logging.info(f"Running with a single process. Device {args.device}.")
+
+    logging.info(f"openai cache dir: {os.path.expanduser(args.openai_model_cache_dir)}")
+
+    # Create CLAP model
+    clap_model, clap_model_cfg = create_model(
+        args.amodel,
+        args.tmodel,
+        args.pretrained,
+        precision=args.precision,
+        device=device,
+        jit=args.torchscript,
+        force_quick_gelu=args.force_quick_gelu,
+        openai_model_cache_dir=os.path.expanduser(args.openai_model_cache_dir),
+        skip_params=False,
+        pretrained_audio=args.pretrained_audio,
+        pretrained_text=args.pretrained_text,
+        enable_fusion=args.enable_fusion,
+        fusion_type=args.fusion_type,
+    )
+
+    args.lp_out_ch = len(list(args.class_index_dict.keys()))
+    # Linear Probe
+    logging.info(f"linear probe using mlp: {args.lp_mlp}")
+    logging.info(f"linear probe using freeze: {args.lp_freeze}")
+    logging.info(f"linear probe act layer: {args.lp_act}")
+    logging.info(f"linear probe out ch: {args.lp_out_ch}")
+    logging.info(f"linear probe learning rate (if applicable): {args.lp_lr}")
+    logging.info(f"linear probe loss func: {args.lp_loss}")
+    logging.info(f"linear probe lp_metrics: {args.lp_metrics}")
+
+    model = LinearProbe(
+        clap_model,
+        mlp=args.lp_mlp,
+        freeze=args.lp_freeze,
+        in_ch=512,
+        out_ch=args.lp_out_ch,
+        act=args.lp_act,
+    )  # in_ch is fixed (i.e., 512)
+    model = model.to(device)
+
+    if args.horovod:
+        with torch.no_grad():
+            for param in model.parameters():
+                param.set_(param.contiguous())
+
+    if args.trace:
+        model = trace_model(model, batch_size=args.batch_size, device=device)
+
+    if is_master(args):
+        logging.info("Linear Probe CLAP Model:")
+        logging.info(f"{str(clap_model)}")
+        logging.info("Params:")
+        params_file = os.path.join(args.logs, args.name, "params.txt")
+        with open(params_file, "w") as f:
+            for name in sorted(vars(args)):
+                val = getattr(args, name)
+                logging.info(f"  {name}: {val}")
+                f.write(f"{name}: {val}\n")
+
+    if args.distributed and not args.horovod:
+        if args.use_bn_sync:
+            model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
+        ddp_args = {}
+        if args.ddp_static_graph:
+            # this doesn't exist in older PyTorch, arg only added if enabled
+            ddp_args["static_graph"] = True
+        model = torch.nn.parallel.DistributedDataParallel(
+            model, device_ids=[device], find_unused_parameters=True, **ddp_args
+        )
+
+    data = get_data(args, clap_model_cfg)
+    assert len(data), "At least one train or eval dataset must be specified."
+    if args.trace:
+        assert "train" not in data, "Cannot train with traced model"
+
+    optimizer, scheduler, text_freeze_parameters = config_lp_optimizer(
+        model, data, args
+    )
+
+    scaler = GradScaler() if args.precision == "amp" else None
+
+    # optionally resume from a checkpoint
+    start_epoch = 0
+    if args.resume is not None:
+        if os.path.isfile(args.resume):
+            checkpoint = torch.load(args.resume, map_location=device)
+            if "epoch" in checkpoint:
+                # resuming a train checkpoint w/ epoch and optimizer state
+                start_epoch = checkpoint["epoch"]
+                sd = checkpoint["state_dict"]
+                if not args.distributed and next(iter(sd.items()))[0].startswith(
+                    "module"
+                ):
+                    sd = {k[len("module.") :]: v for k, v in sd.items()}
+                model.load_state_dict(sd)
+                if args.split_opt:
+                    if optimizer is not None:
+                        for k, o_ in optimizer.items():
+                            o_.load_state_dict(checkpoint[k + "_" + "optimizer"])
+                if optimizer is not None:
+                    optimizer.load_state_dict(checkpoint["optimizer"])
+                if scaler is not None and "scaler" in checkpoint:
+                    scaler.load_state_dict(checkpoint["scaler"])
+                logging.info(
+                    f"=> resuming checkpoint '{args.resume}' (epoch {start_epoch})"
+                )
+            else:
+                # loading a bare (model only) checkpoint for fine-tune or evaluation
+                model.load_state_dict(checkpoint)
+                logging.info(
+                    f"=> loaded checkpoint '{args.resume}' (epoch {start_epoch})"
+                )
+            if args.freeze_text:
+                print("Freeze Text!!!!")
+                for k in text_freeze_parameters:
+                    k.requires_grad = False
+        else:
+            logging.info("=> no checkpoint found at '{}'".format(args.resume))
+
+    cudnn.benchmark = True
+    cudnn.deterministic = False
+
+    # determine if this worker should save logs and checkpoints. only do so if it is rank == 0
+    args.save_logs = args.logs and args.logs.lower() != "none" and is_master(args)
+    writer = None
+    if args.save_logs and args.tensorboard:
+        assert tensorboard is not None, "Please install tensorboard."
+        writer = tensorboard.SummaryWriter(args.tensorboard_path)
+
+    if args.wandb and is_master(args):
+        assert wandb is not None, "Please install wandb."
+        logging.debug("Starting wandb.")
+        args.train_sz = data["train"].dataloader.num_samples
+        if args.val_data is not None:
+            args.val_sz = data["val"].dataloader.num_samples
+        # you will have to configure this for your project!
+        wandb.init(
+            project="clap",
+            notes=args.wandb_notes,
+            name=args.wandb_notes,
+            tags=[],
+            config=vars(args),
+        )
+        if args.debug:
+            wandb.watch(model, log="all")
+        wandb.save(params_file)
+        logging.debug("Finished loading wandb.")
+
+    if "train" not in data:
+        evaluate(model, data, start_epoch, args, writer)
+        return
+    elif start_epoch == 0 and "val" in data and not args.no_eval:
+        evaluate(model, data, 0, args, writer)
+    if args.save_top_performance:
+        current_top_k_ckpt_metrics = {
+            i: 0 for i in range(args.save_top_performance)
+        }  # initialize the top-k metric for ckpts to 0
+
+    for epoch in range(start_epoch, args.epochs):
+        # freeze the text param after (include) args.freeze_text_after, this is -1 by default
+        if epoch == args.freeze_text_after:
+            print("Text pretrained parameters are freezed since this epoch.")
+            for k in text_freeze_parameters:
+                k.requires_grad = False
+        if is_master(args):
+            logging.info(f"Start epoch {epoch}")
+
+        train_one_epoch(model, data, epoch, optimizer, scaler, scheduler, args, writer)
+        completed_epoch = epoch + 1
+
+        if (
+            any(v in data for v in ("val", "imagenet-val", "imagenet-v2"))
+            and not args.no_eval
+        ):
+            metrics = evaluate(model, data, completed_epoch, args, writer)
+            if args.save_top_performance:
+                top_k_dataset = args.top_k_checkpoint_select_dataset
+                top_k_metric = args.top_k_checkpoint_select_metric
+                filtered_metrics = [
+                    v
+                    for k, v in metrics.items()
+                    if top_k_metric in k and top_k_dataset in k
+                ]  # check all R@10 metrics (all dataset) and use it to update the ckpt
+        # Saving checkpoints.
+        if args.save_logs:
+            opt_dict = {
+                k + "_" + "optimizer": v.state_dict() for k, v in optimizer.items()
+            }
+            checkpoint_dict = {
+                "epoch": completed_epoch,
+                "name": args.name,
+                "state_dict": model.state_dict(),
+            }
+            checkpoint_dict.update(opt_dict)
+            if scaler is not None:
+                checkpoint_dict["scaler"] = scaler.state_dict()
+
+            if completed_epoch == args.epochs or (
+                args.save_frequency > 0 and (completed_epoch % args.save_frequency) == 0
+            ):
+                torch.save(
+                    checkpoint_dict,
+                    os.path.join(args.checkpoint_path, f"epoch_{completed_epoch}.pt"),
+                )
+            if args.save_most_recent:
+                torch.save(
+                    checkpoint_dict,
+                    os.path.join(args.checkpoint_path, f"epoch_latest.pt"),
+                )
+            if args.save_top_performance and not args.no_eval:
+                update_top_k_performance(
+                    filtered_metrics,
+                    current_top_k_ckpt_metrics,
+                    args,
+                    checkpoint_dict,
+                    bignumbetter=True,
+                )
+
+    if args.wandb and is_master(args):
+        wandb.finish()
+
+
+def copy_codebase(args):
+    from shutil import copytree, ignore_patterns
+
+    new_code_path = os.path.join(args.logs, args.name, "code")
+    if os.path.exists(new_code_path):
+        print(
+            f"Error. Experiment already exists at {new_code_path}. Use --name to specify a new experiment."
+        )
+        return -1
+    print(f"Copying codebase to {new_code_path}")
+    current_code_path = os.path.realpath(__file__)
+    for _ in range(3):
+        current_code_path = os.path.dirname(current_code_path)
+    copytree(
+        current_code_path, new_code_path, ignore=ignore_patterns("log", "logs", "wandb")
+    )
+    print("Done copying code.")
+    return 1
+
+
+if __name__ == "__main__":
+    main()
diff --git a/picoaudio/audioldm/clap/training/lp_train.py b/picoaudio/audioldm/clap/training/lp_train.py
new file mode 100644
index 0000000000000000000000000000000000000000..24a19bacd0a4b789415cfccbce1f8bc99bc493ed
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/lp_train.py
@@ -0,0 +1,301 @@
+import json
+import logging
+import math
+import os
+import time
+from contextlib import suppress
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+
+try:
+    import wandb
+except ImportError:
+    wandb = None
+
+from open_clip import LPLoss, LPMetrics, lp_gather_features
+from open_clip.utils import do_mixup, get_mix_lambda
+from .distributed import is_master
+from .zero_shot import zero_shot_eval
+
+
+class AverageMeter(object):
+    """Computes and stores the average and current value"""
+
+    def __init__(self):
+        self.reset()
+
+    def reset(self):
+        self.val = 0
+        self.avg = 0
+        self.sum = 0
+        self.count = 0
+
+    def update(self, val, n=1):
+        self.val = val
+        self.sum += val * n
+        self.count += n
+        self.avg = self.sum / self.count
+
+
+def unwrap_model(model):
+    if hasattr(model, "module"):
+        return model.module
+    else:
+        return model
+
+
+def train_one_epoch(
+    model,
+    data,
+    epoch,
+    optimizer,
+    scaler,
+    scheduler,
+    args,
+    tb_writer=None,
+    extra_suffix="",
+):
+    device = torch.device(args.device)
+    autocast = torch.cuda.amp.autocast if args.precision == "amp" else suppress
+    model.train()
+    loss = LPLoss(args.lp_loss)
+
+    dataloader, sampler = data["train"].dataloader, data["train"].sampler
+    if args.distributed and sampler is not None:
+        sampler.set_epoch(epoch)
+    num_batches_per_epoch = dataloader.num_batches
+    sample_digits = math.ceil(math.log(dataloader.num_samples + 1, 10))
+
+    # for toy dataset
+    if args.dataset_type == "toy":
+        dataloader.dataset.generate_queue()
+
+    loss_m = AverageMeter()
+    batch_time_m = AverageMeter()
+    data_time_m = AverageMeter()
+    end = time.time()
+
+    for i, batch in enumerate(dataloader):
+        step = num_batches_per_epoch * epoch + i
+
+        if isinstance(scheduler, dict):
+            for s in scheduler.values():
+                s(step)
+        else:
+            scheduler(step)
+
+        audio = batch  # contains mel_spec, wavform, and longer list
+        class_label = batch["class_label"]
+        # audio = audio.to(device=device, non_blocking=True)
+        class_label = class_label.to(device=device, non_blocking=True)
+
+        if args.mixup:
+            # https://github.com/RetroCirce/HTS-Audio-Transformer/blob/main/utils.py#L146
+            mix_lambda = torch.from_numpy(
+                get_mix_lambda(0.5, len(audio["waveform"]))
+            ).to(device)
+            class_label = do_mixup(class_label, mix_lambda)
+        else:
+            mix_lambda = None
+
+        data_time_m.update(time.time() - end)
+        if isinstance(optimizer, dict):
+            for o_ in optimizer.values():
+                o_.zero_grad()
+        else:
+            optimizer.zero_grad()
+
+        with autocast():
+            pred = model(audio, mix_lambda=mix_lambda, device=device)
+            total_loss = loss(pred, class_label)
+
+        if isinstance(optimizer, dict):
+            if scaler is not None:
+                scaler.scale(total_loss).backward()
+                for o_ in optimizer.values():
+                    if args.horovod:
+                        o_.synchronize()
+                        scaler.unscale_(o_)
+                        with o_.skip_synchronize():
+                            scaler.step(o_)
+                    else:
+                        scaler.step(o_)
+                scaler.update()
+            else:
+                total_loss.backward()
+                for o_ in optimizer.values():
+                    o_.step()
+        else:
+            if scaler is not None:
+                scaler.scale(total_loss).backward()
+                if args.horovod:
+                    optimizer.synchronize()
+                    scaler.unscale_(optimizer)
+                    with optimizer.skip_synchronize():
+                        scaler.step(optimizer)
+                else:
+                    scaler.step(optimizer)
+                scaler.update()
+            else:
+                total_loss.backward()
+                optimizer.step()
+
+        # Note: we clamp to 4.6052 = ln(100), as in the original paper.
+        with torch.no_grad():
+            unwrap_model(model).clap_model.logit_scale_a.clamp_(0, math.log(100))
+            unwrap_model(model).clap_model.logit_scale_t.clamp_(0, math.log(100))
+
+        batch_time_m.update(time.time() - end)
+        end = time.time()
+        batch_count = i + 1
+
+        if is_master(args) and (i % 100 == 0 or batch_count == num_batches_per_epoch):
+            if isinstance(audio, dict):
+                batch_size = len(audio["waveform"])
+            else:
+                batch_size = len(audio)
+            num_samples = batch_count * batch_size * args.world_size
+            samples_per_epoch = dataloader.num_samples
+            percent_complete = 100.0 * batch_count / num_batches_per_epoch
+
+            # NOTE loss is coarsely sampled, just master node and per log update
+            loss_m.update(total_loss.item(), batch_size)
+            if isinstance(optimizer, dict):
+                logging.info(
+                    f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                    f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                    f"Data (t): {data_time_m.avg:.3f} "
+                    f"Batch (t): {batch_time_m.avg:.3f} "
+                    f"LR: {[o_.param_groups[0]['lr'] for o_ in optimizer.values()]}"
+                )
+                log_data = {
+                    "loss": loss_m.val,
+                    "data_time": data_time_m.val,
+                    "batch_time": batch_time_m.val,
+                    "lr": [o_.param_groups[0]["lr"] for o_ in optimizer.values()],
+                }
+            else:
+                logging.info(
+                    f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                    f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                    f"Data (t): {data_time_m.avg:.3f} "
+                    f"Batch (t): {batch_time_m.avg:.3f} "
+                    f"LR: {optimizer.param_groups[0]['lr']:5f} "
+                )
+
+                # Save train loss / etc. Using non avg meter values as loggers have their own smoothing
+                log_data = {
+                    "loss": loss_m.val,
+                    "data_time": data_time_m.val,
+                    "batch_time": batch_time_m.val,
+                    "lr": optimizer.param_groups[0]["lr"],
+                }
+            for name, val in log_data.items():
+                name = f"train{extra_suffix}/{name}"
+                if tb_writer is not None:
+                    tb_writer.add_scalar(name, val, step)
+                if args.wandb:
+                    assert wandb is not None, "Please install wandb."
+                    wandb.log({name: val, "step": step})
+
+            # resetting batch / data time meters per log window
+            batch_time_m.reset()
+            data_time_m.reset()
+    # end for
+
+
+def evaluate(model, data, epoch, args, tb_writer=None, extra_suffix=""):
+    metrics = {}
+    if not args.parallel_eval:
+        if not is_master(args):
+            return metrics
+    device = torch.device(args.device)
+    model.eval()
+
+    # CHANGE
+    # zero_shot_metrics = zero_shot_eval(model, data, epoch, args)
+    # metrics.update(zero_shot_metrics)
+    if is_master(args):
+        print("Evaluating...")
+        metric_names = args.lp_metrics.split(",")
+        eval_tool = LPMetrics(metric_names=metric_names)
+
+    autocast = torch.cuda.amp.autocast if args.precision == "amp" else suppress
+    if "val" in data and (
+        args.val_frequency
+        and ((epoch % args.val_frequency) == 0 or epoch == args.epochs)
+    ):
+        if args.parallel_eval:
+            dataloader, sampler = data["val"].dataloader, data["val"].sampler
+            if args.distributed and sampler is not None:
+                sampler.set_epoch(epoch)
+            samples_per_val = dataloader.num_samples
+        else:
+            dataloader = data["val"].dataloader
+            num_samples = 0
+            samples_per_val = dataloader.num_samples
+
+        eval_info = {"pred": [], "target": []}
+        with torch.no_grad():
+            for i, batch in enumerate(dataloader):
+                audio = batch  # contains mel_spec, wavform, and longer list
+                class_label = batch["class_label"]
+
+                # audio = audio.to(device=device, non_blocking=True)
+                class_label = class_label.to(device=device, non_blocking=True)
+
+                with autocast():
+                    pred = model(audio, device=device)
+                    if args.parallel_eval:
+                        pred, class_label = lp_gather_features(
+                            pred, class_label, args.world_size, args.horovod
+                        )
+                    eval_info["pred"].append(pred)
+                    eval_info["target"].append(class_label)
+
+                num_samples += class_label.shape[0]
+
+                if (i % 100) == 0:  # and i != 0:
+                    logging.info(
+                        f"Eval Epoch: {epoch} [{num_samples} / {samples_per_val}]"
+                    )
+
+            if is_master(args):
+                eval_info["pred"] = torch.cat(eval_info["pred"], 0).cpu()
+                eval_info["target"] = torch.cat(eval_info["target"], 0).cpu()
+                metric_dict = eval_tool.evaluate_mertics(
+                    eval_info["pred"], eval_info["target"]
+                )
+                metrics.update(metric_dict)
+                if "epoch" not in metrics.keys():
+                    metrics.update({"epoch": epoch})
+
+    if is_master(args):
+        if not metrics:
+            return metrics
+
+        logging.info(
+            f"Eval Epoch: {epoch} "
+            + "\n".join(
+                ["\t".join([f"{m}: {round(metrics[m], 4):.4f}"]) for m in metrics]
+            )
+        )
+        if args.save_logs:
+            for name, val in metrics.items():
+                if tb_writer is not None:
+                    tb_writer.add_scalar(f"val{extra_suffix}/{name}", val, epoch)
+
+            with open(os.path.join(args.checkpoint_path, "results.jsonl"), "a+") as f:
+                f.write(json.dumps(metrics))
+                f.write("\n")
+
+        if args.wandb:
+            assert wandb is not None, "Please install wandb."
+            for name, val in metrics.items():
+                wandb.log({f"val{extra_suffix}/{name}": val, "epoch": epoch})
+
+        return metrics
+    else:
+        return metrics
diff --git a/picoaudio/audioldm/clap/training/main.py b/picoaudio/audioldm/clap/training/main.py
new file mode 100644
index 0000000000000000000000000000000000000000..3b563a5d001be7adfbe779dee7ad8ac49aadc50d
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/main.py
@@ -0,0 +1,596 @@
+from inspect import getargs
+import logging
+import os
+import random
+from datetime import datetime
+import bisect
+import copy
+import numpy as np
+import torch
+import torch.backends.cudnn as cudnn
+from torch import optim
+from torch.cuda.amp import GradScaler
+import faulthandler
+import pathlib
+
+try:
+    import wandb
+except ImportError:
+    wandb = None
+
+try:
+    import torch.utils.tensorboard as tensorboard
+except ImportError:
+    tensorboard = None
+
+try:
+    import horovod.torch as hvd
+except ImportError:
+    hvd = None
+
+from open_clip import create_model_and_transforms, trace_model, create_model
+from training.data import get_data
+from training.distributed import is_master, init_distributed_device, world_info_from_env
+from training.logger import setup_logging
+from training.params import parse_args
+from training.scheduler import cosine_lr
+from training.train import train_one_epoch, evaluate
+from open_clip.utils import dataset_split, get_optimizer
+
+
+def maintain_ckpts(args, startidx, all_idx_len):
+    for i in reversed(range(startidx, all_idx_len)):
+        if os.path.exists(os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt")):
+            os.rename(
+                os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt"),
+                os.path.join(args.checkpoint_path, f"epoch_top_{i+1}.pt"),
+            )
+    if os.path.exists(
+        os.path.join(args.checkpoint_path, f"epoch_top_{all_idx_len}.pt")
+    ):
+        os.remove(os.path.join(args.checkpoint_path, f"epoch_top_{all_idx_len}.pt"))
+    return
+
+
+def update_top_k_performance(
+    new_metrics_inputs, current_top_k_ckpt_metrics, args, ckpt, bignumbetter=True
+):
+    """
+    Record the top-k performance of the current epoch.
+    current_top_k_metrics is a dictionary of the form: {1: top_1_ckpt_measure, 2: top_2_ckpt_measure, ...}
+    """
+    if isinstance(new_metrics_inputs, (list, tuple)):
+        new_metrics_inputs = np.mean(new_metrics_inputs)
+        return update_top_k_performance(
+            new_metrics_inputs,
+            current_top_k_ckpt_metrics,
+            args=args,
+            ckpt=ckpt,
+            bignumbetter=bignumbetter,
+        )
+    elif isinstance(new_metrics_inputs, dict):
+        new_metrics_inputs = np.mean(list(new_metrics_inputs.values()))
+        return update_top_k_performance(
+            new_metrics_inputs,
+            current_top_k_ckpt_metrics,
+            args=args,
+            ckpt=ckpt,
+            bignumbetter=bignumbetter,
+        )
+    elif isinstance(new_metrics_inputs, (float, int)):
+        update_flag = {k: False for k in current_top_k_ckpt_metrics.keys()}
+        sorted_keys = sorted(current_top_k_ckpt_metrics.keys())
+        sorted_values = sorted(
+            current_top_k_ckpt_metrics.values(), reverse=bignumbetter
+        )
+        sorted_values_ = copy.deepcopy(sorted_values)
+        sorted_values.append(new_metrics_inputs)
+        sorted_values = sorted(sorted_values, reverse=bignumbetter)
+        sorted_values = sorted_values[:-1]
+
+        if sorted_values == sorted_values_:
+            return current_top_k_ckpt_metrics, new_metrics_inputs
+        else:
+            for i in range(len(sorted_keys)):
+                if current_top_k_ckpt_metrics[sorted_keys[i]] != sorted_values[i]:
+                    current_top_k_ckpt_metrics[sorted_keys[i]] = sorted_values[i]
+                    update_flag[sorted_keys[i]] = True
+            for i in range(len(update_flag)):
+                if update_flag[i]:
+                    maintain_ckpts(args, i, len(sorted_keys))
+                    torch.save(
+                        ckpt,
+                        os.path.join(args.checkpoint_path, f"epoch_top_{i}.pt"),
+                    )
+                    break
+            return current_top_k_ckpt_metrics, new_metrics_inputs
+
+
+# def updateifNone(a, b):
+#     a = b if None else a
+#     return a
+
+
+def is_pretrained_params(n):
+    return (
+        n.startswith("transformer")
+        or n in ["positional_embedding", "text_projection"]
+        or n.startswith("token_embedding")
+        or n.startswith("ln_final")
+        or n.startswith("logit_scale_t")
+    )
+
+
+def random_seed(seed=42, rank=0):
+    torch.manual_seed(seed + rank)
+    np.random.seed(seed + rank)
+    random.seed(seed + rank)
+
+
+def main():
+    args = parse_args()
+    # sanitize model name for filesystem / uri use, easier if we don't use / in name as a rule?
+    args.amodel = args.amodel.replace("/", "-")
+    # download sizes.json file
+
+    # (yusong): the below two lines are for debug
+    # print("setting up faulthandler")
+    # faulthandler.register(10)
+
+    random.seed(args.seed)
+    torch.manual_seed(args.seed)
+    torch.cuda.manual_seed(args.seed)
+    torch.cuda.manual_seed_all(args.seed)
+    np.random.seed(args.seed)
+    if args.tmodel == "bert" or args.tmodel == "roberta" or args.tmodel == "bart":
+        assert (
+            args.pretrained == "" or args.pretrained is None
+        ), "bert/roberta/bart text encoder does not support pretrained models."
+
+    # get the name of the experiments
+    if args.name is None:
+        args.name = "-".join(
+            [
+                datetime.now().strftime("%Y_%m_%d-%H_%M_%S"),
+                f"model_{args.amodel}",
+                f"lr_{args.lr}",
+                f"b_{args.batch_size}",
+                f"j_{args.workers}",
+                f"p_{args.precision}",
+            ]
+        )
+
+    # discover initial world args early so we can log properly
+    args.distributed = False
+    args.local_rank, args.rank, args.world_size = world_info_from_env()
+
+    if args.remotedata and is_master(args):
+        for dataset_name in args.datasetnames:
+            for split in dataset_split[dataset_name]:
+                if not os.path.exists(f"./json_files/{dataset_name}/{split}"):
+                    os.makedirs(f"./json_files/{dataset_name}/{split}")
+                os.system(
+                    f"aws s3 cp s3://s-laion-audio/webdataset_tar/{dataset_name}/{split}/sizes.json ./json_files/{dataset_name}/{split}/sizes.json"
+                )
+
+    args.log_path = None
+    if is_master(args, local=args.log_local):
+        log_base_path = os.path.join(args.logs, args.name)
+        os.makedirs(log_base_path, exist_ok=True)
+        log_filename = f"out-{args.rank}" if args.log_local else "out.log"
+        args.log_path = os.path.join(log_base_path, log_filename)
+        if os.path.exists(args.log_path):
+            print(
+                "Error. Experiment already exists. Use --name {} to specify a new experiment."
+            )
+            return -1
+
+    # Set logger
+    args.log_level = logging.DEBUG if args.debug else logging.INFO
+    setup_logging(args.log_path, args.log_level)
+
+    # fully initialize distributed device environment
+    device = init_distributed_device(args)
+
+    args.wandb = "wandb" in args.report_to or "all" in args.report_to
+    args.tensorboard = "tensorboard" in args.report_to or "all" in args.report_to
+    if is_master(args):
+        args.tensorboard_path = (
+            os.path.join(args.logs, args.name, "tensorboard")
+            if args.tensorboard
+            else ""
+        )
+        args.checkpoint_path = os.path.join(args.logs, args.name, "checkpoints")
+        for dirname in [args.tensorboard_path, args.checkpoint_path]:
+            if dirname:
+                os.makedirs(dirname, exist_ok=True)
+    else:
+        args.tensorboard_path = ""
+        args.checkpoint_path = ""
+
+    if args.copy_codebase:
+        copy_codebase(args)
+
+    assert args.precision in ["amp", "fp16", "fp32"]
+    if args.precision == "fp16":
+        logging.warning(
+            "It is recommended to use AMP mixed-precision instead of FP16. "
+            "FP16 support needs further verification and tuning, especially for train."
+        )
+
+    if args.horovod:
+        logging.info(
+            f"Running in horovod mode with multiple processes / nodes. Device: {args.device}."
+            f"Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}."
+        )
+    elif args.distributed:
+        logging.info(
+            f"Running in distributed mode with multiple processes. Device: {args.device}."
+            f"Process (global: {args.rank}, local {args.local_rank}), total {args.world_size}."
+        )
+    else:
+        logging.info(f"Running with a single process. Device {args.device}.")
+
+    logging.info(f"openai cache dir: {os.path.expanduser(args.openai_model_cache_dir)}")
+
+    model, model_cfg = create_model(
+        args.amodel,
+        args.tmodel,
+        args.pretrained,
+        precision=args.precision,
+        device=device,
+        jit=args.torchscript,
+        force_quick_gelu=args.force_quick_gelu,
+        openai_model_cache_dir=os.path.expanduser(args.openai_model_cache_dir),
+        skip_params=True,
+        pretrained_audio=args.pretrained_audio,
+        pretrained_text=args.pretrained_text,
+        enable_fusion=args.enable_fusion,
+        fusion_type=args.fusion_type,
+    )
+
+    if args.horovod:
+        with torch.no_grad():
+            for param in model.parameters():
+                param.set_(param.contiguous())
+
+    if args.trace:
+        model = trace_model(model, batch_size=args.batch_size, device=device)
+
+    if is_master(args):
+        logging.info("Model:")
+        logging.info(f"{str(model)}")
+        logging.info("Params:")
+        params_file = os.path.join(args.logs, args.name, "params.txt")
+        with open(params_file, "w") as f:
+            for name in sorted(vars(args)):
+                val = getattr(args, name)
+                logging.info(f"  {name}: {val}")
+                f.write(f"{name}: {val}\n")
+
+    if args.distributed and not args.horovod:
+        if args.use_bn_sync:
+            model = torch.nn.SyncBatchNorm.convert_sync_batchnorm(model)
+        ddp_args = {}
+        if args.ddp_static_graph:
+            # this doesn't exist in older PyTorch, arg only added if enabled
+            ddp_args["static_graph"] = True
+        model = torch.nn.parallel.DistributedDataParallel(
+            model, device_ids=[device], find_unused_parameters=True, **ddp_args
+        )
+
+    data = get_data(args, model_cfg)
+    assert len(data), "At least one train or eval dataset must be specified."
+    if args.trace:
+        assert "train" not in data, "Cannot train with traced model"
+
+    exclude = (
+        lambda n, p: p.ndim < 2
+        or "bn" in n
+        or "ln" in n
+        or "bias" in n
+        or "logit_scale" in n
+    )
+    include = lambda n, p: not exclude(n, p)
+
+    named_parameters = list(model.named_parameters())
+
+    # freeze text encoder
+    text_freeze_parameters = [p for n, p in named_parameters if "text_branch" in n]
+
+    if args.freeze_text:
+        print("Freeze Text!!!!")
+        for k in text_freeze_parameters:
+            k.requires_grad = False
+
+    gain_or_bias_params = [
+        p for n, p in named_parameters if exclude(n, p) and p.requires_grad
+    ]
+    rest_params = [p for n, p in named_parameters if include(n, p) and p.requires_grad]
+
+    # set wd-related params to 0 if use adam optimizer
+    if args.optimizer == "adam":
+        args.wd = 0
+        args.wd_pretrained = 0
+        args.wd_new = 0
+
+    if args.train_data is None:
+        optimizer = None
+        scheduler = None
+    else:
+        total_steps = data["train"].dataloader.num_batches * args.epochs
+
+        if args.split_opt:
+            for x in ["lr", "beta1", "beta2", "eps", "wd"]:
+                for y in ["_new", "_pretrained"]:
+                    if getattr(args, x + y) is None:
+                        setattr(args, x + y, getattr(args, x))
+
+            gain_or_bias_pretrained_params = [
+                p
+                for n, p in named_parameters
+                if (exclude(n, p) and p.requires_grad) and is_pretrained_params(n)
+            ]
+            rest_pretrained_params = [
+                p
+                for n, p in named_parameters
+                if (include(n, p) and p.requires_grad) and is_pretrained_params(n)
+            ]
+            gain_or_bias_new_params = [
+                p
+                for n, p in named_parameters
+                if (exclude(n, p) and p.requires_grad) and (not is_pretrained_params(n))
+            ]
+            rest_new_params = [
+                p
+                for n, p in named_parameters
+                if (include(n, p) and p.requires_grad) and (not is_pretrained_params(n))
+            ]
+            pretrained_params_optimizer = get_optimizer(
+                [
+                    {"params": gain_or_bias_pretrained_params, "weight_decay": 0.0},
+                    {
+                        "params": rest_pretrained_params,
+                        "weight_decay": args.wd_pretrained,
+                    },
+                ],
+                lr=args.lr_pretrained,
+                betas=(args.beta1_pretrained, args.beta2_pretrained),
+                eps=args.eps_pretrained,
+                momentum=args.momentum_pretrained,
+                optimizer_name=args.optimizer,
+            )
+            pretrained_params_scheduler = cosine_lr(
+                pretrained_params_optimizer,
+                args.lr_pretrained,
+                args.warmup,
+                total_steps,
+            )
+            new_params_optimizer = get_optimizer(
+                [
+                    {"params": gain_or_bias_new_params, "weight_decay": 0.0},
+                    {"params": rest_new_params, "weight_decay": args.wd_new},
+                ],
+                lr=args.lr_new,
+                betas=(args.beta1_new, args.beta2_new),
+                eps=args.eps_new,
+                momentum=args.momentum_new,
+                optimizer_name=args.optimizer,
+            )
+
+            new_params_scheduler = cosine_lr(
+                new_params_optimizer, args.lr_new, args.warmup, total_steps
+            )
+
+            optimizer = {
+                "pretrained": pretrained_params_optimizer,
+                "new": new_params_optimizer,
+            }
+            scheduler = {
+                "pretrained": pretrained_params_scheduler,
+                "new": new_params_scheduler,
+            }
+
+            if args.horovod:
+                pretrained_params_optimizer = hvd.DistributedOptimizer(
+                    pretrained_params_optimizer,
+                    named_parameters=model.named_parameters(),
+                )
+                new_params_optimizer = hvd.DistributedOptimizer(
+                    new_params_optimizer, named_parameters=model.named_parameters()
+                )
+                hvd.broadcast_parameters(model.state_dict(), root_rank=0)
+                hvd.broadcast_optimizer_state(pretrained_params_optimizer, root_rank=0)
+                hvd.broadcast_optimizer_state(new_params_optimizer, root_rank=0)
+        else:
+            optimizer = get_optimizer(
+                [
+                    {"params": gain_or_bias_params, "weight_decay": 0.0},
+                    {"params": rest_params, "weight_decay": args.wd},
+                ],
+                lr=args.lr,
+                betas=(args.beta1, args.beta2),
+                eps=args.eps,
+                momentum=args.momentum,
+                optimizer_name=args.optimizer,
+            )
+
+            scheduler = cosine_lr(optimizer, args.lr, args.warmup, total_steps)
+
+            if args.horovod:
+                optimizer = hvd.DistributedOptimizer(
+                    optimizer, named_parameters=model.named_parameters()
+                )
+                hvd.broadcast_parameters(model.state_dict(), root_rank=0)
+                hvd.broadcast_optimizer_state(optimizer, root_rank=0)
+
+    scaler = GradScaler() if args.precision == "amp" else None
+
+    # optionally resume from a checkpoint
+    start_epoch = 0
+    if args.resume is not None:
+        if os.path.isfile(args.resume):
+            checkpoint = torch.load(args.resume, map_location=device)
+            if "epoch" in checkpoint:
+                # resuming a train checkpoint w/ epoch and optimizer state
+                start_epoch = checkpoint["epoch"]
+                sd = checkpoint["state_dict"]
+                if not args.distributed and next(iter(sd.items()))[0].startswith(
+                    "module"
+                ):
+                    sd = {k[len("module.") :]: v for k, v in sd.items()}
+                model.load_state_dict(sd)
+                if args.split_opt:
+                    if optimizer is not None:
+                        for k, o_ in optimizer.items():
+                            o_.load_state_dict(checkpoint[k + "_" + "optimizer"])
+                if optimizer is not None:
+                    optimizer.load_state_dict(checkpoint["optimizer"])
+                if scaler is not None and "scaler" in checkpoint:
+                    scaler.load_state_dict(checkpoint["scaler"])
+                logging.info(
+                    f"=> resuming checkpoint '{args.resume}' (epoch {start_epoch})"
+                )
+            else:
+                # loading a bare (model only) checkpoint for fine-tune or evaluation
+                model.load_state_dict(checkpoint)
+                logging.info(
+                    f"=> loaded checkpoint '{args.resume}' (epoch {start_epoch})"
+                )
+            if args.freeze_text:
+                print("Freeze Text!!!!")
+                for k in text_freeze_parameters:
+                    k.requires_grad = False
+        else:
+            logging.info("=> no checkpoint found at '{}'".format(args.resume))
+
+    cudnn.benchmark = True
+    cudnn.deterministic = False
+
+    # determine if this worker should save logs and checkpoints. only do so if it is rank == 0
+    args.save_logs = args.logs and args.logs.lower() != "none" and is_master(args)
+    writer = None
+    if args.save_logs and args.tensorboard:
+        assert tensorboard is not None, "Please install tensorboard."
+        writer = tensorboard.SummaryWriter(args.tensorboard_path)
+
+    if args.wandb and is_master(args):
+        assert wandb is not None, "Please install wandb."
+        logging.debug("Starting wandb.")
+        args.train_sz = data["train"].dataloader.num_samples
+        if args.val_data is not None:
+            args.val_sz = data["val"].dataloader.num_samples
+        # you will have to configure this for your project!
+        wandb.init(
+            project="clap",
+            notes=args.wandb_notes,
+            name=args.wandb_notes,
+            tags=[],
+            config=vars(args),
+        )
+        if args.debug:
+            wandb.watch(model, log="all")
+        wandb.save(params_file)
+        logging.debug("Finished loading wandb.")
+
+    if "train" not in data:
+        evaluate(model, data, start_epoch, args, writer)
+        return
+    elif start_epoch == 0 and "val" in data and not args.no_eval:
+        evaluate(model, data, 0, args, writer)
+        #  print(f'rank {args.rank}, Start First Evaluation')#  (yusong): for debug
+    if args.save_top_performance:
+        current_top_k_ckpt_metrics = {
+            i: 0 for i in range(args.save_top_performance)
+        }  # initialize the top-k metric for ckpts to 0
+
+    #  print(f'rank {args.rank}, Start Training') #  (yusong): for debug
+    for epoch in range(start_epoch, args.epochs):
+        # freeze the text param after (include) args.freeze_text_after, this is -1 by default
+        if epoch == args.freeze_text_after:
+            print("Text pretrained parameters are freezed since this epoch.")
+            for k in text_freeze_parameters:
+                k.requires_grad = False
+        if is_master(args):
+            logging.info(f"Start epoch {epoch}")
+
+        train_one_epoch(model, data, epoch, optimizer, scaler, scheduler, args, writer)
+        completed_epoch = epoch + 1
+
+        if (
+            any(v in data for v in ("val", "imagenet-val", "imagenet-v2"))
+            and not args.no_eval
+        ):
+            metrics = evaluate(model, data, completed_epoch, args, writer)
+            if args.save_top_performance:
+                top_k_dataset = args.top_k_checkpoint_select_dataset
+                top_k_metric = args.top_k_checkpoint_select_metric
+                filtered_metrics = [
+                    v
+                    for k, v in metrics.items()
+                    if top_k_metric in k and top_k_dataset in k
+                ]  # check all R@10 metrics (all dataset) and use it to update the ckpt
+        # Saving checkpoints.
+        if args.save_logs:
+            if args.split_opt:
+                opt_dict = {
+                    k + "_" + "optimizer": v.state_dict() for k, v in optimizer.items()
+                }
+            else:
+                opt_dict = {"optimizer": optimizer.state_dict()}
+            checkpoint_dict = {
+                "epoch": completed_epoch,
+                "name": args.name,
+                "state_dict": model.state_dict(),
+            }
+            checkpoint_dict.update(opt_dict)
+            if scaler is not None:
+                checkpoint_dict["scaler"] = scaler.state_dict()
+
+            if completed_epoch == args.epochs or (
+                args.save_frequency > 0 and (completed_epoch % args.save_frequency) == 0
+            ):
+                torch.save(
+                    checkpoint_dict,
+                    os.path.join(args.checkpoint_path, f"epoch_{completed_epoch}.pt"),
+                )
+            if args.save_most_recent:
+                torch.save(
+                    checkpoint_dict,
+                    os.path.join(args.checkpoint_path, f"epoch_latest.pt"),
+                )
+            if args.save_top_performance and not args.no_eval:
+                update_top_k_performance(
+                    filtered_metrics,
+                    current_top_k_ckpt_metrics,
+                    args,
+                    checkpoint_dict,
+                    bignumbetter=True,
+                )
+
+    if args.wandb and is_master(args):
+        wandb.finish()
+
+
+def copy_codebase(args):
+    from shutil import copytree, ignore_patterns
+
+    new_code_path = os.path.join(args.logs, args.name, "code")
+    if os.path.exists(new_code_path):
+        print(
+            f"Error. Experiment already exists at {new_code_path}. Use --name to specify a new experiment."
+        )
+        return -1
+    print(f"Copying codebase to {new_code_path}")
+    current_code_path = os.path.realpath(__file__)
+    for _ in range(3):
+        current_code_path = os.path.dirname(current_code_path)
+    copytree(
+        current_code_path, new_code_path, ignore=ignore_patterns("log", "logs", "wandb")
+    )
+    print("Done copying code.")
+    return 1
+
+
+if __name__ == "__main__":
+    main()
diff --git a/picoaudio/audioldm/clap/training/params.py b/picoaudio/audioldm/clap/training/params.py
new file mode 100644
index 0000000000000000000000000000000000000000..b1933e3a78ff583733846ea285d56eb0a0b892a5
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/params.py
@@ -0,0 +1,569 @@
+import argparse
+import os
+
+CACHE_DIR = os.getenv(
+    "AUDIOLDM_CACHE_DIR",
+    "~/.cache")
+
+
+
+def get_default_params(model_name):
+    # Params from paper (https://arxiv.org/pdf/2103.00020.pdf)
+    model_name = model_name.lower()
+    if "vit" in model_name:
+        return {"lr": 5.0e-4, "beta1": 0.9, "beta2": 0.98, "eps": 1.0e-6}
+    else:
+        return {"lr": 5.0e-4, "beta1": 0.9, "beta2": 0.999, "eps": 1.0e-8}
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument(
+        "--train-data",
+        type=str,
+        default=None,
+        help="Path to h5 filewith training data",
+    )
+    parser.add_argument(
+        "--val-data",
+        type=str,
+        default=None,
+        help="Path to h5 file with validation data",
+    )
+    parser.add_argument(
+        "--freeze-text",
+        default=False,
+        action="store_true",
+        help="if you need to freeze the text encoder, make this True",
+    )
+    parser.add_argument(
+        "--freeze-text-after",
+        type=int,
+        default=-1,
+        help="if you need to freeze the text encoder after (include) epoch x, set this param to x. Set -1 to disable it",
+    )
+    parser.add_argument(
+        "--train-ipc",
+        type=str,
+        default=None,
+        help="Path to npy file of the number of instance per class in training data",
+    )
+    parser.add_argument(
+        "--val-ipc",
+        type=str,
+        default=None,
+        help="Path to npy file of the number of instance per class in validation data",
+    )
+    parser.add_argument(
+        "--train-num-samples",
+        type=int,
+        default=None,
+        help="Number of samples in dataset. Required for webdataset if not available in info file.",
+    )
+    parser.add_argument(
+        "--val-num-samples",
+        type=int,
+        default=None,
+        help="Number of samples in dataset. Useful for webdataset if not available in info file.",
+    )
+    parser.add_argument(
+        "--dataset-type",
+        choices=["webdataset", "csv", "auto", "toy"],
+        default="auto",
+        help="Which type of dataset to process.",
+    )
+    parser.add_argument(
+        "--csv-separator",
+        type=str,
+        default="\t",
+        help="For csv-like datasets, which separator to use.",
+    )
+    parser.add_argument(
+        "--csv-img-key",
+        type=str,
+        default="filepath",
+        help="For csv-like datasets, the name of the key for the image paths.",
+    )
+    parser.add_argument(
+        "--csv-caption-key",
+        type=str,
+        default="title",
+        help="For csv-like datasets, the name of the key for the captions.",
+    )
+    parser.add_argument(
+        "--imagenet-val",
+        type=str,
+        default=None,
+        help="Path to imagenet val set for conducting zero shot evaluation.",
+    )
+    parser.add_argument(
+        "--imagenet-v2",
+        type=str,
+        default=None,
+        help="Path to imagenet v2 for conducting zero shot evaluation.",
+    )
+    parser.add_argument(
+        "--datasetnames",
+        nargs="+",
+        default=None,
+        help="If loading webdataset, spedify the dataset names to load. Can be some of these: Clotho, audioset, audiocaps, BBCSoundEffects",
+    )
+    parser.add_argument(
+        "--full-train-dataset",
+        nargs="+",
+        default=None,
+        help="Which dataset will be trained with all the subsets. (train+test)",
+    )
+    parser.add_argument(
+        "--exclude-eval-dataset",
+        nargs="+",
+        default=None,
+        help="Which dataset will be excluded with evaluation",
+    )
+    parser.add_argument(
+        "--datasetinfos",
+        nargs="+",
+        default=None,
+        help="If loading webdataset, spedify the dataset types to load. Can be some of these: train, test, valid, unbalanced_train, balanced_train, eval",
+    )
+    parser.add_argument(
+        "--dataset-proportion",
+        type=float,
+        default=1.0,
+        help="How much proportion of dataset we want to train.",
+    )
+    parser.add_argument(
+        "--remotedata",
+        default=False,
+        action="store_true",
+        help="if the dataset is remote, set this flag",
+    )
+    parser.add_argument(
+        "--class-label-path",
+        type=str,
+        default=None,
+        help="The path of the class label pickle or csv.",
+    )
+    parser.add_argument(
+        "--datasetpath",
+        type=str,
+        default="/mnt/audio_clip/webdataset_tar",
+        help="The path to the dataset",
+    )
+    parser.add_argument(
+        "--logs",
+        type=str,
+        default="./logs/",
+        help="Where to store tensorboard logs. Use None to avoid storing logs.",
+    )
+    parser.add_argument(
+        "--log-local",
+        action="store_true",
+        default=False,
+        help="log files on local master, otherwise global master only.",
+    )
+    parser.add_argument(
+        "--name",
+        type=str,
+        default=None,
+        help="Optional identifier for the experiment when storing logs. Otherwise use current time.",
+    )
+    parser.add_argument(
+        "--workers", type=int, default=1, help="Number of workers per GPU."
+    )
+    parser.add_argument(
+        "--batch-size", type=int, default=64, help="Batch size per GPU."
+    )
+    parser.add_argument(
+        "--epochs", type=int, default=32, help="Number of epochs to train for."
+    )
+    parser.add_argument("--lr", type=float, default=None, help="Learning rate.")
+    parser.add_argument("--beta1", type=float, default=None, help="Adam beta 1.")
+    parser.add_argument("--beta2", type=float, default=None, help="Adam beta 2.")
+    parser.add_argument("--eps", type=float, default=None, help="Adam epsilon.")
+    parser.add_argument("--momentum", type=float, default=None, help="SGD epsilon.")
+    parser.add_argument("--wd", type=float, default=0.2, help="Weight decay.")
+
+    parser.add_argument(
+        "--split-opt",
+        action="store_true",
+        default=False,
+        help="Use this flag to skip the learning rate decay.",
+    )
+    parser.add_argument(
+        "--lr-pretrained", type=float, default=None, help="Learning rate for text."
+    )
+    parser.add_argument(
+        "--beta1-pretrained", type=float, default=None, help="Adam beta 1 for text."
+    )
+    parser.add_argument(
+        "--beta2-pretrained", type=float, default=None, help="Adam beta 2 for text."
+    )
+    parser.add_argument(
+        "--eps-pretrained", type=float, default=None, help="Adam epsilon for text."
+    )
+    parser.add_argument(
+        "--wd-pretrained", type=float, default=0.2, help="Weight decay for text."
+    )
+    parser.add_argument(
+        "--momentum-pretrained", type=float, default=0.9, help="Momentum for text."
+    )
+    parser.add_argument(
+        "--lr-new", type=float, default=None, help="Learning rate for audio."
+    )
+    parser.add_argument(
+        "--beta1-new", type=float, default=None, help="Adam beta 1 for audio."
+    )
+    parser.add_argument(
+        "--beta2-new", type=float, default=None, help="Adam beta 2 for audio."
+    )
+    parser.add_argument(
+        "--eps-new", type=float, default=None, help="Adam epsilon for audio."
+    )
+    parser.add_argument(
+        "--wd-new", type=float, default=0.2, help="Weight decay for audio."
+    )
+    parser.add_argument(
+        "--momentum-new", type=float, default=0.9, help="Momentum for audio."
+    )
+    parser.add_argument(
+        "--warmup", type=int, default=10000, help="Number of steps to warmup for."
+    )
+    parser.add_argument(
+        "--use-bn-sync",
+        default=False,
+        action="store_true",
+        help="Whether to use batch norm sync.",
+    )
+    parser.add_argument(
+        "--skip-scheduler",
+        action="store_true",
+        default=False,
+        help="Use this flag to skip the learning rate decay.",
+    )
+    parser.add_argument(
+        "--save-frequency", type=int, default=1, help="How often to save checkpoints."
+    )
+    parser.add_argument(
+        "--save-top-performance",
+        type=int,
+        default=0,
+        help="Save the top x performance weights if the value >0",
+    )
+    parser.add_argument(
+        "--save-most-recent",
+        action="store_true",
+        default=False,
+        help="Always save the most recent model trained to epoch_latest.pt.",
+    )
+    parser.add_argument(
+        "--zeroshot-frequency", type=int, default=2, help="How often to run zero shot."
+    )
+    parser.add_argument(
+        "--val-frequency",
+        type=int,
+        default=1,
+        help="How often to run evaluation with val data.",
+    )
+    parser.add_argument(
+        "--resume",
+        default=None,
+        type=str,
+        help="path to latest checkpoint (default: none)",
+    )
+    parser.add_argument(
+        "--precision",
+        choices=["amp", "fp16", "fp32"],
+        default="amp",
+        help="Floating point precision.",
+    )
+    parser.add_argument(
+        "--amodel",
+        type=str,
+        default="RN50",
+        help="Name of the audio backbone to use.",
+    )
+    parser.add_argument(
+        "--tmodel",
+        type=str,
+        default="transformer",
+        help="Name of the text backbone to use. Can be [transformer, bert, roberta, bart]",
+    )
+    parser.add_argument(
+        "--pretrained-audio",
+        default="",
+        type=str,
+        help="Use a pretrained audio model weights for the audio encoder of CLAP",
+    )
+    parser.add_argument(
+        "--pretrained-text",
+        default="",
+        type=str,
+        help="Use a pretrained text model weights for the text encoder of CLAP",
+    )
+    parser.add_argument(
+        "--pretrained",
+        default="",
+        type=str,
+        help="Use a pretrained CLIP model weights with the specified tag or file path.",
+    )
+    parser.add_argument(
+        "--pretrained-image",
+        default=False,
+        action="store_true",
+        help="Load imagenet pretrained weights for image tower backbone if available.",
+    )
+    parser.add_argument(
+        "--lock-image",
+        default=False,
+        action="store_true",
+        help="Lock full image tower by disabling gradients.",
+    )
+    parser.add_argument(
+        "--lock-image-unlocked-groups",
+        type=int,
+        default=0,
+        help="Leave last n image tower layer groups unlocked.",
+    )
+    parser.add_argument(
+        "--lock-image-freeze-bn-stats",
+        default=False,
+        action="store_true",
+        help="Freeze BatchNorm running stats in image tower for any locked layers.",
+    )
+    parser.add_argument(
+        "--local-loss",
+        default=False,
+        action="store_true",
+        help="calculate loss w/ local features @ global (instead of realizing full global @ global matrix)",
+    )
+    parser.add_argument(
+        "--gather-with-grad",
+        default=False,
+        action="store_true",
+        help="enable full distributed gradient for feature gather",
+    )
+    parser.add_argument(
+        "--force-quick-gelu",
+        default=False,
+        action="store_true",
+        help="Force use of QuickGELU activation for non-OpenAI transformer models.",
+    )
+    parser.add_argument(
+        "--torchscript",
+        default=False,
+        action="store_true",
+        help="torch.jit.script the model, also uses jit version of OpenAI models if pretrained=='openai'",
+    )
+    parser.add_argument(
+        "--trace",
+        default=False,
+        action="store_true",
+        help="torch.jit.trace the model for inference / eval only",
+    )
+    # arguments for distributed training
+    parser.add_argument(
+        "--dist-url",
+        default="env://",
+        type=str,
+        help="url used to set up distributed training",
+    )
+    parser.add_argument(
+        "--dist-backend", default="nccl", type=str, help="distributed backend"
+    )
+    parser.add_argument(
+        "--report-to",
+        default="",
+        type=str,
+        help="Options are ['wandb', 'tensorboard', 'wandb,tensorboard']",
+    )
+    parser.add_argument(
+        "--wandb-notes", default="", type=str, help="Notes if logging with wandb"
+    )
+    parser.add_argument(
+        "--C", type=float, default=3.16, help="inverse regularizer for logistic reg."
+    )
+    parser.add_argument(
+        "--debug",
+        default=False,
+        action="store_true",
+        help="If true, more information is logged.",
+    )
+    parser.add_argument(
+        "--copy-codebase",
+        default=False,
+        action="store_true",
+        help="If true, we copy the entire base on the log diretory, and execute from there.",
+    )
+    parser.add_argument(
+        "--horovod",
+        default=False,
+        action="store_true",
+        help="Use horovod for distributed training.",
+    )
+    parser.add_argument(
+        "--ddp-static-graph",
+        default=False,
+        action="store_true",
+        help="Enable static graph optimization for DDP in PyTorch >= 1.11.",
+    )
+    parser.add_argument(
+        "--no-set-device-rank",
+        default=False,
+        action="store_true",
+        help="Don't set device index from local rank (when CUDA_VISIBLE_DEVICES restricted to one per proc).",
+    )
+    parser.add_argument("--seed", type=int, default=4242, help="Default random seed.")
+
+    parser.add_argument(
+        "--top-k-checkpoint-select-dataset",
+        type=str,
+        default="all",
+        help="The dataset of selecting top-k checkpoint.",
+    )
+
+    # @R10, @R@5, @R1, mAP@10
+    parser.add_argument(
+        "--top-k-checkpoint-select-metric",
+        type=str,
+        default="_R@10",
+        help="The metric for selecting top-k checkpoint.",
+    )
+    parser.add_argument(
+        "--openai-model-cache-dir",
+        type=str,
+        default=f"{CACHE_DIR}/clip",
+        help="Directory to download OpenAI models.",
+    )
+    parser.add_argument(
+        "--optimizer",
+        type=str,
+        default="adamw",
+        help="can be AdamW or SGD",
+    )
+    parser.add_argument(
+        "--parallel-eval",
+        default=False,
+        action="store_true",
+        help="Eval in parallel (multi-GPU, multi-node).",
+    )
+
+    parser.add_argument(
+        "--no-eval",
+        default=False,
+        action="store_true",
+        help="Training without evaluation.",
+    )
+
+    parser.add_argument(
+        "--lp-mlp",
+        default=False,
+        action="store_true",
+        help="Linear Probe using MLP layer or not.",
+    )
+
+    parser.add_argument(
+        "--lp-freeze",
+        default=False,
+        action="store_true",
+        help="Linear Probe using Freeze CLAP or not",
+    )
+
+    parser.add_argument(
+        "--lp-act",
+        default="None",
+        type=str,
+        help="Options are ['relu','elu','prelu','softmax','sigmoid']",
+    )
+
+    parser.add_argument(
+        "--lp-loss", type=str, default="bce", help="Loss func of Linear Probe."
+    )
+
+    parser.add_argument(
+        "--lp-metrics",
+        type=str,
+        default="map,mauc,acc",
+        help="Metrics of Linear Probe.",
+    )
+
+    parser.add_argument(
+        "--lp-lr", type=float, default=1e-4, help="learning rate of linear probe"
+    )
+    parser.add_argument(
+        "--kappa",
+        type=float,
+        default=0,
+        help="the kappa in the weighted contrastive loss, default is to turn off the weighted contrastive loss",
+    )
+
+    parser.add_argument(
+        "--data-filling",
+        type=str,
+        default="pad",
+        help="type of data filling when the audio length is shorter than the max length."
+        "Can be one of the following: repeat, repeatpad, pad",
+    )
+    parser.add_argument(
+        "--data-truncating",
+        type=str,
+        default="rand_trunc",
+        help="type of data truncation when the audio length is longer than the max length."
+        "Can be one of the following: rand_trunc, fusion",
+    )
+
+    parser.add_argument(
+        "--clap-mlploss",
+        default=False,
+        action="store_true",
+        help="Using MLP loss for CLAP model or not",
+    )
+
+    parser.add_argument(
+        "--wandb-id",
+        type=str,
+        default=None,
+        help="the id of wandb experiment to restore.",
+    )
+
+    parser.add_argument(
+        "--sleep", type=float, default=0, help="sleep n seconds before start training"
+    )
+
+    # variable length processing
+    parser.add_argument(
+        "--enable-fusion",
+        default=False,
+        action="store_true",
+        help="Enable feature funsion for variable-length data",
+    )
+
+    parser.add_argument(
+        "--fusion-type",
+        type=str,
+        default="None",
+        help="Type is among ['channel_map', 'daf_1d','aff_1d','iaff_1d','daf_2d','aff_2d','iaff_2d']",
+    )
+
+    parser.add_argument(
+        "--mixup",
+        default=False,
+        action="store_true",
+        help="Enable mixup in finetuning training.",
+    )
+    parser.add_argument(
+        "--text-augment-selection",
+        type=str,
+        default=None,
+        help="For selecting levels of augmented text. Type is among ['all', 'augment_only', 'none']",
+    )
+
+    args = parser.parse_args()
+
+    # If some params are not passed, we use the default values based on model name.
+    default_params = get_default_params(args.amodel)
+    for name, val in default_params.items():
+        if getattr(args, name) is None:
+            setattr(args, name, val)
+
+    return args
diff --git a/picoaudio/audioldm/clap/training/scheduler.py b/picoaudio/audioldm/clap/training/scheduler.py
new file mode 100644
index 0000000000000000000000000000000000000000..7151ffbab25a113673b7627027b443b27f22cb0f
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/scheduler.py
@@ -0,0 +1,24 @@
+import numpy as np
+
+
+def assign_learning_rate(optimizer, new_lr):
+    for param_group in optimizer.param_groups:
+        param_group["lr"] = new_lr
+
+
+def _warmup_lr(base_lr, warmup_length, step):
+    return base_lr * (step + 1) / warmup_length
+
+
+def cosine_lr(optimizer, base_lr, warmup_length, steps):
+    def _lr_adjuster(step):
+        if step < warmup_length:
+            lr = _warmup_lr(base_lr, warmup_length, step)
+        else:
+            e = step - warmup_length
+            es = steps - warmup_length
+            lr = 0.5 * (1 + np.cos(np.pi * e / es)) * base_lr
+        assign_learning_rate(optimizer, lr)
+        return lr
+
+    return _lr_adjuster
diff --git a/picoaudio/audioldm/clap/training/train.py b/picoaudio/audioldm/clap/training/train.py
new file mode 100644
index 0000000000000000000000000000000000000000..f5759c4679d2ee9c0748444adf66b8453cf09728
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/train.py
@@ -0,0 +1,838 @@
+import json
+import logging
+import math
+import os
+import time
+from contextlib import suppress
+
+import numpy as np
+import torch
+import torch.nn.functional as F
+
+try:
+    import wandb
+except ImportError:
+    wandb = None
+
+from open_clip import ClipLoss, gather_features
+from .distributed import is_master
+from .zero_shot import zero_shot_eval
+
+
+class AverageMeter(object):
+    """Computes and stores the average and current value"""
+
+    def __init__(self):
+        self.reset()
+
+    def reset(self):
+        self.val = 0
+        self.avg = 0
+        self.sum = 0
+        self.count = 0
+
+    def update(self, val, n=1):
+        self.val = val
+        self.sum += val * n
+        self.count += n
+        self.avg = self.sum / self.count
+
+
+def unwrap_model(model):
+    if hasattr(model, "module"):
+        return model.module
+    else:
+        return model
+
+
+def train_one_epoch(
+    model, data, epoch, optimizer, scaler, scheduler, args, tb_writer=None
+):
+    device = torch.device(args.device)
+    autocast = torch.cuda.amp.autocast if args.precision == "amp" else suppress
+    model.train()
+    loss = ClipLoss(
+        local_loss=args.local_loss,
+        gather_with_grad=args.gather_with_grad,
+        cache_labels=True,
+        rank=args.rank,
+        world_size=args.world_size,
+        use_horovod=args.horovod,
+        mlp_loss=args.clap_mlploss,
+        weight_loss_kappa=args.kappa,
+    )
+
+    dataloader, sampler = data["train"].dataloader, data["train"].sampler
+    if args.distributed and sampler is not None:
+        sampler.set_epoch(epoch)
+    num_batches_per_epoch = dataloader.num_batches
+    sample_digits = math.ceil(math.log(dataloader.num_samples + 1, 10))
+
+    # for toy dataset
+    if args.dataset_type == "toy":
+        dataloader.dataset.generate_queue()
+
+    loss_m = AverageMeter()
+    batch_time_m = AverageMeter()
+    data_time_m = AverageMeter()
+    end = time.time()
+
+    for i, batch in enumerate(dataloader):
+        # logging.info(f"batch {i} of {num_batches_per_epoch}")
+        step = num_batches_per_epoch * epoch + i
+        if isinstance(scheduler, dict):
+            for s in scheduler.values():
+                s(step)
+        else:
+            scheduler(step)
+        audios = batch  # contains mel_spec, wavform, and longer list
+        texts = batch["text"]
+        # audios = audios.to(device=device, non_blocking=True)
+        # texts = texts.to(device=device, non_blocking=True)
+
+        data_time_m.update(time.time() - end)
+        if isinstance(optimizer, dict):
+            for o_ in optimizer.values():
+                o_.zero_grad()
+        else:
+            optimizer.zero_grad()
+
+        with autocast():
+            (
+                audio_features,
+                text_features,
+                audio_features_mlp,
+                text_features_mlp,
+                logit_scale_a,
+                logit_scale_t,
+            ) = model(audios, texts, device)
+
+            if args.clap_mlploss:
+                total_loss = loss(
+                    audio_features=audio_features,
+                    text_features=text_features,
+                    logit_scale_a=logit_scale_a,
+                    logit_scale_t=logit_scale_t,
+                    audio_features_mlp=audio_features_mlp,
+                    text_features_mlp=text_features_mlp,
+                )
+            else:
+                total_loss = loss(
+                    audio_features=audio_features,
+                    text_features=text_features,
+                    logit_scale_a=logit_scale_a,
+                )
+        if isinstance(optimizer, dict):
+            if scaler is not None:
+                scaler.scale(total_loss).backward()
+                for o_ in optimizer.values():
+                    if args.horovod:
+                        o_.synchronize()
+                        scaler.unscale_(o_)
+                        with o_.skip_synchronize():
+                            scaler.step(o_)
+                    else:
+                        scaler.step(o_)
+                scaler.update()
+            else:
+                total_loss.backward()
+                for o_ in optimizer.values():
+                    o_.step()
+        else:
+            if scaler is not None:
+                scaler.scale(total_loss).backward()
+                if args.horovod:
+                    optimizer.synchronize()
+                    scaler.unscale_(optimizer)
+                    with optimizer.skip_synchronize():
+                        scaler.step(optimizer)
+                else:
+                    scaler.step(optimizer)
+                scaler.update()
+            else:
+                total_loss.backward()
+                optimizer.step()
+
+        # Note: we clamp to 4.6052 = ln(100), as in the original paper.
+        with torch.no_grad():
+            unwrap_model(model).logit_scale_a.clamp_(0, math.log(100))
+            if args.clap_mlploss:
+                unwrap_model(model).logit_scale_t.clamp_(0, math.log(100))
+
+        batch_time_m.update(time.time() - end)
+        end = time.time()
+        batch_count = i + 1
+        if is_master(args) and (i % 100 == 0 or batch_count == num_batches_per_epoch):
+            if isinstance(audios, dict):
+                batch_size = len(audios["waveform"])
+            else:
+                batch_size = len(audios)
+            num_samples = batch_count * batch_size * args.world_size
+            samples_per_epoch = dataloader.num_samples
+            percent_complete = 100.0 * batch_count / num_batches_per_epoch
+
+            # NOTE loss is coarsely sampled, just master node and per log update
+            loss_m.update(total_loss.item(), batch_size)
+            logit_scale_scalar_a = logit_scale_a.item()
+            logit_scale_scalar_t = logit_scale_t.item()
+            if isinstance(optimizer, dict):
+                if args.clap_mlploss:
+                    logging.info(
+                        f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                        f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                        f"Data (t): {data_time_m.avg:.3f} "
+                        f"Batch (t): {batch_time_m.avg:.3f} "
+                        f"LR: {[o_.param_groups[0]['lr'] for o_ in optimizer.values()]} "
+                        f"Logit Scale Audio: {logit_scale_scalar_a:.3f}"
+                        f"Logit Scale Text: {logit_scale_scalar_t:.3f}"
+                    )
+                    log_data = {
+                        "loss": loss_m.val,
+                        "data_time": data_time_m.val,
+                        "batch_time": batch_time_m.val,
+                        "scale_audio": logit_scale_scalar_a,
+                        "scale_text": logit_scale_scalar_t,
+                        "lr": [o_.param_groups[0]["lr"] for o_ in optimizer.values()],
+                    }
+                else:
+                    logging.info(
+                        f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                        f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                        f"Data (t): {data_time_m.avg:.3f} "
+                        f"Batch (t): {batch_time_m.avg:.3f} "
+                        f"LR: {[o_.param_groups[0]['lr'] for o_ in optimizer.values()]} "
+                        f"Logit Scale Audio: {logit_scale_scalar_a:.3f}"
+                    )
+                    log_data = {
+                        "loss": loss_m.val,
+                        "data_time": data_time_m.val,
+                        "batch_time": batch_time_m.val,
+                        "scale_audio": logit_scale_scalar_a,
+                        "lr": [o_.param_groups[0]["lr"] for o_ in optimizer.values()],
+                    }
+
+            else:
+                if args.clap_mlploss:
+                    logging.info(
+                        f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                        f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                        f"Data (t): {data_time_m.avg:.3f} "
+                        f"Batch (t): {batch_time_m.avg:.3f} "
+                        f"LR: {optimizer.param_groups[0]['lr']:5f} "
+                        f"Logit Scale Audio: {logit_scale_scalar_a:.3f}"
+                        f"Logit Scale Text: {logit_scale_scalar_t:.3f}"
+                    )
+
+                    # Save train loss / etc. Using non avg meter values as loggers have their own smoothing
+                    log_data = {
+                        "loss": loss_m.val,
+                        "data_time": data_time_m.val,
+                        "batch_time": batch_time_m.val,
+                        "scale_audio": logit_scale_scalar_a,
+                        "scale_text": logit_scale_scalar_t,
+                        "lr": optimizer.param_groups[0]["lr"],
+                    }
+                else:
+                    logging.info(
+                        f"Train Epoch: {epoch} [{num_samples:>{sample_digits}}/{samples_per_epoch} ({percent_complete:.0f}%)] "
+                        f"Loss: {loss_m.val:#.5g} ({loss_m.avg:#.4g}) "
+                        f"Data (t): {data_time_m.avg:.3f} "
+                        f"Batch (t): {batch_time_m.avg:.3f} "
+                        f"LR: {optimizer.param_groups[0]['lr']:5f} "
+                        f"Logit Scale Audio: {logit_scale_scalar_a:.3f}"
+                    )
+
+                    # Save train loss / etc. Using non avg meter values as loggers have their own smoothing
+                    log_data = {
+                        "loss": loss_m.val,
+                        "data_time": data_time_m.val,
+                        "batch_time": batch_time_m.val,
+                        "scale_audio": logit_scale_scalar_a,
+                        "lr": optimizer.param_groups[0]["lr"],
+                    }
+            for name, val in log_data.items():
+                name = "train/" + name
+                if tb_writer is not None:
+                    tb_writer.add_scalar(name, val, step)
+                if args.wandb:
+                    assert wandb is not None, "Please install wandb."
+                    wandb.log({name: val, "step": step})
+
+            # resetting batch / data time meters per log window
+            batch_time_m.reset()
+            data_time_m.reset()
+    # end for
+
+
+def evaluate(model, data, epoch, args, tb_writer=None):
+    metrics = {}
+    if not args.parallel_eval:
+        if not is_master(args):
+            return metrics
+    device = torch.device(args.device)
+    model.eval()
+
+    # CHANGE
+    # zero_shot_metrics = zero_shot_eval(model, data, epoch, args)
+    # metrics.update(zero_shot_metrics)
+    if is_master(args):
+        print("Evaluating...")
+    autocast = torch.cuda.amp.autocast if args.precision == "amp" else suppress
+    if args.val_dataset_names == ["Clotho", "audiocaps"]:
+        # if only clotho and audiocaps are used, then we will use a different evaluation function.
+        # This is because in the Clotho and audiocaps valid and test set, there are 5 text for 1 audio.
+        if args.parallel_eval:
+            # (yusong): just a hack here. Don't use parallel eval when evaluating only clotho and audiocaps.
+            raise NotImplementedError(
+                "Parallel evaluation not supported for eval only Clotho and audiocaps."
+            )
+        val_metrics_per_dataset = evaluate_clotho_audiocaps(
+            model, data, epoch, args, autocast, device, tb_writer
+        )
+        for m in val_metrics_per_dataset.values():
+            metrics.update(m)
+        if "epoch" not in metrics.keys():
+            metrics.update({"epoch": epoch})
+        metrics = select_top_metric_clotho_audiocaps(
+            metrics, val_metrics_per_dataset, args
+        )
+    elif "val" in data and (
+        args.val_frequency
+        and ((epoch % args.val_frequency) == 0 or epoch == args.epochs)
+    ):
+        dataloader = data["val"].dataloader
+        num_samples = 0
+        samples_per_val = dataloader.num_samples
+
+        # FIXME this does not scale past small eval datasets
+        # all_audio_features @ all_text_features will blow up memory and compute very quickly
+        eval_info = {}
+        if args.clap_mlploss:
+            eval_info["all"] = {
+                "cumulative_loss": 0.0,
+                "num_samples": 0,
+                "all_audio_features": [],
+                "all_text_features": [],
+                "all_audio_features_mlp": [],
+                "all_text_features_mlp": [],
+            }  # cumulative_loss = 0.0
+        else:
+            eval_info["all"] = {
+                "cumulative_loss": 0.0,
+                "num_samples": 0,
+                "all_audio_features": [],
+                "all_text_features": [],
+            }  # cumu
+        # all_audio_features, all_text_features, all_audio_features_mlp, all_text_features_mlp = [], [], [], []
+        with torch.no_grad():
+            for i, batch in enumerate(dataloader):
+                audios = batch  # contains mel_spec, wavform, and longer list
+                texts = batch["text"]
+                # audios = audios.to(device=device, non_blocking=True)
+
+                all_names = list(
+                    set(["-".join(b.split("/")[-3:-1]) for b in batch["__url__"]])
+                )
+                for name in all_names:
+                    if name not in eval_info.keys():
+                        if args.clap_mlploss:
+                            eval_info[name] = {
+                                "cumulative_loss": 0.0,
+                                "num_samples": 0,
+                                "all_audio_features": [],
+                                "all_text_features": [],
+                                "all_audio_features_mlp": [],
+                                "all_text_features_mlp": [],
+                            }
+                        else:
+                            eval_info[name] = {
+                                "cumulative_loss": 0.0,
+                                "num_samples": 0,
+                                "all_audio_features": [],
+                                "all_text_features": [],
+                            }
+                with autocast():
+                    (
+                        audio_features,
+                        text_features,
+                        audio_features_mlp,
+                        text_features_mlp,
+                        logit_scale_a,
+                        logit_scale_t,
+                    ) = model(audios, texts, device)
+
+                    if args.parallel_eval:
+                        # multi-GPU eval
+                        if args.clap_mlploss:
+                            (
+                                audio_features,
+                                text_features,
+                                audio_features_mlp,
+                                text_features_mlp,
+                            ) = gather_features(
+                                audio_features=audio_features,
+                                text_features=text_features,
+                                audio_features_mlp=audio_features_mlp,
+                                text_features_mlp=text_features_mlp,
+                                local_loss=False,
+                                gather_with_grad=False,
+                                rank=args.rank,
+                                world_size=args.world_size,
+                                use_horovod=args.horovod,
+                                mlp_loss=args.clap_mlploss,
+                            )
+                        else:
+                            (audio_features, text_features,) = gather_features(
+                                audio_features=audio_features,
+                                text_features=text_features,
+                                local_loss=False,
+                                gather_with_grad=False,
+                                rank=args.rank,
+                                world_size=args.world_size,
+                                use_horovod=args.horovod,
+                                mlp_loss=args.clap_mlploss,
+                            )
+
+                    if is_master(args):
+                        num_samples += audio_features.shape[0]
+                        for n in [*all_names, "all"]:
+                            if n == "all":
+                                eval_info[n]["all_audio_features"].append(
+                                    audio_features.cpu()
+                                )
+                                eval_info[n]["all_text_features"].append(
+                                    text_features.cpu()
+                                )
+                                if args.clap_mlploss:
+                                    eval_info[n]["all_audio_features_mlp"].append(
+                                        audio_features_mlp.cpu()
+                                    )
+                                    eval_info[n]["all_text_features_mlp"].append(
+                                        text_features_mlp.cpu()
+                                    )
+                            else:
+                                idx = np.where(
+                                    np.array(
+                                        [
+                                            "-".join(b.split("/")[-3:-1])
+                                            for b in batch["__url__"]
+                                        ]
+                                    )
+                                    == n
+                                )[0]
+                                eval_info[n]["all_audio_features"].append(
+                                    audio_features.cpu().index_select(
+                                        0, torch.tensor(idx).long()
+                                    )
+                                )
+                                eval_info[n]["all_text_features"].append(
+                                    text_features.cpu().index_select(
+                                        0, torch.tensor(idx).long()
+                                    )
+                                )
+                                if args.clap_mlploss:
+                                    eval_info[n]["all_audio_features_mlp"].append(
+                                        audio_features_mlp.cpu().index_select(
+                                            0, torch.tensor(idx).long()
+                                        )
+                                    )
+                                    eval_info[n]["all_text_features_mlp"].append(
+                                        text_features_mlp.cpu().index_select(
+                                            0, torch.tensor(idx).long()
+                                        )
+                                    )
+                        #  print(f'eval step {i}') #  (yusong): for debug
+
+                # cumulative_loss += total_loss * batch_size
+                # num_samples += batch_size
+                if is_master(args) and (i % 100) == 0:  # and i != 0:
+                    logging.info(
+                        f"Eval Epoch: {epoch} [{num_samples} / {samples_per_val}]"
+                    )
+            if is_master(args):
+                val_metrics_per_dataset = {}
+                for n in eval_info.keys():
+                    if args.clap_mlploss:
+                        metrics_single_dataset = get_metrics(
+                            audio_features=torch.cat(
+                                eval_info[n]["all_audio_features"]
+                            ),
+                            text_features=torch.cat(eval_info[n]["all_text_features"]),
+                            logit_scale_a=logit_scale_a.cpu(),
+                            audio_features_mlp=torch.cat(
+                                eval_info[n]["all_audio_features_mlp"]
+                            ),
+                            text_features_mlp=torch.cat(
+                                eval_info[n]["all_text_features_mlp"]
+                            ),
+                            logit_scale_t=logit_scale_t.cpu(),
+                            mlp_loss=args.clap_mlploss,
+                        )
+                    else:
+                        metrics_single_dataset = get_metrics(
+                            audio_features=torch.cat(
+                                eval_info[n]["all_audio_features"]
+                            ),
+                            text_features=torch.cat(eval_info[n]["all_text_features"]),
+                            logit_scale_a=logit_scale_a.cpu(),
+                            mlp_loss=args.clap_mlploss,
+                        )
+                    val_metrics_per_dataset[n] = {
+                        n + "/" + k: v for k, v in metrics_single_dataset.items()
+                    }
+                    metrics.update(val_metrics_per_dataset[n])
+                    if "epoch" not in metrics.keys():
+                        metrics.update({"epoch": epoch})
+    if is_master(args):
+        if not metrics:
+            return metrics
+
+        logging.info(
+            f"Eval Epoch: {epoch} "
+            + "\n".join(
+                [
+                    "\t".join([f"{k}: {round(v, 4):.4f}" for k, v in m.items()])
+                    for m in val_metrics_per_dataset.values()
+                ]
+            )
+        )
+
+        if args.save_logs:
+            for name, val in metrics.items():
+                if tb_writer is not None:
+                    tb_writer.add_scalar(f"val/{name}", val, epoch)
+
+            with open(os.path.join(args.checkpoint_path, "results.jsonl"), "a+") as f:
+                f.write(json.dumps(metrics))
+                f.write("\n")
+
+        if args.wandb:
+            assert wandb is not None, "Please install wandb."
+            for name, val in metrics.items():
+                wandb.log({f"val/{name}": val, "epoch": epoch})
+
+        return metrics
+    else:
+        return metrics
+
+
+def get_metrics(
+    audio_features,
+    text_features,
+    logit_scale_a,
+    audio_features_mlp=None,
+    text_features_mlp=None,
+    logit_scale_t=None,
+    mlp_loss=False,
+):
+    metrics = {}
+    if mlp_loss:
+        # Set up audio to text & text to audio similary matrice
+        a_logits_per_audio = (
+            (logit_scale_a * audio_features @ text_features_mlp.t()).detach().cpu()
+        )
+        a_logits_per_text = a_logits_per_audio.t().detach().cpu()
+        t_logits_per_audio = (
+            (logit_scale_t * audio_features_mlp @ text_features.t()).detach().cpu()
+        )
+        t_logits_per_text = t_logits_per_audio.t().detach().cpu()
+
+        labels = torch.arange(audio_features.shape[0]).long()
+        # Change the loss from two terms into four terms with 2x2 combined CE loss
+        total_loss = (
+            F.cross_entropy(a_logits_per_audio, labels)
+            + F.cross_entropy(a_logits_per_text, labels)
+            + F.cross_entropy(t_logits_per_audio, labels)
+            + F.cross_entropy(t_logits_per_text, labels)
+        ) / 4
+
+        metrics[f"cumulative_loss"] = total_loss.item()
+        metrics[f"num_samples"] = audio_features.shape[0]
+
+        logits = {
+            "audio_to_text": (a_logits_per_audio + t_logits_per_audio) / 2,
+            "text_to_audio": (a_logits_per_text + t_logits_per_text) / 2,
+        }
+        ground_truth = torch.arange(len(text_features)).view(-1, 1)
+
+    else:
+        # print("text_features", text_features)
+        # print("text_features.shape", text_features.shape)
+        logits_per_audio = (
+            (logit_scale_a * audio_features @ text_features.t()).detach().cpu()
+        )
+        logits_per_text = logits_per_audio.t().detach().cpu()
+
+        labels = torch.arange(audio_features.shape[0]).long()
+        # Change the loss from two terms into four terms with 2x2 combined CE loss
+        total_loss = (
+            F.cross_entropy(logits_per_audio, labels)
+            + F.cross_entropy(logits_per_text, labels)
+        ) / 2
+
+        metrics[f"cumulative_loss"] = total_loss.item()
+        metrics[f"num_samples"] = audio_features.shape[0]
+
+        logits = {"audio_to_text": logits_per_audio, "text_to_audio": logits_per_text}
+
+        ground_truth = torch.arange(len(text_features)).view(-1, 1)
+
+    for name, logit in logits.items():
+        ranking = torch.argsort(logit, descending=True)
+        preds = torch.where(ranking == ground_truth)[
+            1
+        ]  # (yusong) this line is slow because it uses single thread
+        preds = preds.detach().cpu().numpy()
+        metrics[f"{name}_mean_rank"] = preds.mean() + 1
+        metrics[f"{name}_median_rank"] = np.floor(np.median(preds)) + 1
+        for k in [1, 5, 10]:
+            metrics[f"{name}_R@{k}"] = np.mean(preds < k)
+        # map@10
+        metrics[f"{name}_mAP@10"] = np.mean(np.where(preds < 10, 1 / (preds + 1), 0.0))
+
+    return metrics
+
+
+def evaluate_clotho_audiocaps(
+    model, data, epoch, args, autocast, device, tb_writer=None
+):
+    """
+    Adapted from https://github.com/XinhaoMei/audio-text_retrieval/blob/main/tools/utils.py.
+    1. for text-to-audio retrieval, do 5 times and average the results
+    2. for R@1, R@5, R@10 in audio-to-text retrieval, take the best rank among 5 text
+    3. for map@10 in audio-to-text retrieval:
+        3.1: sort the rank of 5 text
+        3.2: exclude the rank >=10 (0-index)
+        3.3: compute the map regarding the remaining ranks: np.mean(np.arange(1, len(ranks)+1) / ranks).
+        (3.3) That is, take the top ranks of 5 text that is < 10, and assign the descending number as ground truth.
+        (3.3) E.g.: the ground truth of first rank of the 5 text should be 1, the second rank should be 2, etc.
+    """
+    # TODO: (yusong) only support single GPU evaluation and only support non-mlp case for now.
+    dataloader = data["val"].dataloader
+    with torch.no_grad():
+        eval_info = {}
+        for i, batch in enumerate(dataloader):
+            audios = batch  # contains mel_spec, wavform, and longer list
+
+            # each item in the list has 5 texts
+            if args.tmodel == "transformer":
+                from open_clip import tokenize
+
+                texts = [tokenize(t) for t in batch["full_text"]]
+                texts = torch.cat(texts)
+            else:
+                from .data import tokenizer
+
+                texts = [
+                    tokenizer(t) for t in batch["full_text"]
+                ]  # 5 texts for each audio
+                texts = {
+                    k: torch.cat([t[k] for t in texts]) for k in texts[0].keys()
+                }  # 5 x batch
+
+            # audios = audios.to(device=device, non_blocking=True)
+
+            all_names = list(
+                set(["-".join(b.split("/")[-3:-1]) for b in batch["__url__"]])
+            )
+            for name in all_names:
+                if name not in eval_info.keys():
+                    # we will not use mlp outputs even if args.clap_mlploss=True
+                    eval_info[name] = {
+                        "cumulative_loss": 0.0,
+                        "num_samples": 0,
+                        "all_audio_features": [],
+                        "all_text_features": [],
+                    }
+            with autocast():
+                audio_features = model(audios, None, device)
+                text_features = model(None, texts, device)
+                audio_features = F.normalize(audio_features, dim=-1)
+                text_features = F.normalize(text_features, dim=-1)
+
+                all_names = list(
+                    set(["-".join(b.split("/")[-3:-1]) for b in batch["__url__"]])
+                )
+                for n in all_names:
+                    idx = np.where(
+                        np.array(
+                            ["-".join(b.split("/")[-3:-1]) for b in batch["__url__"]]
+                        )
+                        == n
+                    )[0]
+                    eval_info[n]["all_audio_features"].append(
+                        audio_features.cpu().index_select(0, torch.tensor(idx).long())
+                    )
+                    # (yusong) please double-check. This is for selecting 5 text features at once.
+                    # because idx is a list of indices in size of num_samples,
+                    # and text_features is a tensor of size (5*num_samples, dim)
+                    # so we need to select 5 consecutive indices at once for a single index in idx.
+                    eval_info[n]["all_text_features"].append(
+                        text_features.cpu()
+                        .reshape([-1, 5, text_features.shape[1]])
+                        .index_select(0, torch.tensor(idx).long())
+                        .reshape([-1, text_features.shape[1]])
+                    )
+
+        val_metrics_all = {}
+
+        for n in eval_info.keys():
+            logit_scale_a, logit_scale_t = model(None, None, device)
+            logit_scale_a = logit_scale_a.cpu()
+
+            audio_features = torch.cat(eval_info[n]["all_audio_features"], dim=0)
+            text_features = torch.cat(eval_info[n]["all_text_features"], dim=0)
+
+            logits_per_audio = (
+                (logit_scale_a * audio_features @ text_features.t()).detach().cpu()
+            )
+            logits_per_text = logits_per_audio.t().detach().cpu()
+
+            # logits_per_audio shape: [num_samples, num_samples*5]
+            # logits_per_text shape: [num_samples*5, num_samples]
+
+            logging.info(
+                f"dataset {n}, logits_per_audio shape: {logits_per_audio.shape}, "
+                f"logits_per_text shape: {logits_per_text.shape}"
+            )
+
+            metrics = {}
+            num_samples = audio_features.shape[0]
+            metrics[f"num_samples"] = num_samples
+
+            # (yusong) the following code is very important, please double-check:
+            # logits_per_audio.reshape(num_samples, num_samples, 5)[:, :, d]
+            # logits_per_text.reshape(num_samples, 5, num_samples)[:, d, :]
+            # Those two are retrieving one of the 5 text for each audio.
+            labels = torch.arange(audio_features.shape[0]).long()
+            audio_to_text_loss = [
+                F.cross_entropy(
+                    logits_per_audio.reshape(num_samples, num_samples, 5)[:, :, d],
+                    labels,
+                )
+                for d in range(5)
+            ]
+            text_to_audio_loss = [
+                F.cross_entropy(
+                    logits_per_text.reshape(num_samples, 5, num_samples)[:, d, :],
+                    labels,
+                )
+                for d in range(5)
+            ]
+            total_loss = (np.mean(audio_to_text_loss) + np.mean(text_to_audio_loss)) / 2
+
+            metrics[f"cumulative_loss"] = total_loss.item()
+
+            # text to audio: do 5 times
+            pred_text = []
+            for d in range(5):
+                logit = logits_per_text.reshape(num_samples, 5, num_samples)[:, d, :]
+                ground_truth = torch.arange(len(logit)).view(-1, 1)
+                ranking = torch.argsort(
+                    logit, descending=True
+                )  # [num_samples, num_samples]
+                preds = torch.where(ranking == ground_truth)[1]
+                pred_text.append(preds.detach().cpu().numpy())
+            pred_text_concat = np.concatenate(pred_text, axis=0)  # [5*num_samples]
+            metrics[f"text_to_audio_mean_rank"] = pred_text_concat.mean() + 1
+            metrics[f"text_to_audio_median_rank"] = (
+                np.floor(np.median(pred_text_concat)) + 1
+            )
+            for k in [1, 5, 10]:
+                metrics[f"text_to_audio_R@{k}"] = np.mean(pred_text_concat < k)
+            # map@10
+            metrics[f"text_to_audio_mAP@10"] = np.mean(
+                np.where(pred_text_concat < 10, 1 / (pred_text_concat + 1), 0.0)
+            )
+
+            # audio to text: take the best result
+            # for audio to text map 10, sort and assign descending ground truth.
+            # see https://github.com/XinhaoMei/audio-text_retrieval/blob/main/tools/utils.py#L103
+            # map@10
+            map_all = []
+            pred_audio_all = []
+            for d in range(num_samples):
+                # logits_per_audio: [num_samples, num_samples*5]
+                logit_single = logits_per_audio[d, :]  # [5*num_samples]
+                # Ground-truth index: [d*5, d*5+1, d*5+2, d*5+3, d*5+4]
+                ranking = torch.argsort(
+                    logit_single, descending=True
+                )  # [5*num_samples]
+                # ranking: the index of first match, second match, ...
+                ground_truth = torch.arange(d * 5, d * 5 + 5)[None]
+                all_pred = torch.where(
+                    torch.stack([ranking] * 5) == ground_truth.view(-1, 1)
+                )[1]
+                min_pred = torch.min(all_pred)
+                pred_audio_all.append(min_pred.detach().cpu().numpy())
+                all_pred_filter = all_pred[all_pred < 10].detach().cpu().numpy()
+                # /5 because we have 5 text, so it means for the text rank >=10 we count as 0.
+                map_single = (
+                    np.sum(
+                        (np.arange(1, len(all_pred_filter) + 1) / (all_pred_filter + 1))
+                    )
+                    / 5
+                )
+                map_all.append(map_single)
+            metrics[f"audio_to_text_mAP@10"] = np.mean(map_all)
+            for k in [1, 5, 10]:
+                metrics[f"audio_to_text_R@{k}"] = np.mean(np.array(pred_audio_all) < k)
+
+            val_metrics_all[n] = {n + "/" + k: v for k, v in metrics.items()}
+    return val_metrics_all
+
+
+def calculate_selection_performance_clotho_audiocaps(val_metrics_per_dataset):
+    """
+    Calculate performance for Clotho+AudioCaps for model selection.
+    """
+    selection_performance_all = []
+    for n in val_metrics_per_dataset.keys():
+        selection_performance = (
+            val_metrics_per_dataset[n][f"{n}/audio_to_text_mAP@10"]
+            + val_metrics_per_dataset[n][f"{n}/text_to_audio_mAP@10"]
+        ) / 2
+        selection_performance_all.append(selection_performance)
+    return np.mean(selection_performance_all)
+
+
+def select_top_metric_clotho_audiocaps(metrics, val_metrics_per_dataset, args):
+    # val_metrics_per_dataset: dict, key: dataset name, value: dict, key: metric name, value: metric value
+    # metrics: dict, key: metric name, value: metric value
+    # Hack: use args to save the top performance
+    if not hasattr(args, "top_selection_performance"):
+        selection_performance = calculate_selection_performance_clotho_audiocaps(
+            val_metrics_per_dataset
+        )
+        # TODO: write the if and else together
+        metric_update = {}
+        for n in val_metrics_per_dataset.keys():
+            for k in val_metrics_per_dataset[n].keys():
+                metric_update[
+                    k.split("/")[0] + "-top" + "/" + k.split("/")[1]
+                ] = val_metrics_per_dataset[n][k]
+        metric_update["top_selection_performance"] = selection_performance
+        metric_update["top-selection-epoch"] = metrics["epoch"]
+        metrics.update(metric_update)
+        args.top_metric = metric_update
+        args.top_selection_performance = selection_performance
+    else:
+        selection_performance_new = calculate_selection_performance_clotho_audiocaps(
+            val_metrics_per_dataset
+        )
+        selection_performance_old = args.top_selection_performance
+        if selection_performance_new > selection_performance_old:
+            metric_update = {}
+            for n in val_metrics_per_dataset.keys():
+                for k in val_metrics_per_dataset[n].keys():
+                    metric_update[
+                        k.split("/")[0] + "-top" + "/" + k.split("/")[1]
+                    ] = val_metrics_per_dataset[n][k]
+            metric_update["top_selection_performance"] = selection_performance_new
+            metric_update["top-selection-epoch"] = metrics["epoch"]
+            metrics.update(metric_update)
+            args.top_metric = metric_update
+            args.top_selection_performance = selection_performance_new
+        else:
+            metrics.update(args.top_metric)
+    return metrics
diff --git a/picoaudio/audioldm/clap/training/zero_shot.py b/picoaudio/audioldm/clap/training/zero_shot.py
new file mode 100644
index 0000000000000000000000000000000000000000..28b8fccc1af17fc69002857a7f529ac041c374f2
--- /dev/null
+++ b/picoaudio/audioldm/clap/training/zero_shot.py
@@ -0,0 +1,95 @@
+# NOTE: This script is currently not supported for CLAP.
+import logging
+from contextlib import suppress
+
+import torch
+import torch.nn.functional as F
+from tqdm import tqdm
+
+from open_clip import tokenize
+from .imagenet_zeroshot_data import imagenet_classnames, openai_imagenet_template
+
+
+def zero_shot_classifier(model, classnames, templates, args):
+    with torch.no_grad():
+        zeroshot_weights = []
+        for classname in tqdm(classnames):
+            texts = [template(classname) for template in templates]  # format with class
+            texts = tokenize(texts).to(args.device)  # tokenize
+            if args.distributed and not args.horovod:
+                class_embeddings = model.module.encode_text(texts)
+            else:
+                class_embeddings = model.encode_text(texts)
+            class_embedding = F.normalize(class_embeddings, dim=-1).mean(dim=0)
+            class_embedding /= class_embedding.norm()
+            zeroshot_weights.append(class_embedding)
+        zeroshot_weights = torch.stack(zeroshot_weights, dim=1).to(args.device)
+    return zeroshot_weights
+
+
+def accuracy(output, target, topk=(1,)):
+    pred = output.topk(max(topk), 1, True, True)[1].t()
+    correct = pred.eq(target.view(1, -1).expand_as(pred))
+    return [
+        float(correct[:k].reshape(-1).float().sum(0, keepdim=True).cpu().numpy())
+        for k in topk
+    ]
+
+
+def run(model, classifier, dataloader, args):
+    autocast = torch.cuda.amp.autocast if args.precision == "amp" else suppress
+    with torch.no_grad():
+        top1, top5, n = 0.0, 0.0, 0.0
+        for images, target in tqdm(dataloader, unit_scale=args.batch_size):
+            images = images.to(args.device)
+            target = target.to(args.device)
+
+            with autocast():
+                # predict
+                if args.distributed and not args.horovod:
+                    image_features = model.module.encode_image(images)
+                else:
+                    image_features = model.encode_image(images)
+                image_features = F.normalize(image_features, dim=-1)
+                logits = 100.0 * image_features @ classifier
+
+            # measure accuracy
+            acc1, acc5 = accuracy(logits, target, topk=(1, 5))
+            top1 += acc1
+            top5 += acc5
+            n += images.size(0)
+
+    top1 = top1 / n
+    top5 = top5 / n
+    return top1, top5
+
+
+def zero_shot_eval(model, data, epoch, args):
+    if "imagenet-val" not in data and "imagenet-v2" not in data:
+        return {}
+    if args.zeroshot_frequency == 0:
+        return {}
+    if (epoch % args.zeroshot_frequency) != 0 and epoch != args.epochs:
+        return {}
+
+    logging.info("Starting zero-shot imagenet.")
+
+    logging.info("Building zero-shot classifier")
+    classifier = zero_shot_classifier(
+        model, imagenet_classnames, openai_imagenet_template, args
+    )
+
+    logging.info("Using classifier")
+    results = {}
+    if "imagenet-val" in data:
+        top1, top5 = run(model, classifier, data["imagenet-val"].dataloader, args)
+        results["imagenet-zeroshot-val-top1"] = top1
+        results["imagenet-zeroshot-val-top5"] = top5
+    if "imagenet-v2" in data:
+        top1, top5 = run(model, classifier, data["imagenet-v2"].dataloader, args)
+        results["imagenetv2-zeroshot-val-top1"] = top1
+        results["imagenetv2-zeroshot-val-top5"] = top5
+
+    logging.info("Finished zero-shot imagenet.")
+
+    return results
diff --git a/picoaudio/audioldm/hifigan/__init__.py b/picoaudio/audioldm/hifigan/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e0ae476fe58c48e998c56234a55b871beba4042d
--- /dev/null
+++ b/picoaudio/audioldm/hifigan/__init__.py
@@ -0,0 +1,7 @@
+from .models import Generator
+
+
+class AttrDict(dict):
+    def __init__(self, *args, **kwargs):
+        super(AttrDict, self).__init__(*args, **kwargs)
+        self.__dict__ = self
diff --git a/picoaudio/audioldm/hifigan/models.py b/picoaudio/audioldm/hifigan/models.py
new file mode 100644
index 0000000000000000000000000000000000000000..c4382cc39de0463f9b7c0f33f037dbc233e7cb36
--- /dev/null
+++ b/picoaudio/audioldm/hifigan/models.py
@@ -0,0 +1,174 @@
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from torch.nn import Conv1d, ConvTranspose1d
+from torch.nn.utils import weight_norm, remove_weight_norm
+
+LRELU_SLOPE = 0.1
+
+
+def init_weights(m, mean=0.0, std=0.01):
+    classname = m.__class__.__name__
+    if classname.find("Conv") != -1:
+        m.weight.data.normal_(mean, std)
+
+
+def get_padding(kernel_size, dilation=1):
+    return int((kernel_size * dilation - dilation) / 2)
+
+
+class ResBlock(torch.nn.Module):
+    def __init__(self, h, channels, kernel_size=3, dilation=(1, 3, 5)):
+        super(ResBlock, self).__init__()
+        self.h = h
+        self.convs1 = nn.ModuleList(
+            [
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[0],
+                        padding=get_padding(kernel_size, dilation[0]),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[1],
+                        padding=get_padding(kernel_size, dilation[1]),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=dilation[2],
+                        padding=get_padding(kernel_size, dilation[2]),
+                    )
+                ),
+            ]
+        )
+        self.convs1.apply(init_weights)
+
+        self.convs2 = nn.ModuleList(
+            [
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+                weight_norm(
+                    Conv1d(
+                        channels,
+                        channels,
+                        kernel_size,
+                        1,
+                        dilation=1,
+                        padding=get_padding(kernel_size, 1),
+                    )
+                ),
+            ]
+        )
+        self.convs2.apply(init_weights)
+
+    def forward(self, x):
+        for c1, c2 in zip(self.convs1, self.convs2):
+            xt = F.leaky_relu(x, LRELU_SLOPE)
+            xt = c1(xt)
+            xt = F.leaky_relu(xt, LRELU_SLOPE)
+            xt = c2(xt)
+            x = xt + x
+        return x
+
+    def remove_weight_norm(self):
+        for l in self.convs1:
+            remove_weight_norm(l)
+        for l in self.convs2:
+            remove_weight_norm(l)
+
+
+class Generator(torch.nn.Module):
+    def __init__(self, h):
+        super(Generator, self).__init__()
+        self.h = h
+        self.num_kernels = len(h.resblock_kernel_sizes)
+        self.num_upsamples = len(h.upsample_rates)
+        self.conv_pre = weight_norm(
+            Conv1d(h.num_mels, h.upsample_initial_channel, 7, 1, padding=3)
+        )
+        resblock = ResBlock
+
+        self.ups = nn.ModuleList()
+        for i, (u, k) in enumerate(zip(h.upsample_rates, h.upsample_kernel_sizes)):
+            self.ups.append(
+                weight_norm(
+                    ConvTranspose1d(
+                        h.upsample_initial_channel // (2**i),
+                        h.upsample_initial_channel // (2 ** (i + 1)),
+                        k,
+                        u,
+                        padding=(k - u) // 2,
+                    )
+                )
+            )
+
+        self.resblocks = nn.ModuleList()
+        for i in range(len(self.ups)):
+            ch = h.upsample_initial_channel // (2 ** (i + 1))
+            for j, (k, d) in enumerate(
+                zip(h.resblock_kernel_sizes, h.resblock_dilation_sizes)
+            ):
+                self.resblocks.append(resblock(h, ch, k, d))
+
+        self.conv_post = weight_norm(Conv1d(ch, 1, 7, 1, padding=3))
+        self.ups.apply(init_weights)
+        self.conv_post.apply(init_weights)
+
+    def forward(self, x):
+        x = self.conv_pre(x)
+        for i in range(self.num_upsamples):
+            x = F.leaky_relu(x, LRELU_SLOPE)
+            x = self.ups[i](x)
+            xs = None
+            for j in range(self.num_kernels):
+                if xs is None:
+                    xs = self.resblocks[i * self.num_kernels + j](x)
+                else:
+                    xs += self.resblocks[i * self.num_kernels + j](x)
+            x = xs / self.num_kernels
+        x = F.leaky_relu(x)
+        x = self.conv_post(x)
+        x = torch.tanh(x)
+
+        return x
+
+    def remove_weight_norm(self):
+        # print("Removing weight norm...")
+        for l in self.ups:
+            remove_weight_norm(l)
+        for l in self.resblocks:
+            l.remove_weight_norm()
+        remove_weight_norm(self.conv_pre)
+        remove_weight_norm(self.conv_post)
diff --git a/picoaudio/audioldm/hifigan/utilities.py b/picoaudio/audioldm/hifigan/utilities.py
new file mode 100644
index 0000000000000000000000000000000000000000..ea9f958e460a77fd4936a6edf59403dd3ea617ab
--- /dev/null
+++ b/picoaudio/audioldm/hifigan/utilities.py
@@ -0,0 +1,86 @@
+import os
+import json
+
+import torch
+import numpy as np
+
+import audioldm.hifigan as hifigan
+
+HIFIGAN_16K_64 = {
+    "resblock": "1",
+    "num_gpus": 6,
+    "batch_size": 16,
+    "learning_rate": 0.0002,
+    "adam_b1": 0.8,
+    "adam_b2": 0.99,
+    "lr_decay": 0.999,
+    "seed": 1234,
+    "upsample_rates": [5, 4, 2, 2, 2],
+    "upsample_kernel_sizes": [16, 16, 8, 4, 4],
+    "upsample_initial_channel": 1024,
+    "resblock_kernel_sizes": [3, 7, 11],
+    "resblock_dilation_sizes": [[1, 3, 5], [1, 3, 5], [1, 3, 5]],
+    "segment_size": 8192,
+    "num_mels": 64,
+    "num_freq": 1025,
+    "n_fft": 1024,
+    "hop_size": 160,
+    "win_size": 1024,
+    "sampling_rate": 16000,
+    "fmin": 0,
+    "fmax": 8000,
+    "fmax_for_loss": None,
+    "num_workers": 4,
+    "dist_config": {
+        "dist_backend": "nccl",
+        "dist_url": "tcp://localhost:54321",
+        "world_size": 1,
+    },
+}
+
+
+def get_available_checkpoint_keys(model, ckpt):
+    print("==> Attemp to reload from %s" % ckpt)
+    state_dict = torch.load(ckpt)["state_dict"]
+    current_state_dict = model.state_dict()
+    new_state_dict = {}
+    for k in state_dict.keys():
+        if (
+            k in current_state_dict.keys()
+            and current_state_dict[k].size() == state_dict[k].size()
+        ):
+            new_state_dict[k] = state_dict[k]
+        else:
+            print("==> WARNING: Skipping %s" % k)
+    print(
+        "%s out of %s keys are matched"
+        % (len(new_state_dict.keys()), len(state_dict.keys()))
+    )
+    return new_state_dict
+
+
+def get_param_num(model):
+    num_param = sum(param.numel() for param in model.parameters())
+    return num_param
+
+
+def get_vocoder(config, device):
+    config = hifigan.AttrDict(HIFIGAN_16K_64)
+    vocoder = hifigan.Generator(config)
+    vocoder.eval()
+    vocoder.remove_weight_norm()
+    vocoder.to(device)
+    return vocoder
+
+
+def vocoder_infer(mels, vocoder, lengths=None):
+    vocoder.eval()
+    with torch.no_grad():
+        wavs = vocoder(mels).squeeze(1)
+
+    wavs = (wavs.cpu().numpy() * 32768).astype("int16")
+
+    if lengths is not None:
+        wavs = wavs[:, :lengths]
+
+    return wavs
diff --git a/picoaudio/audioldm/latent_diffusion/__init__.py b/picoaudio/audioldm/latent_diffusion/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..e69de29bb2d1d6434b8b29ae775ad8c2e48c5391
diff --git a/picoaudio/audioldm/latent_diffusion/attention.py b/picoaudio/audioldm/latent_diffusion/attention.py
new file mode 100644
index 0000000000000000000000000000000000000000..27886f5ee3c7eb856100503b838399106ef00051
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/attention.py
@@ -0,0 +1,469 @@
+from inspect import isfunction
+import math
+import torch
+import torch.nn.functional as F
+from torch import nn
+from einops import rearrange
+
+from audioldm.latent_diffusion.util import checkpoint
+
+
+def exists(val):
+    return val is not None
+
+
+def uniq(arr):
+    return {el: True for el in arr}.keys()
+
+
+def default(val, d):
+    if exists(val):
+        return val
+    return d() if isfunction(d) else d
+
+
+def max_neg_value(t):
+    return -torch.finfo(t.dtype).max
+
+
+def init_(tensor):
+    dim = tensor.shape[-1]
+    std = 1 / math.sqrt(dim)
+    tensor.uniform_(-std, std)
+    return tensor
+
+
+# feedforward
+class GEGLU(nn.Module):
+    def __init__(self, dim_in, dim_out):
+        super().__init__()
+        self.proj = nn.Linear(dim_in, dim_out * 2)
+
+    def forward(self, x):
+        x, gate = self.proj(x).chunk(2, dim=-1)
+        return x * F.gelu(gate)
+
+
+class FeedForward(nn.Module):
+    def __init__(self, dim, dim_out=None, mult=4, glu=False, dropout=0.0):
+        super().__init__()
+        inner_dim = int(dim * mult)
+        dim_out = default(dim_out, dim)
+        project_in = (
+            nn.Sequential(nn.Linear(dim, inner_dim), nn.GELU())
+            if not glu
+            else GEGLU(dim, inner_dim)
+        )
+
+        self.net = nn.Sequential(
+            project_in, nn.Dropout(dropout), nn.Linear(inner_dim, dim_out)
+        )
+
+    def forward(self, x):
+        return self.net(x)
+
+
+def zero_module(module):
+    """
+    Zero out the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def Normalize(in_channels):
+    return torch.nn.GroupNorm(
+        num_groups=32, num_channels=in_channels, eps=1e-6, affine=True
+    )
+
+
+class LinearAttention(nn.Module):
+    def __init__(self, dim, heads=4, dim_head=32):
+        super().__init__()
+        self.heads = heads
+        hidden_dim = dim_head * heads
+        self.to_qkv = nn.Conv2d(dim, hidden_dim * 3, 1, bias=False)
+        self.to_out = nn.Conv2d(hidden_dim, dim, 1)
+
+    def forward(self, x):
+        b, c, h, w = x.shape
+        qkv = self.to_qkv(x)
+        q, k, v = rearrange(
+            qkv, "b (qkv heads c) h w -> qkv b heads c (h w)", heads=self.heads, qkv=3
+        )
+        k = k.softmax(dim=-1)
+        context = torch.einsum("bhdn,bhen->bhde", k, v)
+        out = torch.einsum("bhde,bhdn->bhen", context, q)
+        out = rearrange(
+            out, "b heads c (h w) -> b (heads c) h w", heads=self.heads, h=h, w=w
+        )
+        return self.to_out(out)
+
+
+class SpatialSelfAttention(nn.Module):
+    def __init__(self, in_channels):
+        super().__init__()
+        self.in_channels = in_channels
+
+        self.norm = Normalize(in_channels)
+        self.q = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.k = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.v = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.proj_out = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+
+    def forward(self, x):
+        h_ = x
+        h_ = self.norm(h_)
+        q = self.q(h_)
+        k = self.k(h_)
+        v = self.v(h_)
+
+        # compute attention
+        b, c, h, w = q.shape
+        q = rearrange(q, "b c h w -> b (h w) c")
+        k = rearrange(k, "b c h w -> b c (h w)")
+        w_ = torch.einsum("bij,bjk->bik", q, k)
+
+        w_ = w_ * (int(c) ** (-0.5))
+        w_ = torch.nn.functional.softmax(w_, dim=2)
+
+        # attend to values
+        v = rearrange(v, "b c h w -> b c (h w)")
+        w_ = rearrange(w_, "b i j -> b j i")
+        h_ = torch.einsum("bij,bjk->bik", v, w_)
+        h_ = rearrange(h_, "b c (h w) -> b c h w", h=h)
+        h_ = self.proj_out(h_)
+
+        return x + h_
+
+
+class CrossAttention(nn.Module):
+    """
+    ### Cross Attention Layer
+    This falls-back to self-attention when conditional embeddings are not specified.
+    """
+
+    # use_flash_attention: bool = True
+    use_flash_attention: bool = False
+
+    def __init__(
+        self,
+        query_dim,
+        context_dim=None,
+        heads=8,
+        dim_head=64,
+        dropout=0.0,
+        is_inplace: bool = True,
+    ):
+        # def __init__(self, d_model: int, d_cond: int, n_heads: int, d_head: int, is_inplace: bool = True):
+        """
+        :param d_model: is the input embedding size
+        :param n_heads: is the number of attention heads
+        :param d_head: is the size of a attention head
+        :param d_cond: is the size of the conditional embeddings
+        :param is_inplace: specifies whether to perform the attention softmax computation inplace to
+            save memory
+        """
+        super().__init__()
+
+        self.is_inplace = is_inplace
+        self.n_heads = heads
+        self.d_head = dim_head
+
+        # Attention scaling factor
+        self.scale = dim_head**-0.5
+
+        # The normal self-attention layer
+        if context_dim is None:
+            context_dim = query_dim
+
+        # Query, key and value mappings
+        d_attn = dim_head * heads
+        self.to_q = nn.Linear(query_dim, d_attn, bias=False)
+        self.to_k = nn.Linear(context_dim, d_attn, bias=False)
+        self.to_v = nn.Linear(context_dim, d_attn, bias=False)
+
+        # Final linear layer
+        self.to_out = nn.Sequential(nn.Linear(d_attn, query_dim), nn.Dropout(dropout))
+
+        # Setup [flash attention](https://github.com/HazyResearch/flash-attention).
+        # Flash attention is only used if it's installed
+        # and `CrossAttention.use_flash_attention` is set to `True`.
+        try:
+            # You can install flash attention by cloning their Github repo,
+            # [https://github.com/HazyResearch/flash-attention](https://github.com/HazyResearch/flash-attention)
+            # and then running `python setup.py install`
+            from flash_attn.flash_attention import FlashAttention
+
+            self.flash = FlashAttention()
+            # Set the scale for scaled dot-product attention.
+            self.flash.softmax_scale = self.scale
+        # Set to `None` if it's not installed
+        except ImportError:
+            self.flash = None
+
+    def forward(self, x, context=None, mask=None):
+        """
+        :param x: are the input embeddings of shape `[batch_size, height * width, d_model]`
+        :param cond: is the conditional embeddings of shape `[batch_size, n_cond, d_cond]`
+        """
+
+        # If `cond` is `None` we perform self attention
+        has_cond = context is not None
+        if not has_cond:
+            context = x
+
+        # Get query, key and value vectors
+        q = self.to_q(x)
+        k = self.to_k(context)
+        v = self.to_v(context)
+
+        # Use flash attention if it's available and the head size is less than or equal to `128`
+        if (
+            CrossAttention.use_flash_attention
+            and self.flash is not None
+            and not has_cond
+            and self.d_head <= 128
+        ):
+            return self.flash_attention(q, k, v)
+        # Otherwise, fallback to normal attention
+        else:
+            return self.normal_attention(q, k, v)
+
+    def flash_attention(self, q: torch.Tensor, k: torch.Tensor, v: torch.Tensor):
+        """
+        #### Flash Attention
+        :param q: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        :param k: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        :param v: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        """
+
+        # Get batch size and number of elements along sequence axis (`width * height`)
+        batch_size, seq_len, _ = q.shape
+
+        # Stack `q`, `k`, `v` vectors for flash attention, to get a single tensor of
+        # shape `[batch_size, seq_len, 3, n_heads * d_head]`
+        qkv = torch.stack((q, k, v), dim=2)
+        # Split the heads
+        qkv = qkv.view(batch_size, seq_len, 3, self.n_heads, self.d_head)
+
+        # Flash attention works for head sizes `32`, `64` and `128`, so we have to pad the heads to
+        # fit this size.
+        if self.d_head <= 32:
+            pad = 32 - self.d_head
+        elif self.d_head <= 64:
+            pad = 64 - self.d_head
+        elif self.d_head <= 128:
+            pad = 128 - self.d_head
+        else:
+            raise ValueError(f"Head size ${self.d_head} too large for Flash Attention")
+
+        # Pad the heads
+        if pad:
+            qkv = torch.cat(
+                (qkv, qkv.new_zeros(batch_size, seq_len, 3, self.n_heads, pad)), dim=-1
+            )
+
+        # Compute attention
+        # $$\underset{seq}{softmax}\Bigg(\frac{Q K^\top}{\sqrt{d_{key}}}\Bigg)V$$
+        # This gives a tensor of shape `[batch_size, seq_len, n_heads, d_padded]`
+        # TODO here I add the dtype changing
+        out, _ = self.flash(qkv.type(torch.float16))
+        # Truncate the extra head size
+        out = out[:, :, :, : self.d_head].float()
+        # Reshape to `[batch_size, seq_len, n_heads * d_head]`
+        out = out.reshape(batch_size, seq_len, self.n_heads * self.d_head)
+
+        # Map to `[batch_size, height * width, d_model]` with a linear layer
+        return self.to_out(out)
+
+    def normal_attention(self, q: torch.Tensor, k: torch.Tensor, v: torch.Tensor):
+        """
+        #### Normal Attention
+
+        :param q: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        :param k: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        :param v: are the query vectors before splitting heads, of shape `[batch_size, seq, d_attn]`
+        """
+
+        # Split them to heads of shape `[batch_size, seq_len, n_heads, d_head]`
+        q = q.view(*q.shape[:2], self.n_heads, -1)  # [bs, 64, 20, 32]
+        k = k.view(*k.shape[:2], self.n_heads, -1)  # [bs, 1, 20, 32]
+        v = v.view(*v.shape[:2], self.n_heads, -1)
+
+        # Calculate attention $\frac{Q K^\top}{\sqrt{d_{key}}}$
+        attn = torch.einsum("bihd,bjhd->bhij", q, k) * self.scale
+
+        # Compute softmax
+        # $$\underset{seq}{softmax}\Bigg(\frac{Q K^\top}{\sqrt{d_{key}}}\Bigg)$$
+        if self.is_inplace:
+            half = attn.shape[0] // 2
+            attn[half:] = attn[half:].softmax(dim=-1)
+            attn[:half] = attn[:half].softmax(dim=-1)
+        else:
+            attn = attn.softmax(dim=-1)
+
+        # Compute attention output
+        # $$\underset{seq}{softmax}\Bigg(\frac{Q K^\top}{\sqrt{d_{key}}}\Bigg)V$$
+        # attn: [bs, 20, 64, 1]
+        # v: [bs, 1, 20, 32]
+        out = torch.einsum("bhij,bjhd->bihd", attn, v)
+        # Reshape to `[batch_size, height * width, n_heads * d_head]`
+        out = out.reshape(*out.shape[:2], -1)
+        # Map to `[batch_size, height * width, d_model]` with a linear layer
+        return self.to_out(out)
+
+
+# class CrossAttention(nn.Module):
+# def __init__(self, query_dim, context_dim=None, heads=8, dim_head=64, dropout=0.):
+#     super().__init__()
+#     inner_dim = dim_head * heads
+#     context_dim = default(context_dim, query_dim)
+
+#     self.scale = dim_head ** -0.5
+#     self.heads = heads
+
+#     self.to_q = nn.Linear(query_dim, inner_dim, bias=False)
+#     self.to_k = nn.Linear(context_dim, inner_dim, bias=False)
+#     self.to_v = nn.Linear(context_dim, inner_dim, bias=False)
+
+#     self.to_out = nn.Sequential(
+#         nn.Linear(inner_dim, query_dim),
+#         nn.Dropout(dropout)
+#     )
+
+# def forward(self, x, context=None, mask=None):
+#     h = self.heads
+
+#     q = self.to_q(x)
+#     context = default(context, x)
+#     k = self.to_k(context)
+#     v = self.to_v(context)
+
+#     q, k, v = map(lambda t: rearrange(t, 'b n (h d) -> (b h) n d', h=h), (q, k, v))
+
+#     sim = einsum('b i d, b j d -> b i j', q, k) * self.scale
+
+#     if exists(mask):
+#         mask = rearrange(mask, 'b ... -> b (...)')
+#         max_neg_value = -torch.finfo(sim.dtype).max
+#         mask = repeat(mask, 'b j -> (b h) () j', h=h)
+#         sim.masked_fill_(~mask, max_neg_value)
+
+#     # attention, what we cannot get enough of
+#     attn = sim.softmax(dim=-1)
+
+#     out = einsum('b i j, b j d -> b i d', attn, v)
+#     out = rearrange(out, '(b h) n d -> b n (h d)', h=h)
+#     return self.to_out(out)
+
+
+class BasicTransformerBlock(nn.Module):
+    def __init__(
+        self,
+        dim,
+        n_heads,
+        d_head,
+        dropout=0.0,
+        context_dim=None,
+        gated_ff=True,
+        checkpoint=True,
+    ):
+        super().__init__()
+        self.attn1 = CrossAttention(
+            query_dim=dim, heads=n_heads, dim_head=d_head, dropout=dropout
+        )  # is a self-attention
+        self.ff = FeedForward(dim, dropout=dropout, glu=gated_ff)
+        self.attn2 = CrossAttention(
+            query_dim=dim,
+            context_dim=context_dim,
+            heads=n_heads,
+            dim_head=d_head,
+            dropout=dropout,
+        )  # is self-attn if context is none
+        self.norm1 = nn.LayerNorm(dim)
+        self.norm2 = nn.LayerNorm(dim)
+        self.norm3 = nn.LayerNorm(dim)
+        self.checkpoint = checkpoint
+
+    def forward(self, x, context=None):
+        if context is None:
+            return checkpoint(self._forward, (x,), self.parameters(), self.checkpoint)
+        else:
+            return checkpoint(
+                self._forward, (x, context), self.parameters(), self.checkpoint
+            )
+
+    def _forward(self, x, context=None):
+        x = self.attn1(self.norm1(x)) + x
+        x = self.attn2(self.norm2(x), context=context) + x
+        x = self.ff(self.norm3(x)) + x
+        return x
+
+
+class SpatialTransformer(nn.Module):
+    """
+    Transformer block for image-like data.
+    First, project the input (aka embedding)
+    and reshape to b, t, d.
+    Then apply standard transformer action.
+    Finally, reshape to image
+    """
+
+    def __init__(
+        self,
+        in_channels,
+        n_heads,
+        d_head,
+        depth=1,
+        dropout=0.0,
+        context_dim=None,
+        no_context=False,
+    ):
+        super().__init__()
+
+        if no_context:
+            context_dim = None
+
+        self.in_channels = in_channels
+        inner_dim = n_heads * d_head
+        self.norm = Normalize(in_channels)
+
+        self.proj_in = nn.Conv2d(
+            in_channels, inner_dim, kernel_size=1, stride=1, padding=0
+        )
+
+        self.transformer_blocks = nn.ModuleList(
+            [
+                BasicTransformerBlock(
+                    inner_dim, n_heads, d_head, dropout=dropout, context_dim=context_dim
+                )
+                for d in range(depth)
+            ]
+        )
+
+        self.proj_out = zero_module(
+            nn.Conv2d(inner_dim, in_channels, kernel_size=1, stride=1, padding=0)
+        )
+
+    def forward(self, x, context=None):
+        # note: if no context is given, cross-attention defaults to self-attention
+        b, c, h, w = x.shape
+        x_in = x
+        x = self.norm(x)
+        x = self.proj_in(x)
+        x = rearrange(x, "b c h w -> b (h w) c")
+        for block in self.transformer_blocks:
+            x = block(x, context=context)
+        x = rearrange(x, "b (h w) c -> b c h w", h=h, w=w)
+        x = self.proj_out(x)
+        return x + x_in
diff --git a/picoaudio/audioldm/latent_diffusion/ddim.py b/picoaudio/audioldm/latent_diffusion/ddim.py
new file mode 100644
index 0000000000000000000000000000000000000000..732002b048e9a193313aa0ef9a353d4fc078be72
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/ddim.py
@@ -0,0 +1,377 @@
+"""SAMPLING ONLY."""
+
+import torch
+import numpy as np
+from tqdm import tqdm
+
+from audioldm.latent_diffusion.util import (
+    make_ddim_sampling_parameters,
+    make_ddim_timesteps,
+    noise_like,
+    extract_into_tensor,
+)
+
+
+class DDIMSampler(object):
+    def __init__(self, model, schedule="linear", **kwargs):
+        super().__init__()
+        self.model = model
+        self.ddpm_num_timesteps = model.num_timesteps
+        self.schedule = schedule
+
+    def register_buffer(self, name, attr):
+        if type(attr) == torch.Tensor:
+            if attr.device != torch.device("cuda"):
+                attr = attr.to(torch.device("cuda"))
+        setattr(self, name, attr)
+
+    def make_schedule(
+        self, ddim_num_steps, ddim_discretize="uniform", ddim_eta=0.0, verbose=True
+    ):
+        self.ddim_timesteps = make_ddim_timesteps(
+            ddim_discr_method=ddim_discretize,
+            num_ddim_timesteps=ddim_num_steps,
+            num_ddpm_timesteps=self.ddpm_num_timesteps,
+            verbose=verbose,
+        )
+        alphas_cumprod = self.model.alphas_cumprod
+        assert (
+            alphas_cumprod.shape[0] == self.ddpm_num_timesteps
+        ), "alphas have to be defined for each timestep"
+        to_torch = lambda x: x.clone().detach().to(torch.float32).to(self.model.device)
+
+        self.register_buffer("betas", to_torch(self.model.betas))
+        self.register_buffer("alphas_cumprod", to_torch(alphas_cumprod))
+        self.register_buffer(
+            "alphas_cumprod_prev", to_torch(self.model.alphas_cumprod_prev)
+        )
+
+        # calculations for diffusion q(x_t | x_{t-1}) and others
+        self.register_buffer(
+            "sqrt_alphas_cumprod", to_torch(np.sqrt(alphas_cumprod.cpu()))
+        )
+        self.register_buffer(
+            "sqrt_one_minus_alphas_cumprod",
+            to_torch(np.sqrt(1.0 - alphas_cumprod.cpu())),
+        )
+        self.register_buffer(
+            "log_one_minus_alphas_cumprod", to_torch(np.log(1.0 - alphas_cumprod.cpu()))
+        )
+        self.register_buffer(
+            "sqrt_recip_alphas_cumprod", to_torch(np.sqrt(1.0 / alphas_cumprod.cpu()))
+        )
+        self.register_buffer(
+            "sqrt_recipm1_alphas_cumprod",
+            to_torch(np.sqrt(1.0 / alphas_cumprod.cpu() - 1)),
+        )
+
+        # ddim sampling parameters
+        ddim_sigmas, ddim_alphas, ddim_alphas_prev = make_ddim_sampling_parameters(
+            alphacums=alphas_cumprod.cpu(),
+            ddim_timesteps=self.ddim_timesteps,
+            eta=ddim_eta,
+            verbose=verbose,
+        )
+        self.register_buffer("ddim_sigmas", ddim_sigmas)
+        self.register_buffer("ddim_alphas", ddim_alphas)
+        self.register_buffer("ddim_alphas_prev", ddim_alphas_prev)
+        self.register_buffer("ddim_sqrt_one_minus_alphas", np.sqrt(1.0 - ddim_alphas))
+        sigmas_for_original_sampling_steps = ddim_eta * torch.sqrt(
+            (1 - self.alphas_cumprod_prev)
+            / (1 - self.alphas_cumprod)
+            * (1 - self.alphas_cumprod / self.alphas_cumprod_prev)
+        )
+        self.register_buffer(
+            "ddim_sigmas_for_original_num_steps", sigmas_for_original_sampling_steps
+        )
+
+    @torch.no_grad()
+    def sample(
+        self,
+        S,
+        batch_size,
+        shape,
+        conditioning=None,
+        callback=None,
+        normals_sequence=None,
+        img_callback=None,
+        quantize_x0=False,
+        eta=0.0,
+        mask=None,
+        x0=None,
+        temperature=1.0,
+        noise_dropout=0.0,
+        score_corrector=None,
+        corrector_kwargs=None,
+        verbose=True,
+        x_T=None,
+        log_every_t=100,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+        # this has to come in the same format as the conditioning, # e.g. as encoded tokens, ...
+        **kwargs,
+    ):
+        if conditioning is not None:
+            if isinstance(conditioning, dict):
+                cbs = conditioning[list(conditioning.keys())[0]].shape[0]
+                if cbs != batch_size:
+                    print(
+                        f"Warning: Got {cbs} conditionings but batch-size is {batch_size}"
+                    )
+            else:
+                if conditioning.shape[0] != batch_size:
+                    print(
+                        f"Warning: Got {conditioning.shape[0]} conditionings but batch-size is {batch_size}"
+                    )
+
+        self.make_schedule(ddim_num_steps=S, ddim_eta=eta, verbose=verbose)
+        # sampling
+        C, H, W = shape
+        size = (batch_size, C, H, W)
+        samples, intermediates = self.ddim_sampling(
+            conditioning,
+            size,
+            callback=callback,
+            img_callback=img_callback,
+            quantize_denoised=quantize_x0,
+            mask=mask,
+            x0=x0,
+            ddim_use_original_steps=False,
+            noise_dropout=noise_dropout,
+            temperature=temperature,
+            score_corrector=score_corrector,
+            corrector_kwargs=corrector_kwargs,
+            x_T=x_T,
+            log_every_t=log_every_t,
+            unconditional_guidance_scale=unconditional_guidance_scale,
+            unconditional_conditioning=unconditional_conditioning,
+        )
+        return samples, intermediates
+
+    @torch.no_grad()
+    def ddim_sampling(
+        self,
+        cond,
+        shape,
+        x_T=None,
+        ddim_use_original_steps=False,
+        callback=None,
+        timesteps=None,
+        quantize_denoised=False,
+        mask=None,
+        x0=None,
+        img_callback=None,
+        log_every_t=100,
+        temperature=1.0,
+        noise_dropout=0.0,
+        score_corrector=None,
+        corrector_kwargs=None,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+    ):
+        device = self.model.betas.device
+        b = shape[0]
+        if x_T is None:
+            img = torch.randn(shape, device=device)
+        else:
+            img = x_T
+
+        if timesteps is None:
+            timesteps = (
+                self.ddpm_num_timesteps
+                if ddim_use_original_steps
+                else self.ddim_timesteps
+            )
+        elif timesteps is not None and not ddim_use_original_steps:
+            subset_end = (
+                int(
+                    min(timesteps / self.ddim_timesteps.shape[0], 1)
+                    * self.ddim_timesteps.shape[0]
+                )
+                - 1
+            )
+            timesteps = self.ddim_timesteps[:subset_end]
+
+        intermediates = {"x_inter": [img], "pred_x0": [img]}
+        time_range = (
+            reversed(range(0, timesteps))
+            if ddim_use_original_steps
+            else np.flip(timesteps)
+        )
+        total_steps = timesteps if ddim_use_original_steps else timesteps.shape[0]
+        # print(f"Running DDIM Sampling with {total_steps} timesteps")
+
+        # iterator = gr.Progress().tqdm(time_range, desc="DDIM Sampler", total=total_steps)
+        iterator = tqdm(time_range, desc="DDIM Sampler", total=total_steps, leave=False)
+
+        for i, step in enumerate(iterator):
+            index = total_steps - i - 1
+            ts = torch.full((b,), step, device=device, dtype=torch.long)
+            if mask is not None:
+                assert x0 is not None
+                img_orig = self.model.q_sample(
+                    x0, ts
+                )  # TODO deterministic forward pass?
+                img = (
+                    img_orig * mask + (1.0 - mask) * img
+                )  # In the first sampling step, img is pure gaussian noise
+
+            outs = self.p_sample_ddim(
+                img,
+                cond,
+                ts,
+                index=index,
+                use_original_steps=ddim_use_original_steps,
+                quantize_denoised=quantize_denoised,
+                temperature=temperature,
+                noise_dropout=noise_dropout,
+                score_corrector=score_corrector,
+                corrector_kwargs=corrector_kwargs,
+                unconditional_guidance_scale=unconditional_guidance_scale,
+                unconditional_conditioning=unconditional_conditioning,
+            )
+            img, pred_x0 = outs
+            if callback:
+                callback(i)
+            if img_callback:
+                img_callback(pred_x0, i)
+
+            if index % log_every_t == 0 or index == total_steps - 1:
+                intermediates["x_inter"].append(img)
+                intermediates["pred_x0"].append(pred_x0)
+
+        return img, intermediates
+
+    @torch.no_grad()
+    def stochastic_encode(self, x0, t, use_original_steps=False, noise=None):
+        # fast, but does not allow for exact reconstruction
+        # t serves as an index to gather the correct alphas
+        if use_original_steps:
+            sqrt_alphas_cumprod = self.sqrt_alphas_cumprod
+            sqrt_one_minus_alphas_cumprod = self.sqrt_one_minus_alphas_cumprod
+        else:
+            sqrt_alphas_cumprod = torch.sqrt(self.ddim_alphas)
+            sqrt_one_minus_alphas_cumprod = self.ddim_sqrt_one_minus_alphas
+
+        if noise is None:
+            noise = torch.randn_like(x0)
+
+        return (
+            extract_into_tensor(sqrt_alphas_cumprod, t, x0.shape) * x0
+            + extract_into_tensor(sqrt_one_minus_alphas_cumprod, t, x0.shape) * noise
+        )
+
+    @torch.no_grad()
+    def decode(
+        self,
+        x_latent,
+        cond,
+        t_start,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+        use_original_steps=False,
+    ):
+
+        timesteps = (
+            np.arange(self.ddpm_num_timesteps)
+            if use_original_steps
+            else self.ddim_timesteps
+        )
+        timesteps = timesteps[:t_start]
+
+        time_range = np.flip(timesteps)
+        total_steps = timesteps.shape[0]
+        # print(f"Running DDIM Sampling with {total_steps} timesteps")
+
+        # iterator = gr.Progress().tqdm(time_range, desc="Decoding image", total=total_steps)
+        iterator = tqdm(time_range, desc="Decoding image", total=total_steps)
+        x_dec = x_latent
+
+        for i, step in enumerate(iterator):
+            index = total_steps - i - 1
+            ts = torch.full(
+                (x_latent.shape[0],), step, device=x_latent.device, dtype=torch.long
+            )
+            x_dec, _ = self.p_sample_ddim(
+                x_dec,
+                cond,
+                ts,
+                index=index,
+                use_original_steps=use_original_steps,
+                unconditional_guidance_scale=unconditional_guidance_scale,
+                unconditional_conditioning=unconditional_conditioning,
+            )
+        return x_dec
+
+    @torch.no_grad()
+    def p_sample_ddim(
+        self,
+        x,
+        c,
+        t,
+        index,
+        repeat_noise=False,
+        use_original_steps=False,
+        quantize_denoised=False,
+        temperature=1.0,
+        noise_dropout=0.0,
+        score_corrector=None,
+        corrector_kwargs=None,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+    ):
+        b, *_, device = *x.shape, x.device
+
+        if unconditional_conditioning is None or unconditional_guidance_scale == 1.0:
+            e_t = self.model.apply_model(x, t, c)
+        else:
+            x_in = torch.cat([x] * 2)
+            t_in = torch.cat([t] * 2)
+            c_in = torch.cat([unconditional_conditioning, c])
+            e_t_uncond, e_t = self.model.apply_model(x_in, t_in, c_in).chunk(2)
+            # When unconditional_guidance_scale == 1: only e_t
+            # When unconditional_guidance_scale == 0: only unconditional
+            # When unconditional_guidance_scale > 1: add more unconditional guidance
+            e_t = e_t_uncond + unconditional_guidance_scale * (e_t - e_t_uncond)
+
+        if score_corrector is not None:
+            assert self.model.parameterization == "eps"
+            e_t = score_corrector.modify_score(
+                self.model, e_t, x, t, c, **corrector_kwargs
+            )
+
+        alphas = self.model.alphas_cumprod if use_original_steps else self.ddim_alphas
+        alphas_prev = (
+            self.model.alphas_cumprod_prev
+            if use_original_steps
+            else self.ddim_alphas_prev
+        )
+        sqrt_one_minus_alphas = (
+            self.model.sqrt_one_minus_alphas_cumprod
+            if use_original_steps
+            else self.ddim_sqrt_one_minus_alphas
+        )
+        sigmas = (
+            self.model.ddim_sigmas_for_original_num_steps
+            if use_original_steps
+            else self.ddim_sigmas
+        )
+        # select parameters corresponding to the currently considered timestep
+        a_t = torch.full((b, 1, 1, 1), alphas[index], device=device)
+        a_prev = torch.full((b, 1, 1, 1), alphas_prev[index], device=device)
+        sigma_t = torch.full((b, 1, 1, 1), sigmas[index], device=device)
+        sqrt_one_minus_at = torch.full(
+            (b, 1, 1, 1), sqrt_one_minus_alphas[index], device=device
+        )
+
+        # current prediction for x_0
+        pred_x0 = (x - sqrt_one_minus_at * e_t) / a_t.sqrt()
+        if quantize_denoised:
+            pred_x0, _, *_ = self.model.first_stage_model.quantize(pred_x0)
+        # direction pointing to x_t
+        dir_xt = (1.0 - a_prev - sigma_t**2).sqrt() * e_t
+        noise = sigma_t * noise_like(x.shape, device, repeat_noise) * temperature
+        if noise_dropout > 0.0:
+            noise = torch.nn.functional.dropout(noise, p=noise_dropout)
+        x_prev = a_prev.sqrt() * pred_x0 + dir_xt + noise  # TODO
+        return x_prev, pred_x0
diff --git a/picoaudio/audioldm/latent_diffusion/ddpm.py b/picoaudio/audioldm/latent_diffusion/ddpm.py
new file mode 100644
index 0000000000000000000000000000000000000000..ffca031c27d413698adee5a58547b7d0ea4069c3
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/ddpm.py
@@ -0,0 +1,441 @@
+"""
+wild mixture of
+https://github.com/lucidrains/denoising-diffusion-pytorch/blob/7706bdfc6f527f58d33f84b7b522e61e6e3164b3/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py
+https://github.com/openai/improved-diffusion/blob/e94489283bb876ac1477d5dd7709bbbd2d9902ce/improved_diffusion/gaussian_diffusion.py
+https://github.com/CompVis/taming-transformers
+-- merci
+"""
+import sys
+import os
+
+import torch
+import torch.nn as nn
+import numpy as np
+from contextlib import contextmanager
+from functools import partial
+from tqdm import tqdm
+
+from audioldm.utils import exists, default, count_params, instantiate_from_config
+from audioldm.latent_diffusion.ema import LitEma
+from audioldm.latent_diffusion.util import (
+    make_beta_schedule,
+    extract_into_tensor,
+    noise_like,
+)
+import soundfile as sf
+import os
+
+
+__conditioning_keys__ = {"concat": "c_concat", "crossattn": "c_crossattn", "adm": "y"}
+
+
+def disabled_train(self, mode=True):
+    """Overwrite model.train with this function to make sure train/eval mode
+    does not change anymore."""
+    return self
+
+
+def uniform_on_device(r1, r2, shape, device):
+    return (r1 - r2) * torch.rand(*shape, device=device) + r2
+
+
+class DiffusionWrapper(nn.Module):
+    def __init__(self, diff_model_config, conditioning_key):
+        super().__init__()
+        self.diffusion_model = instantiate_from_config(diff_model_config)
+        self.conditioning_key = conditioning_key
+        assert self.conditioning_key in [
+            None,
+            "concat",
+            "crossattn",
+            "hybrid",
+            "adm",
+            "film",
+        ]
+
+    def forward(
+        self, x, t, c_concat: list = None, c_crossattn: list = None, c_film: list = None
+    ):
+        x = x.contiguous()
+        t = t.contiguous()
+
+        if self.conditioning_key is None:
+            out = self.diffusion_model(x, t)
+        elif self.conditioning_key == "concat":
+            xc = torch.cat([x] + c_concat, dim=1)
+            out = self.diffusion_model(xc, t)
+        elif self.conditioning_key == "crossattn":
+            cc = torch.cat(c_crossattn, 1)
+            out = self.diffusion_model(x, t, context=cc)
+        elif self.conditioning_key == "hybrid":
+            xc = torch.cat([x] + c_concat, dim=1)
+            cc = torch.cat(c_crossattn, 1)
+            out = self.diffusion_model(xc, t, context=cc)
+        elif (
+            self.conditioning_key == "film"
+        ):  # The condition is assumed to be a global token, which wil pass through a linear layer and added with the time embedding for the FILM
+            cc = c_film[0].squeeze(1)  # only has one token
+            out = self.diffusion_model(x, t, y=cc)
+        elif self.conditioning_key == "adm":
+            cc = c_crossattn[0]
+            out = self.diffusion_model(x, t, y=cc)
+        else:
+            raise NotImplementedError()
+
+        return out
+
+
+class DDPM(nn.Module):
+    # classic DDPM with Gaussian diffusion, in image space
+    def __init__(
+        self,
+        unet_config,
+        timesteps=1000,
+        beta_schedule="linear",
+        loss_type="l2",
+        ckpt_path=None,
+        ignore_keys=[],
+        load_only_unet=False,
+        monitor="val/loss",
+        use_ema=True,
+        first_stage_key="image",
+        latent_t_size=256,
+        latent_f_size=16,
+        channels=3,
+        log_every_t=100,
+        clip_denoised=True,
+        linear_start=1e-4,
+        linear_end=2e-2,
+        cosine_s=8e-3,
+        given_betas=None,
+        original_elbo_weight=0.0,
+        v_posterior=0.0,  # weight for choosing posterior variance as sigma = (1-v) * beta_tilde + v * beta
+        l_simple_weight=1.0,
+        conditioning_key=None,
+        parameterization="eps",  # all assuming fixed variance schedules
+        scheduler_config=None,
+        use_positional_encodings=False,
+        learn_logvar=False,
+        logvar_init=0.0,
+    ):
+        super().__init__()
+        assert parameterization in [
+            "eps",
+            "x0",
+        ], 'currently only supporting "eps" and "x0"'
+        self.parameterization = parameterization
+        self.state = None
+        # print(f"{self.__class__.__name__}: Running in {self.parameterization}-prediction mode")
+        self.cond_stage_model = None
+        self.clip_denoised = clip_denoised
+        self.log_every_t = log_every_t
+        self.first_stage_key = first_stage_key
+
+        self.latent_t_size = latent_t_size
+        self.latent_f_size = latent_f_size
+
+        self.channels = channels
+        self.use_positional_encodings = use_positional_encodings
+        self.model = DiffusionWrapper(unet_config, conditioning_key)
+        count_params(self.model, verbose=True)
+        self.use_ema = use_ema
+        if self.use_ema:
+            self.model_ema = LitEma(self.model)
+            # print(f"Keeping EMAs of {len(list(self.model_ema.buffers()))}.")
+
+        self.use_scheduler = scheduler_config is not None
+        if self.use_scheduler:
+            self.scheduler_config = scheduler_config
+
+        self.v_posterior = v_posterior
+        self.original_elbo_weight = original_elbo_weight
+        self.l_simple_weight = l_simple_weight
+
+        if monitor is not None:
+            self.monitor = monitor
+
+        self.register_schedule(
+            given_betas=given_betas,
+            beta_schedule=beta_schedule,
+            timesteps=timesteps,
+            linear_start=linear_start,
+            linear_end=linear_end,
+            cosine_s=cosine_s,
+        )
+
+        self.loss_type = loss_type
+
+        self.learn_logvar = learn_logvar
+        self.logvar = torch.full(fill_value=logvar_init, size=(self.num_timesteps,))
+        if self.learn_logvar:
+            self.logvar = nn.Parameter(self.logvar, requires_grad=True)
+        else:
+            self.logvar = nn.Parameter(self.logvar, requires_grad=False)
+
+        self.logger_save_dir = None
+        self.logger_project = None
+        self.logger_version = None
+        self.label_indices_total = None
+        # To avoid the system cannot find metric value for checkpoint
+        self.metrics_buffer = {
+            "val/kullback_leibler_divergence_sigmoid": 15.0,
+            "val/kullback_leibler_divergence_softmax": 10.0,
+            "val/psnr": 0.0,
+            "val/ssim": 0.0,
+            "val/inception_score_mean": 1.0,
+            "val/inception_score_std": 0.0,
+            "val/kernel_inception_distance_mean": 0.0,
+            "val/kernel_inception_distance_std": 0.0,
+            "val/frechet_inception_distance": 133.0,
+            "val/frechet_audio_distance": 32.0,
+        }
+        self.initial_learning_rate = None
+
+    def get_log_dir(self):
+        if (
+            self.logger_save_dir is None
+            and self.logger_project is None
+            and self.logger_version is None
+        ):
+            return os.path.join(
+                self.logger.save_dir, self.logger._project, self.logger.version
+            )
+        else:
+            return os.path.join(
+                self.logger_save_dir, self.logger_project, self.logger_version
+            )
+
+    def set_log_dir(self, save_dir, project, version):
+        self.logger_save_dir = save_dir
+        self.logger_project = project
+        self.logger_version = version
+
+    def register_schedule(
+        self,
+        given_betas=None,
+        beta_schedule="linear",
+        timesteps=1000,
+        linear_start=1e-4,
+        linear_end=2e-2,
+        cosine_s=8e-3,
+    ):
+        if exists(given_betas):
+            betas = given_betas
+        else:
+            betas = make_beta_schedule(
+                beta_schedule,
+                timesteps,
+                linear_start=linear_start,
+                linear_end=linear_end,
+                cosine_s=cosine_s,
+            )
+        alphas = 1.0 - betas
+        alphas_cumprod = np.cumprod(alphas, axis=0)
+        alphas_cumprod_prev = np.append(1.0, alphas_cumprod[:-1])
+
+        (timesteps,) = betas.shape
+        self.num_timesteps = int(timesteps)
+        self.linear_start = linear_start
+        self.linear_end = linear_end
+        assert (
+            alphas_cumprod.shape[0] == self.num_timesteps
+        ), "alphas have to be defined for each timestep"
+
+        to_torch = partial(torch.tensor, dtype=torch.float32)
+
+        self.register_buffer("betas", to_torch(betas))
+        self.register_buffer("alphas_cumprod", to_torch(alphas_cumprod))
+        self.register_buffer("alphas_cumprod_prev", to_torch(alphas_cumprod_prev))
+
+        # calculations for diffusion q(x_t | x_{t-1}) and others
+        self.register_buffer("sqrt_alphas_cumprod", to_torch(np.sqrt(alphas_cumprod)))
+        self.register_buffer(
+            "sqrt_one_minus_alphas_cumprod", to_torch(np.sqrt(1.0 - alphas_cumprod))
+        )
+        self.register_buffer(
+            "log_one_minus_alphas_cumprod", to_torch(np.log(1.0 - alphas_cumprod))
+        )
+        self.register_buffer(
+            "sqrt_recip_alphas_cumprod", to_torch(np.sqrt(1.0 / alphas_cumprod))
+        )
+        self.register_buffer(
+            "sqrt_recipm1_alphas_cumprod", to_torch(np.sqrt(1.0 / alphas_cumprod - 1))
+        )
+
+        # calculations for posterior q(x_{t-1} | x_t, x_0)
+        posterior_variance = (1 - self.v_posterior) * betas * (
+            1.0 - alphas_cumprod_prev
+        ) / (1.0 - alphas_cumprod) + self.v_posterior * betas
+        # above: equal to 1. / (1. / (1. - alpha_cumprod_tm1) + alpha_t / beta_t)
+        self.register_buffer("posterior_variance", to_torch(posterior_variance))
+        # below: log calculation clipped because the posterior variance is 0 at the beginning of the diffusion chain
+        self.register_buffer(
+            "posterior_log_variance_clipped",
+            to_torch(np.log(np.maximum(posterior_variance, 1e-20))),
+        )
+        self.register_buffer(
+            "posterior_mean_coef1",
+            to_torch(betas * np.sqrt(alphas_cumprod_prev) / (1.0 - alphas_cumprod)),
+        )
+        self.register_buffer(
+            "posterior_mean_coef2",
+            to_torch(
+                (1.0 - alphas_cumprod_prev) * np.sqrt(alphas) / (1.0 - alphas_cumprod)
+            ),
+        )
+
+        if self.parameterization == "eps":
+            lvlb_weights = self.betas**2 / (
+                2
+                * self.posterior_variance
+                * to_torch(alphas)
+                * (1 - self.alphas_cumprod)
+            )
+        elif self.parameterization == "x0":
+            lvlb_weights = (
+                0.5
+                * np.sqrt(torch.Tensor(alphas_cumprod))
+                / (2.0 * 1 - torch.Tensor(alphas_cumprod))
+            )
+        else:
+            raise NotImplementedError("mu not supported")
+        # TODO how to choose this term
+        lvlb_weights[0] = lvlb_weights[1]
+        self.register_buffer("lvlb_weights", lvlb_weights, persistent=False)
+        assert not torch.isnan(self.lvlb_weights).all()
+
+    @contextmanager
+    def ema_scope(self, context=None):
+        if self.use_ema:
+            self.model_ema.store(self.model.parameters())
+            self.model_ema.copy_to(self.model)
+            if context is not None:
+                # print(f"{context}: Switched to EMA weights")
+                pass
+        try:
+            yield None
+        finally:
+            if self.use_ema:
+                self.model_ema.restore(self.model.parameters())
+                if context is not None:
+                    # print(f"{context}: Restored training weights")
+                    pass
+
+    def q_mean_variance(self, x_start, t):
+        """
+        Get the distribution q(x_t | x_0).
+        :param x_start: the [N x C x ...] tensor of noiseless inputs.
+        :param t: the number of diffusion steps (minus 1). Here, 0 means one step.
+        :return: A tuple (mean, variance, log_variance), all of x_start's shape.
+        """
+        mean = extract_into_tensor(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start
+        variance = extract_into_tensor(1.0 - self.alphas_cumprod, t, x_start.shape)
+        log_variance = extract_into_tensor(
+            self.log_one_minus_alphas_cumprod, t, x_start.shape
+        )
+        return mean, variance, log_variance
+
+    def predict_start_from_noise(self, x_t, t, noise):
+        return (
+            extract_into_tensor(self.sqrt_recip_alphas_cumprod, t, x_t.shape) * x_t
+            - extract_into_tensor(self.sqrt_recipm1_alphas_cumprod, t, x_t.shape)
+            * noise
+        )
+
+    def q_posterior(self, x_start, x_t, t):
+        posterior_mean = (
+            extract_into_tensor(self.posterior_mean_coef1, t, x_t.shape) * x_start
+            + extract_into_tensor(self.posterior_mean_coef2, t, x_t.shape) * x_t
+        )
+        posterior_variance = extract_into_tensor(self.posterior_variance, t, x_t.shape)
+        posterior_log_variance_clipped = extract_into_tensor(
+            self.posterior_log_variance_clipped, t, x_t.shape
+        )
+        return posterior_mean, posterior_variance, posterior_log_variance_clipped
+
+    def p_mean_variance(self, x, t, clip_denoised: bool):
+        model_out = self.model(x, t)
+        if self.parameterization == "eps":
+            x_recon = self.predict_start_from_noise(x, t=t, noise=model_out)
+        elif self.parameterization == "x0":
+            x_recon = model_out
+        if clip_denoised:
+            x_recon.clamp_(-1.0, 1.0)
+
+        model_mean, posterior_variance, posterior_log_variance = self.q_posterior(
+            x_start=x_recon, x_t=x, t=t
+        )
+        return model_mean, posterior_variance, posterior_log_variance
+
+    @torch.no_grad()
+    def p_sample(self, x, t, clip_denoised=True, repeat_noise=False):
+        b, *_, device = *x.shape, x.device
+        model_mean, _, model_log_variance = self.p_mean_variance(
+            x=x, t=t, clip_denoised=clip_denoised
+        )
+        noise = noise_like(x.shape, device, repeat_noise)
+        # no noise when t == 0
+        nonzero_mask = (
+            (1 - (t == 0).float()).reshape(b, *((1,) * (len(x.shape) - 1))).contiguous()
+        )
+        return model_mean + nonzero_mask * (0.5 * model_log_variance).exp() * noise
+
+    @torch.no_grad()
+    def p_sample_loop(self, shape, return_intermediates=False):
+        device = self.betas.device
+        b = shape[0]
+        img = torch.randn(shape, device=device)
+        intermediates = [img]
+        for i in tqdm(
+            reversed(range(0, self.num_timesteps)),
+            desc="Sampling t",
+            total=self.num_timesteps,
+        ):
+            img = self.p_sample(
+                img,
+                torch.full((b,), i, device=device, dtype=torch.long),
+                clip_denoised=self.clip_denoised,
+            )
+            if i % self.log_every_t == 0 or i == self.num_timesteps - 1:
+                intermediates.append(img)
+        if return_intermediates:
+            return img, intermediates
+        return img
+
+    @torch.no_grad()
+    def sample(self, batch_size=16, return_intermediates=False):
+        shape = (batch_size, channels, self.latent_t_size, self.latent_f_size)
+        channels = self.channels
+        return self.p_sample_loop(shape, return_intermediates=return_intermediates)
+
+    def q_sample(self, x_start, t, noise=None):
+        noise = default(noise, lambda: torch.randn_like(x_start))
+        return (
+            extract_into_tensor(self.sqrt_alphas_cumprod, t, x_start.shape) * x_start
+            + extract_into_tensor(self.sqrt_one_minus_alphas_cumprod, t, x_start.shape)
+            * noise
+        )
+
+    def forward(self, x, *args, **kwargs):
+        t = torch.randint(
+            0, self.num_timesteps, (x.shape[0],), device=self.device
+        ).long()
+        return self.p_losses(x, t, *args, **kwargs)
+
+    def get_input(self, batch, k):
+        # fbank, log_magnitudes_stft, label_indices, fname, waveform, clip_label, text = batch
+        fbank, log_magnitudes_stft, label_indices, fname, waveform, text = batch
+        ret = {}
+
+        ret["fbank"] = (
+            fbank.unsqueeze(1).to(memory_format=torch.contiguous_format).float()
+        )
+        ret["stft"] = log_magnitudes_stft.to(
+            memory_format=torch.contiguous_format
+        ).float()
+        # ret["clip_label"] = clip_label.to(memory_format=torch.contiguous_format).float()
+        ret["waveform"] = waveform.to(memory_format=torch.contiguous_format).float()
+        ret["text"] = list(text)
+        ret["fname"] = fname
+
+        return ret[k]
diff --git a/picoaudio/audioldm/latent_diffusion/ema.py b/picoaudio/audioldm/latent_diffusion/ema.py
new file mode 100644
index 0000000000000000000000000000000000000000..880ca3d205d9b4d7450e146930a93f2e63c58b70
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/ema.py
@@ -0,0 +1,82 @@
+import torch
+from torch import nn
+
+
+class LitEma(nn.Module):
+    def __init__(self, model, decay=0.9999, use_num_upates=True):
+        super().__init__()
+        if decay < 0.0 or decay > 1.0:
+            raise ValueError("Decay must be between 0 and 1")
+
+        self.m_name2s_name = {}
+        self.register_buffer("decay", torch.tensor(decay, dtype=torch.float32))
+        self.register_buffer(
+            "num_updates",
+            torch.tensor(0, dtype=torch.int)
+            if use_num_upates
+            else torch.tensor(-1, dtype=torch.int),
+        )
+
+        for name, p in model.named_parameters():
+            if p.requires_grad:
+                # remove as '.'-character is not allowed in buffers
+                s_name = name.replace(".", "")
+                self.m_name2s_name.update({name: s_name})
+                self.register_buffer(s_name, p.clone().detach().data)
+
+        self.collected_params = []
+
+    def forward(self, model):
+        decay = self.decay
+
+        if self.num_updates >= 0:
+            self.num_updates += 1
+            decay = min(self.decay, (1 + self.num_updates) / (10 + self.num_updates))
+
+        one_minus_decay = 1.0 - decay
+
+        with torch.no_grad():
+            m_param = dict(model.named_parameters())
+            shadow_params = dict(self.named_buffers())
+
+            for key in m_param:
+                if m_param[key].requires_grad:
+                    sname = self.m_name2s_name[key]
+                    shadow_params[sname] = shadow_params[sname].type_as(m_param[key])
+                    shadow_params[sname].sub_(
+                        one_minus_decay * (shadow_params[sname] - m_param[key])
+                    )
+                else:
+                    assert not key in self.m_name2s_name
+
+    def copy_to(self, model):
+        m_param = dict(model.named_parameters())
+        shadow_params = dict(self.named_buffers())
+        for key in m_param:
+            if m_param[key].requires_grad:
+                m_param[key].data.copy_(shadow_params[self.m_name2s_name[key]].data)
+            else:
+                assert not key in self.m_name2s_name
+
+    def store(self, parameters):
+        """
+        Save the current parameters for restoring later.
+        Args:
+          parameters: Iterable of `torch.nn.Parameter`; the parameters to be
+            temporarily stored.
+        """
+        self.collected_params = [param.clone() for param in parameters]
+
+    def restore(self, parameters):
+        """
+        Restore the parameters stored with the `store` method.
+        Useful to validate the model with EMA parameters without affecting the
+        original optimization process. Store the parameters before the
+        `copy_to` method. After validation (or model saving), use this to
+        restore the former parameters.
+        Args:
+          parameters: Iterable of `torch.nn.Parameter`; the parameters to be
+            updated with the stored parameters.
+        """
+        for c_param, param in zip(self.collected_params, parameters):
+            param.data.copy_(c_param.data)
diff --git a/picoaudio/audioldm/latent_diffusion/openaimodel.py b/picoaudio/audioldm/latent_diffusion/openaimodel.py
new file mode 100644
index 0000000000000000000000000000000000000000..831d7aafb36bba16888e4389153979a6c13639f5
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/openaimodel.py
@@ -0,0 +1,1069 @@
+from abc import abstractmethod
+import math
+
+import numpy as np
+import torch as th
+import torch.nn as nn
+import torch.nn.functional as F
+
+from audioldm.latent_diffusion.util import (
+    checkpoint,
+    conv_nd,
+    linear,
+    avg_pool_nd,
+    zero_module,
+    normalization,
+    timestep_embedding,
+)
+from audioldm.latent_diffusion.attention import SpatialTransformer
+
+
+# dummy replace
+def convert_module_to_f16(x):
+    pass
+
+
+def convert_module_to_f32(x):
+    pass
+
+
+## go
+class AttentionPool2d(nn.Module):
+    """
+    Adapted from CLIP: https://github.com/openai/CLIP/blob/main/clip/model.py
+    """
+
+    def __init__(
+        self,
+        spacial_dim: int,
+        embed_dim: int,
+        num_heads_channels: int,
+        output_dim: int = None,
+    ):
+        super().__init__()
+        self.positional_embedding = nn.Parameter(
+            th.randn(embed_dim, spacial_dim**2 + 1) / embed_dim**0.5
+        )
+        self.qkv_proj = conv_nd(1, embed_dim, 3 * embed_dim, 1)
+        self.c_proj = conv_nd(1, embed_dim, output_dim or embed_dim, 1)
+        self.num_heads = embed_dim // num_heads_channels
+        self.attention = QKVAttention(self.num_heads)
+
+    def forward(self, x):
+        b, c, *_spatial = x.shape
+        x = x.reshape(b, c, -1).contiguous()  # NC(HW)
+        x = th.cat([x.mean(dim=-1, keepdim=True), x], dim=-1)  # NC(HW+1)
+        x = x + self.positional_embedding[None, :, :].to(x.dtype)  # NC(HW+1)
+        x = self.qkv_proj(x)
+        x = self.attention(x)
+        x = self.c_proj(x)
+        return x[:, :, 0]
+
+
+class TimestepBlock(nn.Module):
+    """
+    Any module where forward() takes timestep embeddings as a second argument.
+    """
+
+    @abstractmethod
+    def forward(self, x, emb):
+        """
+        Apply the module to `x` given `emb` timestep embeddings.
+        """
+
+
+class TimestepEmbedSequential(nn.Sequential, TimestepBlock):
+    """
+    A sequential module that passes timestep embeddings to the children that
+    support it as an extra input.
+    """
+
+    def forward(self, x, emb, context=None):
+        for layer in self:
+            if isinstance(layer, TimestepBlock):
+                x = layer(x, emb)
+            elif isinstance(layer, SpatialTransformer):
+                x = layer(x, context)
+            else:
+                x = layer(x)
+        return x
+
+
+class Upsample(nn.Module):
+    """
+    An upsampling layer with an optional convolution.
+    :param channels: channels in the inputs and outputs.
+    :param use_conv: a bool determining if a convolution is applied.
+    :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
+                 upsampling occurs in the inner-two dimensions.
+    """
+
+    def __init__(self, channels, use_conv, dims=2, out_channels=None, padding=1):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.dims = dims
+        if use_conv:
+            self.conv = conv_nd(
+                dims, self.channels, self.out_channels, 3, padding=padding
+            )
+
+    def forward(self, x):
+        assert x.shape[1] == self.channels
+        if self.dims == 3:
+            x = F.interpolate(
+                x, (x.shape[2], x.shape[3] * 2, x.shape[4] * 2), mode="nearest"
+            )
+        else:
+            x = F.interpolate(x, scale_factor=2, mode="nearest")
+        if self.use_conv:
+            x = self.conv(x)
+        return x
+
+
+class TransposedUpsample(nn.Module):
+    "Learned 2x upsampling without padding"
+
+    def __init__(self, channels, out_channels=None, ks=5):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+
+        self.up = nn.ConvTranspose2d(
+            self.channels, self.out_channels, kernel_size=ks, stride=2
+        )
+
+    def forward(self, x):
+        return self.up(x)
+
+
+class Downsample(nn.Module):
+    """
+    A downsampling layer with an optional convolution.
+    :param channels: channels in the inputs and outputs.
+    :param use_conv: a bool determining if a convolution is applied.
+    :param dims: determines if the signal is 1D, 2D, or 3D. If 3D, then
+                 downsampling occurs in the inner-two dimensions.
+    """
+
+    def __init__(self, channels, use_conv, dims=2, out_channels=None, padding=1):
+        super().__init__()
+        self.channels = channels
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.dims = dims
+        stride = 2 if dims != 3 else (1, 2, 2)
+        if use_conv:
+            self.op = conv_nd(
+                dims,
+                self.channels,
+                self.out_channels,
+                3,
+                stride=stride,
+                padding=padding,
+            )
+        else:
+            assert self.channels == self.out_channels
+            self.op = avg_pool_nd(dims, kernel_size=stride, stride=stride)
+
+    def forward(self, x):
+        assert x.shape[1] == self.channels
+        return self.op(x)
+
+
+class ResBlock(TimestepBlock):
+    """
+    A residual block that can optionally change the number of channels.
+    :param channels: the number of input channels.
+    :param emb_channels: the number of timestep embedding channels.
+    :param dropout: the rate of dropout.
+    :param out_channels: if specified, the number of out channels.
+    :param use_conv: if True and out_channels is specified, use a spatial
+        convolution instead of a smaller 1x1 convolution to change the
+        channels in the skip connection.
+    :param dims: determines if the signal is 1D, 2D, or 3D.
+    :param use_checkpoint: if True, use gradient checkpointing on this module.
+    :param up: if True, use this block for upsampling.
+    :param down: if True, use this block for downsampling.
+    """
+
+    def __init__(
+        self,
+        channels,
+        emb_channels,
+        dropout,
+        out_channels=None,
+        use_conv=False,
+        use_scale_shift_norm=False,
+        dims=2,
+        use_checkpoint=False,
+        up=False,
+        down=False,
+    ):
+        super().__init__()
+        self.channels = channels
+        self.emb_channels = emb_channels
+        self.dropout = dropout
+        self.out_channels = out_channels or channels
+        self.use_conv = use_conv
+        self.use_checkpoint = use_checkpoint
+        self.use_scale_shift_norm = use_scale_shift_norm
+
+        self.in_layers = nn.Sequential(
+            normalization(channels),
+            nn.SiLU(),
+            conv_nd(dims, channels, self.out_channels, 3, padding=1),
+        )
+
+        self.updown = up or down
+
+        if up:
+            self.h_upd = Upsample(channels, False, dims)
+            self.x_upd = Upsample(channels, False, dims)
+        elif down:
+            self.h_upd = Downsample(channels, False, dims)
+            self.x_upd = Downsample(channels, False, dims)
+        else:
+            self.h_upd = self.x_upd = nn.Identity()
+
+        self.emb_layers = nn.Sequential(
+            nn.SiLU(),
+            linear(
+                emb_channels,
+                2 * self.out_channels if use_scale_shift_norm else self.out_channels,
+            ),
+        )
+        self.out_layers = nn.Sequential(
+            normalization(self.out_channels),
+            nn.SiLU(),
+            nn.Dropout(p=dropout),
+            zero_module(
+                conv_nd(dims, self.out_channels, self.out_channels, 3, padding=1)
+            ),
+        )
+
+        if self.out_channels == channels:
+            self.skip_connection = nn.Identity()
+        elif use_conv:
+            self.skip_connection = conv_nd(
+                dims, channels, self.out_channels, 3, padding=1
+            )
+        else:
+            self.skip_connection = conv_nd(dims, channels, self.out_channels, 1)
+
+    def forward(self, x, emb):
+        """
+        Apply the block to a Tensor, conditioned on a timestep embedding.
+        :param x: an [N x C x ...] Tensor of features.
+        :param emb: an [N x emb_channels] Tensor of timestep embeddings.
+        :return: an [N x C x ...] Tensor of outputs.
+        """
+        return checkpoint(
+            self._forward, (x, emb), self.parameters(), self.use_checkpoint
+        )
+
+    def _forward(self, x, emb):
+        if self.updown:
+            in_rest, in_conv = self.in_layers[:-1], self.in_layers[-1]
+            h = in_rest(x)
+            h = self.h_upd(h)
+            x = self.x_upd(x)
+            h = in_conv(h)
+        else:
+            h = self.in_layers(x)
+        emb_out = self.emb_layers(emb).type(h.dtype)
+        while len(emb_out.shape) < len(h.shape):
+            emb_out = emb_out[..., None]
+        if self.use_scale_shift_norm:
+            out_norm, out_rest = self.out_layers[0], self.out_layers[1:]
+            scale, shift = th.chunk(emb_out, 2, dim=1)
+            h = out_norm(h) * (1 + scale) + shift
+            h = out_rest(h)
+        else:
+            h = h + emb_out
+            h = self.out_layers(h)
+        return self.skip_connection(x) + h
+
+
+class AttentionBlock(nn.Module):
+    """
+    An attention block that allows spatial positions to attend to each other.
+    Originally ported from here, but adapted to the N-d case.
+    https://github.com/hojonathanho/diffusion/blob/1e0dceb3b3495bbe19116a5e1b3596cd0706c543/diffusion_tf/models/unet.py#L66.
+    """
+
+    def __init__(
+        self,
+        channels,
+        num_heads=1,
+        num_head_channels=-1,
+        use_checkpoint=False,
+        use_new_attention_order=False,
+    ):
+        super().__init__()
+        self.channels = channels
+        if num_head_channels == -1:
+            self.num_heads = num_heads
+        else:
+            assert (
+                channels % num_head_channels == 0
+            ), f"q,k,v channels {channels} is not divisible by num_head_channels {num_head_channels}"
+            self.num_heads = channels // num_head_channels
+        self.use_checkpoint = use_checkpoint
+        self.norm = normalization(channels)
+        self.qkv = conv_nd(1, channels, channels * 3, 1)
+        if use_new_attention_order:
+            # split qkv before split heads
+            self.attention = QKVAttention(self.num_heads)
+        else:
+            # split heads before split qkv
+            self.attention = QKVAttentionLegacy(self.num_heads)
+
+        self.proj_out = zero_module(conv_nd(1, channels, channels, 1))
+
+    def forward(self, x):
+        return checkpoint(
+            self._forward, (x,), self.parameters(), True
+        )  # TODO: check checkpoint usage, is True # TODO: fix the .half call!!!
+        # return pt_checkpoint(self._forward, x)  # pytorch
+
+    def _forward(self, x):
+        b, c, *spatial = x.shape
+        x = x.reshape(b, c, -1).contiguous()
+        qkv = self.qkv(self.norm(x)).contiguous()
+        h = self.attention(qkv).contiguous()
+        h = self.proj_out(h).contiguous()
+        return (x + h).reshape(b, c, *spatial).contiguous()
+
+
+def count_flops_attn(model, _x, y):
+    """
+    A counter for the `thop` package to count the operations in an
+    attention operation.
+    Meant to be used like:
+        macs, params = thop.profile(
+            model,
+            inputs=(inputs, timestamps),
+            custom_ops={QKVAttention: QKVAttention.count_flops},
+        )
+    """
+    b, c, *spatial = y[0].shape
+    num_spatial = int(np.prod(spatial))
+    # We perform two matmuls with the same number of ops.
+    # The first computes the weight matrix, the second computes
+    # the combination of the value vectors.
+    matmul_ops = 2 * b * (num_spatial**2) * c
+    model.total_ops += th.DoubleTensor([matmul_ops])
+
+
+class QKVAttentionLegacy(nn.Module):
+    """
+    A module which performs QKV attention. Matches legacy QKVAttention + input/ouput heads shaping
+    """
+
+    def __init__(self, n_heads):
+        super().__init__()
+        self.n_heads = n_heads
+
+    def forward(self, qkv):
+        """
+        Apply QKV attention.
+        :param qkv: an [N x (H * 3 * C) x T] tensor of Qs, Ks, and Vs.
+        :return: an [N x (H * C) x T] tensor after attention.
+        """
+        bs, width, length = qkv.shape
+        assert width % (3 * self.n_heads) == 0
+        ch = width // (3 * self.n_heads)
+        q, k, v = (
+            qkv.reshape(bs * self.n_heads, ch * 3, length).contiguous().split(ch, dim=1)
+        )
+        scale = 1 / math.sqrt(math.sqrt(ch))
+        weight = th.einsum(
+            "bct,bcs->bts", q * scale, k * scale
+        )  # More stable with f16 than dividing afterwards
+        weight = th.softmax(weight.float(), dim=-1).type(weight.dtype)
+        a = th.einsum("bts,bcs->bct", weight, v)
+        return a.reshape(bs, -1, length).contiguous()
+
+    @staticmethod
+    def count_flops(model, _x, y):
+        return count_flops_attn(model, _x, y)
+
+
+class QKVAttention(nn.Module):
+    """
+    A module which performs QKV attention and splits in a different order.
+    """
+
+    def __init__(self, n_heads):
+        super().__init__()
+        self.n_heads = n_heads
+
+    def forward(self, qkv):
+        """
+        Apply QKV attention.
+        :param qkv: an [N x (3 * H * C) x T] tensor of Qs, Ks, and Vs.
+        :return: an [N x (H * C) x T] tensor after attention.
+        """
+        bs, width, length = qkv.shape
+        assert width % (3 * self.n_heads) == 0
+        ch = width // (3 * self.n_heads)
+        q, k, v = qkv.chunk(3, dim=1)
+        scale = 1 / math.sqrt(math.sqrt(ch))
+        weight = th.einsum(
+            "bct,bcs->bts",
+            (q * scale).view(bs * self.n_heads, ch, length),
+            (k * scale).view(bs * self.n_heads, ch, length),
+        )  # More stable with f16 than dividing afterwards
+        weight = th.softmax(weight.float(), dim=-1).type(weight.dtype)
+        a = th.einsum(
+            "bts,bcs->bct",
+            weight,
+            v.reshape(bs * self.n_heads, ch, length).contiguous(),
+        )
+        return a.reshape(bs, -1, length).contiguous()
+
+    @staticmethod
+    def count_flops(model, _x, y):
+        return count_flops_attn(model, _x, y)
+
+
+class UNetModel(nn.Module):
+    """
+    The full UNet model with attention and timestep embedding.
+    :param in_channels: channels in the input Tensor.
+    :param model_channels: base channel count for the model.
+    :param out_channels: channels in the output Tensor.
+    :param num_res_blocks: number of residual blocks per downsample.
+    :param attention_resolutions: a collection of downsample rates at which
+        attention will take place. May be a set, list, or tuple.
+        For example, if this contains 4, then at 4x downsampling, attention
+        will be used.
+    :param dropout: the dropout probability.
+    :param channel_mult: channel multiplier for each level of the UNet.
+    :param conv_resample: if True, use learned convolutions for upsampling and
+        downsampling.
+    :param dims: determines if the signal is 1D, 2D, or 3D.
+    :param num_classes: if specified (as an int), then this model will be
+        class-conditional with `num_classes` classes.
+    :param use_checkpoint: use gradient checkpointing to reduce memory usage.
+    :param num_heads: the number of attention heads in each attention layer.
+    :param num_heads_channels: if specified, ignore num_heads and instead use
+                               a fixed channel width per attention head.
+    :param num_heads_upsample: works with num_heads to set a different number
+                               of heads for upsampling. Deprecated.
+    :param use_scale_shift_norm: use a FiLM-like conditioning mechanism.
+    :param resblock_updown: use residual blocks for up/downsampling.
+    :param use_new_attention_order: use a different attention pattern for potentially
+                                    increased efficiency.
+    """
+
+    def __init__(
+        self,
+        image_size,
+        in_channels,
+        model_channels,
+        out_channels,
+        num_res_blocks,
+        attention_resolutions,
+        dropout=0,
+        channel_mult=(1, 2, 4, 8),
+        conv_resample=True,
+        dims=2,
+        num_classes=None,
+        extra_film_condition_dim=None,
+        use_checkpoint=False,
+        use_fp16=False,
+        num_heads=-1,
+        num_head_channels=-1,
+        num_heads_upsample=-1,
+        use_scale_shift_norm=False,
+        extra_film_use_concat=False,  # If true, concatenate extrafilm condition with time embedding, else addition
+        resblock_updown=False,
+        use_new_attention_order=False,
+        use_spatial_transformer=False,  # custom transformer support
+        transformer_depth=1,  # custom transformer support
+        context_dim=None,  # custom transformer support
+        n_embed=None,  # custom support for prediction of discrete ids into codebook of first stage vq model
+        legacy=True,
+    ):
+        super().__init__()
+        if num_heads_upsample == -1:
+            num_heads_upsample = num_heads
+
+        if num_heads == -1:
+            assert (
+                num_head_channels != -1
+            ), "Either num_heads or num_head_channels has to be set"
+
+        if num_head_channels == -1:
+            assert (
+                num_heads != -1
+            ), "Either num_heads or num_head_channels has to be set"
+
+        self.image_size = image_size
+        self.in_channels = in_channels
+        self.model_channels = model_channels
+        self.out_channels = out_channels
+        self.num_res_blocks = num_res_blocks
+        self.attention_resolutions = attention_resolutions
+        self.dropout = dropout
+        self.channel_mult = channel_mult
+        self.conv_resample = conv_resample
+        self.num_classes = num_classes
+        self.extra_film_condition_dim = extra_film_condition_dim
+        self.use_checkpoint = use_checkpoint
+        self.dtype = th.float16 if use_fp16 else th.float32
+        self.num_heads = num_heads
+        self.num_head_channels = num_head_channels
+        self.num_heads_upsample = num_heads_upsample
+        self.predict_codebook_ids = n_embed is not None
+        self.extra_film_use_concat = extra_film_use_concat
+        time_embed_dim = model_channels * 4
+        self.time_embed = nn.Sequential(
+            linear(model_channels, time_embed_dim),
+            nn.SiLU(),
+            linear(time_embed_dim, time_embed_dim),
+        )
+
+        assert not (
+            self.num_classes is not None and self.extra_film_condition_dim is not None
+        ), "As for the condition of theh UNet model, you can only set using class label or an extra embedding vector (such as from CLAP). You cannot set both num_classes and extra_film_condition_dim."
+
+        if self.num_classes is not None:
+            self.label_emb = nn.Embedding(num_classes, time_embed_dim)
+
+        self.use_extra_film_by_concat = (
+            self.extra_film_condition_dim is not None and self.extra_film_use_concat
+        )
+        self.use_extra_film_by_addition = (
+            self.extra_film_condition_dim is not None and not self.extra_film_use_concat
+        )
+
+        if self.extra_film_condition_dim is not None:
+            self.film_emb = nn.Linear(self.extra_film_condition_dim, time_embed_dim)
+            # print("+ Use extra condition on UNet channel using Film. Extra condition dimension is %s. " % self.extra_film_condition_dim)
+            # if(self.use_extra_film_by_concat):
+            #     print("\t By concatenation with time embedding")
+            # elif(self.use_extra_film_by_concat):
+            #     print("\t By addition with time embedding")
+
+        if use_spatial_transformer and (
+            self.use_extra_film_by_concat or self.use_extra_film_by_addition
+        ):
+            # print("+ Spatial transformer will only be used as self-attention. Because you have choose to use film as your global condition.")
+            spatial_transformer_no_context = True
+        else:
+            spatial_transformer_no_context = False
+
+        if use_spatial_transformer and not spatial_transformer_no_context:
+            assert (
+                context_dim is not None
+            ), "Fool!! You forgot to include the dimension of your cross-attention conditioning..."
+
+        if context_dim is not None and not spatial_transformer_no_context:
+            assert (
+                use_spatial_transformer
+            ), "Fool!! You forgot to use the spatial transformer for your cross-attention conditioning..."
+            from omegaconf.listconfig import ListConfig
+
+            if type(context_dim) == ListConfig:
+                context_dim = list(context_dim)
+
+        self.input_blocks = nn.ModuleList(
+            [
+                TimestepEmbedSequential(
+                    conv_nd(dims, in_channels, model_channels, 3, padding=1)
+                )
+            ]
+        )
+        self._feature_size = model_channels
+        input_block_chans = [model_channels]
+        ch = model_channels
+        ds = 1
+        for level, mult in enumerate(channel_mult):
+            for _ in range(num_res_blocks):
+                layers = [
+                    ResBlock(
+                        ch,
+                        time_embed_dim
+                        if (not self.use_extra_film_by_concat)
+                        else time_embed_dim * 2,
+                        dropout,
+                        out_channels=mult * model_channels,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                    )
+                ]
+                ch = mult * model_channels
+                if ds in attention_resolutions:
+                    if num_head_channels == -1:
+                        dim_head = ch // num_heads
+                    else:
+                        num_heads = ch // num_head_channels
+                        dim_head = num_head_channels
+                    if legacy:
+                        dim_head = (
+                            ch // num_heads
+                            if use_spatial_transformer
+                            else num_head_channels
+                        )
+                    layers.append(
+                        AttentionBlock(
+                            ch,
+                            use_checkpoint=use_checkpoint,
+                            num_heads=num_heads,
+                            num_head_channels=dim_head,
+                            use_new_attention_order=use_new_attention_order,
+                        )
+                        if not use_spatial_transformer
+                        else SpatialTransformer(
+                            ch,
+                            num_heads,
+                            dim_head,
+                            depth=transformer_depth,
+                            context_dim=context_dim,
+                            no_context=spatial_transformer_no_context,
+                        )
+                    )
+                self.input_blocks.append(TimestepEmbedSequential(*layers))
+                self._feature_size += ch
+                input_block_chans.append(ch)
+            if level != len(channel_mult) - 1:
+                out_ch = ch
+                self.input_blocks.append(
+                    TimestepEmbedSequential(
+                        ResBlock(
+                            ch,
+                            time_embed_dim
+                            if (not self.use_extra_film_by_concat)
+                            else time_embed_dim * 2,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            down=True,
+                        )
+                        if resblock_updown
+                        else Downsample(
+                            ch, conv_resample, dims=dims, out_channels=out_ch
+                        )
+                    )
+                )
+                ch = out_ch
+                input_block_chans.append(ch)
+                ds *= 2
+                self._feature_size += ch
+
+        if num_head_channels == -1:
+            dim_head = ch // num_heads
+        else:
+            num_heads = ch // num_head_channels
+            dim_head = num_head_channels
+        if legacy:
+            # num_heads = 1
+            dim_head = ch // num_heads if use_spatial_transformer else num_head_channels
+        self.middle_block = TimestepEmbedSequential(
+            ResBlock(
+                ch,
+                time_embed_dim
+                if (not self.use_extra_film_by_concat)
+                else time_embed_dim * 2,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+            ),
+            AttentionBlock(
+                ch,
+                use_checkpoint=use_checkpoint,
+                num_heads=num_heads,
+                num_head_channels=dim_head,
+                use_new_attention_order=use_new_attention_order,
+            )
+            if not use_spatial_transformer
+            else SpatialTransformer(
+                ch,
+                num_heads,
+                dim_head,
+                depth=transformer_depth,
+                context_dim=context_dim,
+                no_context=spatial_transformer_no_context,
+            ),
+            ResBlock(
+                ch,
+                time_embed_dim
+                if (not self.use_extra_film_by_concat)
+                else time_embed_dim * 2,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+            ),
+        )
+        self._feature_size += ch
+
+        self.output_blocks = nn.ModuleList([])
+        for level, mult in list(enumerate(channel_mult))[::-1]:
+            for i in range(num_res_blocks + 1):
+                ich = input_block_chans.pop()
+                layers = [
+                    ResBlock(
+                        ch + ich,
+                        time_embed_dim
+                        if (not self.use_extra_film_by_concat)
+                        else time_embed_dim * 2,
+                        dropout,
+                        out_channels=model_channels * mult,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                    )
+                ]
+                ch = model_channels * mult
+                if ds in attention_resolutions:
+                    if num_head_channels == -1:
+                        dim_head = ch // num_heads
+                    else:
+                        num_heads = ch // num_head_channels
+                        dim_head = num_head_channels
+                    if legacy:
+                        # num_heads = 1
+                        dim_head = (
+                            ch // num_heads
+                            if use_spatial_transformer
+                            else num_head_channels
+                        )
+                    layers.append(
+                        AttentionBlock(
+                            ch,
+                            use_checkpoint=use_checkpoint,
+                            num_heads=num_heads_upsample,
+                            num_head_channels=dim_head,
+                            use_new_attention_order=use_new_attention_order,
+                        )
+                        if not use_spatial_transformer
+                        else SpatialTransformer(
+                            ch,
+                            num_heads,
+                            dim_head,
+                            depth=transformer_depth,
+                            context_dim=context_dim,
+                            no_context=spatial_transformer_no_context,
+                        )
+                    )
+                if level and i == num_res_blocks:
+                    out_ch = ch
+                    layers.append(
+                        ResBlock(
+                            ch,
+                            time_embed_dim
+                            if (not self.use_extra_film_by_concat)
+                            else time_embed_dim * 2,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            up=True,
+                        )
+                        if resblock_updown
+                        else Upsample(ch, conv_resample, dims=dims, out_channels=out_ch)
+                    )
+                    ds //= 2
+                self.output_blocks.append(TimestepEmbedSequential(*layers))
+                self._feature_size += ch
+
+        self.out = nn.Sequential(
+            normalization(ch),
+            nn.SiLU(),
+            zero_module(conv_nd(dims, model_channels, out_channels, 3, padding=1)),
+        )
+        if self.predict_codebook_ids:
+            self.id_predictor = nn.Sequential(
+                normalization(ch),
+                conv_nd(dims, model_channels, n_embed, 1),
+                # nn.LogSoftmax(dim=1)  # change to cross_entropy and produce non-normalized logits
+            )
+
+        self.shape_reported = False
+
+    def convert_to_fp16(self):
+        """
+        Convert the torso of the model to float16.
+        """
+        self.input_blocks.apply(convert_module_to_f16)
+        self.middle_block.apply(convert_module_to_f16)
+        self.output_blocks.apply(convert_module_to_f16)
+
+    def convert_to_fp32(self):
+        """
+        Convert the torso of the model to float32.
+        """
+        self.input_blocks.apply(convert_module_to_f32)
+        self.middle_block.apply(convert_module_to_f32)
+        self.output_blocks.apply(convert_module_to_f32)
+
+    def forward(self, x, timesteps=None, context=None, y=None, **kwargs):
+        """
+        Apply the model to an input batch.
+        :param x: an [N x C x ...] Tensor of inputs.
+        :param timesteps: a 1-D batch of timesteps.
+        :param context: conditioning plugged in via crossattn
+        :param y: an [N] Tensor of labels, if class-conditional. an [N, extra_film_condition_dim] Tensor if film-embed conditional
+        :return: an [N x C x ...] Tensor of outputs.
+        """
+        if not self.shape_reported:
+            # print("The shape of UNet input is", x.size())
+            self.shape_reported = True
+
+        assert (y is not None) == (
+            self.num_classes is not None or self.extra_film_condition_dim is not None
+        ), "must specify y if and only if the model is class-conditional or film embedding conditional"
+        hs = []
+        t_emb = timestep_embedding(timesteps, self.model_channels, repeat_only=False)
+        emb = self.time_embed(t_emb)
+
+        if self.num_classes is not None:
+            assert y.shape == (x.shape[0],)
+            emb = emb + self.label_emb(y)
+
+        if self.use_extra_film_by_addition:
+            emb = emb + self.film_emb(y)
+        elif self.use_extra_film_by_concat:
+            emb = th.cat([emb, self.film_emb(y)], dim=-1)
+
+        h = x.type(self.dtype)
+        for module in self.input_blocks:
+            h = module(h, emb, context)
+            hs.append(h)
+        h = self.middle_block(h, emb, context)
+        for module in self.output_blocks:
+            h = th.cat([h, hs.pop()], dim=1)
+            h = module(h, emb, context)
+        h = h.type(x.dtype)
+        if self.predict_codebook_ids:
+            return self.id_predictor(h)
+        else:
+            return self.out(h)
+
+
+class EncoderUNetModel(nn.Module):
+    """
+    The half UNet model with attention and timestep embedding.
+    For usage, see UNet.
+    """
+
+    def __init__(
+        self,
+        image_size,
+        in_channels,
+        model_channels,
+        out_channels,
+        num_res_blocks,
+        attention_resolutions,
+        dropout=0,
+        channel_mult=(1, 2, 4, 8),
+        conv_resample=True,
+        dims=2,
+        use_checkpoint=False,
+        use_fp16=False,
+        num_heads=1,
+        num_head_channels=-1,
+        num_heads_upsample=-1,
+        use_scale_shift_norm=False,
+        resblock_updown=False,
+        use_new_attention_order=False,
+        pool="adaptive",
+        *args,
+        **kwargs,
+    ):
+        super().__init__()
+
+        if num_heads_upsample == -1:
+            num_heads_upsample = num_heads
+
+        self.in_channels = in_channels
+        self.model_channels = model_channels
+        self.out_channels = out_channels
+        self.num_res_blocks = num_res_blocks
+        self.attention_resolutions = attention_resolutions
+        self.dropout = dropout
+        self.channel_mult = channel_mult
+        self.conv_resample = conv_resample
+        self.use_checkpoint = use_checkpoint
+        self.dtype = th.float16 if use_fp16 else th.float32
+        self.num_heads = num_heads
+        self.num_head_channels = num_head_channels
+        self.num_heads_upsample = num_heads_upsample
+
+        time_embed_dim = model_channels * 4
+        self.time_embed = nn.Sequential(
+            linear(model_channels, time_embed_dim),
+            nn.SiLU(),
+            linear(time_embed_dim, time_embed_dim),
+        )
+
+        self.input_blocks = nn.ModuleList(
+            [
+                TimestepEmbedSequential(
+                    conv_nd(dims, in_channels, model_channels, 3, padding=1)
+                )
+            ]
+        )
+        self._feature_size = model_channels
+        input_block_chans = [model_channels]
+        ch = model_channels
+        ds = 1
+        for level, mult in enumerate(channel_mult):
+            for _ in range(num_res_blocks):
+                layers = [
+                    ResBlock(
+                        ch,
+                        time_embed_dim,
+                        dropout,
+                        out_channels=mult * model_channels,
+                        dims=dims,
+                        use_checkpoint=use_checkpoint,
+                        use_scale_shift_norm=use_scale_shift_norm,
+                    )
+                ]
+                ch = mult * model_channels
+                if ds in attention_resolutions:
+                    layers.append(
+                        AttentionBlock(
+                            ch,
+                            use_checkpoint=use_checkpoint,
+                            num_heads=num_heads,
+                            num_head_channels=num_head_channels,
+                            use_new_attention_order=use_new_attention_order,
+                        )
+                    )
+                self.input_blocks.append(TimestepEmbedSequential(*layers))
+                self._feature_size += ch
+                input_block_chans.append(ch)
+            if level != len(channel_mult) - 1:
+                out_ch = ch
+                self.input_blocks.append(
+                    TimestepEmbedSequential(
+                        ResBlock(
+                            ch,
+                            time_embed_dim,
+                            dropout,
+                            out_channels=out_ch,
+                            dims=dims,
+                            use_checkpoint=use_checkpoint,
+                            use_scale_shift_norm=use_scale_shift_norm,
+                            down=True,
+                        )
+                        if resblock_updown
+                        else Downsample(
+                            ch, conv_resample, dims=dims, out_channels=out_ch
+                        )
+                    )
+                )
+                ch = out_ch
+                input_block_chans.append(ch)
+                ds *= 2
+                self._feature_size += ch
+
+        self.middle_block = TimestepEmbedSequential(
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+            ),
+            AttentionBlock(
+                ch,
+                use_checkpoint=use_checkpoint,
+                num_heads=num_heads,
+                num_head_channels=num_head_channels,
+                use_new_attention_order=use_new_attention_order,
+            ),
+            ResBlock(
+                ch,
+                time_embed_dim,
+                dropout,
+                dims=dims,
+                use_checkpoint=use_checkpoint,
+                use_scale_shift_norm=use_scale_shift_norm,
+            ),
+        )
+        self._feature_size += ch
+        self.pool = pool
+        if pool == "adaptive":
+            self.out = nn.Sequential(
+                normalization(ch),
+                nn.SiLU(),
+                nn.AdaptiveAvgPool2d((1, 1)),
+                zero_module(conv_nd(dims, ch, out_channels, 1)),
+                nn.Flatten(),
+            )
+        elif pool == "attention":
+            assert num_head_channels != -1
+            self.out = nn.Sequential(
+                normalization(ch),
+                nn.SiLU(),
+                AttentionPool2d(
+                    (image_size // ds), ch, num_head_channels, out_channels
+                ),
+            )
+        elif pool == "spatial":
+            self.out = nn.Sequential(
+                nn.Linear(self._feature_size, 2048),
+                nn.ReLU(),
+                nn.Linear(2048, self.out_channels),
+            )
+        elif pool == "spatial_v2":
+            self.out = nn.Sequential(
+                nn.Linear(self._feature_size, 2048),
+                normalization(2048),
+                nn.SiLU(),
+                nn.Linear(2048, self.out_channels),
+            )
+        else:
+            raise NotImplementedError(f"Unexpected {pool} pooling")
+
+    def convert_to_fp16(self):
+        """
+        Convert the torso of the model to float16.
+        """
+        self.input_blocks.apply(convert_module_to_f16)
+        self.middle_block.apply(convert_module_to_f16)
+
+    def convert_to_fp32(self):
+        """
+        Convert the torso of the model to float32.
+        """
+        self.input_blocks.apply(convert_module_to_f32)
+        self.middle_block.apply(convert_module_to_f32)
+
+    def forward(self, x, timesteps):
+        """
+        Apply the model to an input batch.
+        :param x: an [N x C x ...] Tensor of inputs.
+        :param timesteps: a 1-D batch of timesteps.
+        :return: an [N x K] Tensor of outputs.
+        """
+        emb = self.time_embed(timestep_embedding(timesteps, self.model_channels))
+
+        results = []
+        h = x.type(self.dtype)
+        for module in self.input_blocks:
+            h = module(h, emb)
+            if self.pool.startswith("spatial"):
+                results.append(h.type(x.dtype).mean(dim=(2, 3)))
+        h = self.middle_block(h, emb)
+        if self.pool.startswith("spatial"):
+            results.append(h.type(x.dtype).mean(dim=(2, 3)))
+            h = th.cat(results, axis=-1)
+            return self.out(h)
+        else:
+            h = h.type(x.dtype)
+            return self.out(h)
diff --git a/picoaudio/audioldm/latent_diffusion/util.py b/picoaudio/audioldm/latent_diffusion/util.py
new file mode 100644
index 0000000000000000000000000000000000000000..8b289f6aa7f22a070870d8a706f944dc8547e936
--- /dev/null
+++ b/picoaudio/audioldm/latent_diffusion/util.py
@@ -0,0 +1,295 @@
+# adopted from
+# https://github.com/openai/improved-diffusion/blob/main/improved_diffusion/gaussian_diffusion.py
+# and
+# https://github.com/lucidrains/denoising-diffusion-pytorch/blob/7706bdfc6f527f58d33f84b7b522e61e6e3164b3/denoising_diffusion_pytorch/denoising_diffusion_pytorch.py
+# and
+# https://github.com/openai/guided-diffusion/blob/0ba878e517b276c45d1195eb29f6f5f72659a05b/guided_diffusion/nn.py
+#
+# thanks!
+
+
+import os
+import math
+import torch
+import torch.nn as nn
+import numpy as np
+from einops import repeat
+
+from audioldm.utils import instantiate_from_config
+
+
+def make_beta_schedule(
+    schedule, n_timestep, linear_start=1e-4, linear_end=2e-2, cosine_s=8e-3
+):
+    if schedule == "linear":
+        betas = (
+            torch.linspace(
+                linear_start**0.5, linear_end**0.5, n_timestep, dtype=torch.float64
+            )
+            ** 2
+        )
+
+    elif schedule == "cosine":
+        timesteps = (
+            torch.arange(n_timestep + 1, dtype=torch.float64) / n_timestep + cosine_s
+        )
+        alphas = timesteps / (1 + cosine_s) * np.pi / 2
+        alphas = torch.cos(alphas).pow(2)
+        alphas = alphas / alphas[0]
+        betas = 1 - alphas[1:] / alphas[:-1]
+        betas = np.clip(betas, a_min=0, a_max=0.999)
+
+    elif schedule == "sqrt_linear":
+        betas = torch.linspace(
+            linear_start, linear_end, n_timestep, dtype=torch.float64
+        )
+    elif schedule == "sqrt":
+        betas = (
+            torch.linspace(linear_start, linear_end, n_timestep, dtype=torch.float64)
+            ** 0.5
+        )
+    else:
+        raise ValueError(f"schedule '{schedule}' unknown.")
+    return betas.numpy()
+
+
+def make_ddim_timesteps(
+    ddim_discr_method, num_ddim_timesteps, num_ddpm_timesteps, verbose=True
+):
+    if ddim_discr_method == "uniform":
+        c = num_ddpm_timesteps // num_ddim_timesteps
+        ddim_timesteps = np.asarray(list(range(0, num_ddpm_timesteps, c)))
+    elif ddim_discr_method == "quad":
+        ddim_timesteps = (
+            (np.linspace(0, np.sqrt(num_ddpm_timesteps * 0.8), num_ddim_timesteps)) ** 2
+        ).astype(int)
+    else:
+        raise NotImplementedError(
+            f'There is no ddim discretization method called "{ddim_discr_method}"'
+        )
+
+    # assert ddim_timesteps.shape[0] == num_ddim_timesteps
+    # add one to get the final alpha values right (the ones from first scale to data during sampling)
+    steps_out = ddim_timesteps + 1
+    if verbose:
+        print(f"Selected timesteps for ddim sampler: {steps_out}")
+    return steps_out
+
+
+def make_ddim_sampling_parameters(alphacums, ddim_timesteps, eta, verbose=True):
+    # select alphas for computing the variance schedule
+    alphas = alphacums[ddim_timesteps]
+    alphas_prev = np.asarray([alphacums[0]] + alphacums[ddim_timesteps[:-1]].tolist())
+
+    # according the the formula provided in https://arxiv.org/abs/2010.02502
+    sigmas = eta * np.sqrt(
+        (1 - alphas_prev) / (1 - alphas) * (1 - alphas / alphas_prev)
+    )
+    if verbose:
+        print(
+            f"Selected alphas for ddim sampler: a_t: {alphas}; a_(t-1): {alphas_prev}"
+        )
+        print(
+            f"For the chosen value of eta, which is {eta}, "
+            f"this results in the following sigma_t schedule for ddim sampler {sigmas}"
+        )
+    return sigmas, alphas, alphas_prev
+
+
+def betas_for_alpha_bar(num_diffusion_timesteps, alpha_bar, max_beta=0.999):
+    """
+    Create a beta schedule that discretizes the given alpha_t_bar function,
+    which defines the cumulative product of (1-beta) over time from t = [0,1].
+    :param num_diffusion_timesteps: the number of betas to produce.
+    :param alpha_bar: a lambda that takes an argument t from 0 to 1 and
+                      produces the cumulative product of (1-beta) up to that
+                      part of the diffusion process.
+    :param max_beta: the maximum beta to use; use values lower than 1 to
+                     prevent singularities.
+    """
+    betas = []
+    for i in range(num_diffusion_timesteps):
+        t1 = i / num_diffusion_timesteps
+        t2 = (i + 1) / num_diffusion_timesteps
+        betas.append(min(1 - alpha_bar(t2) / alpha_bar(t1), max_beta))
+    return np.array(betas)
+
+
+def extract_into_tensor(a, t, x_shape):
+    b, *_ = t.shape
+    out = a.gather(-1, t).contiguous()
+    return out.reshape(b, *((1,) * (len(x_shape) - 1))).contiguous()
+
+
+def checkpoint(func, inputs, params, flag):
+    """
+    Evaluate a function without caching intermediate activations, allowing for
+    reduced memory at the expense of extra compute in the backward pass.
+    :param func: the function to evaluate.
+    :param inputs: the argument sequence to pass to `func`.
+    :param params: a sequence of parameters `func` depends on but does not
+                   explicitly take as arguments.
+    :param flag: if False, disable gradient checkpointing.
+    """
+    if flag:
+        args = tuple(inputs) + tuple(params)
+        return CheckpointFunction.apply(func, len(inputs), *args)
+    else:
+        return func(*inputs)
+
+
+class CheckpointFunction(torch.autograd.Function):
+    @staticmethod
+    def forward(ctx, run_function, length, *args):
+        ctx.run_function = run_function
+        ctx.input_tensors = list(args[:length])
+        ctx.input_params = list(args[length:])
+
+        with torch.no_grad():
+            output_tensors = ctx.run_function(*ctx.input_tensors)
+        return output_tensors
+
+    @staticmethod
+    def backward(ctx, *output_grads):
+        ctx.input_tensors = [x.detach().requires_grad_(True) for x in ctx.input_tensors]
+        with torch.enable_grad():
+            # Fixes a bug where the first op in run_function modifies the
+            # Tensor storage in place, which is not allowed for detach()'d
+            # Tensors.
+            shallow_copies = [x.view_as(x) for x in ctx.input_tensors]
+            output_tensors = ctx.run_function(*shallow_copies)
+        input_grads = torch.autograd.grad(
+            output_tensors,
+            ctx.input_tensors + ctx.input_params,
+            output_grads,
+            allow_unused=True,
+        )
+        del ctx.input_tensors
+        del ctx.input_params
+        del output_tensors
+        return (None, None) + input_grads
+
+
+def timestep_embedding(timesteps, dim, max_period=10000, repeat_only=False):
+    """
+    Create sinusoidal timestep embeddings.
+    :param timesteps: a 1-D Tensor of N indices, one per batch element.
+                      These may be fractional.
+    :param dim: the dimension of the output.
+    :param max_period: controls the minimum frequency of the embeddings.
+    :return: an [N x dim] Tensor of positional embeddings.
+    """
+    if not repeat_only:
+        half = dim // 2
+        freqs = torch.exp(
+            -math.log(max_period)
+            * torch.arange(start=0, end=half, dtype=torch.float32)
+            / half
+        ).to(device=timesteps.device)
+        args = timesteps[:, None].float() * freqs[None]
+        embedding = torch.cat([torch.cos(args), torch.sin(args)], dim=-1)
+        if dim % 2:
+            embedding = torch.cat(
+                [embedding, torch.zeros_like(embedding[:, :1])], dim=-1
+            )
+    else:
+        embedding = repeat(timesteps, "b -> b d", d=dim)
+    return embedding
+
+
+def zero_module(module):
+    """
+    Zero out the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().zero_()
+    return module
+
+
+def scale_module(module, scale):
+    """
+    Scale the parameters of a module and return it.
+    """
+    for p in module.parameters():
+        p.detach().mul_(scale)
+    return module
+
+
+def mean_flat(tensor):
+    """
+    Take the mean over all non-batch dimensions.
+    """
+    return tensor.mean(dim=list(range(1, len(tensor.shape))))
+
+
+def normalization(channels):
+    """
+    Make a standard normalization layer.
+    :param channels: number of input channels.
+    :return: an nn.Module for normalization.
+    """
+    return GroupNorm32(32, channels)
+
+
+# PyTorch 1.7 has SiLU, but we support PyTorch 1.5.
+class SiLU(nn.Module):
+    def forward(self, x):
+        return x * torch.sigmoid(x)
+
+
+class GroupNorm32(nn.GroupNorm):
+    def forward(self, x):
+        return super().forward(x.float()).type(x.dtype)
+
+
+def conv_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D convolution module.
+    """
+    if dims == 1:
+        return nn.Conv1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.Conv2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.Conv3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+def linear(*args, **kwargs):
+    """
+    Create a linear module.
+    """
+    return nn.Linear(*args, **kwargs)
+
+
+def avg_pool_nd(dims, *args, **kwargs):
+    """
+    Create a 1D, 2D, or 3D average pooling module.
+    """
+    if dims == 1:
+        return nn.AvgPool1d(*args, **kwargs)
+    elif dims == 2:
+        return nn.AvgPool2d(*args, **kwargs)
+    elif dims == 3:
+        return nn.AvgPool3d(*args, **kwargs)
+    raise ValueError(f"unsupported dimensions: {dims}")
+
+
+class HybridConditioner(nn.Module):
+    def __init__(self, c_concat_config, c_crossattn_config):
+        super().__init__()
+        self.concat_conditioner = instantiate_from_config(c_concat_config)
+        self.crossattn_conditioner = instantiate_from_config(c_crossattn_config)
+
+    def forward(self, c_concat, c_crossattn):
+        c_concat = self.concat_conditioner(c_concat)
+        c_crossattn = self.crossattn_conditioner(c_crossattn)
+        return {"c_concat": [c_concat], "c_crossattn": [c_crossattn]}
+
+
+def noise_like(shape, device, repeat=False):
+    repeat_noise = lambda: torch.randn((1, *shape[1:]), device=device).repeat(
+        shape[0], *((1,) * (len(shape) - 1))
+    )
+    noise = lambda: torch.randn(shape, device=device)
+    return repeat_noise() if repeat else noise()
diff --git a/picoaudio/audioldm/ldm.py b/picoaudio/audioldm/ldm.py
new file mode 100644
index 0000000000000000000000000000000000000000..e0179fd5a506052ac9db22bd37f3db6b910aded5
--- /dev/null
+++ b/picoaudio/audioldm/ldm.py
@@ -0,0 +1,818 @@
+import os
+
+import torch
+import numpy as np
+from tqdm import tqdm
+from audioldm.utils import default, instantiate_from_config, save_wave
+from audioldm.latent_diffusion.ddpm import DDPM
+from audioldm.variational_autoencoder.distributions import DiagonalGaussianDistribution
+from audioldm.latent_diffusion.util import noise_like
+from audioldm.latent_diffusion.ddim import DDIMSampler
+import os
+
+
+def disabled_train(self, mode=True):
+    """Overwrite model.train with this function to make sure train/eval mode
+    does not change anymore."""
+    return self
+
+
+class LatentDiffusion(DDPM):
+    """main class"""
+
+    def __init__(
+        self,
+        device="cuda",
+        first_stage_config=None,
+        cond_stage_config=None,
+        num_timesteps_cond=None,
+        cond_stage_key="image",
+        cond_stage_trainable=False,
+        concat_mode=True,
+        cond_stage_forward=None,
+        conditioning_key=None,
+        scale_factor=1.0,
+        scale_by_std=False,
+        base_learning_rate=None,
+        *args,
+        **kwargs,
+    ):
+        self.device = device
+        self.learning_rate = base_learning_rate
+        self.num_timesteps_cond = default(num_timesteps_cond, 1)
+        self.scale_by_std = scale_by_std
+        assert self.num_timesteps_cond <= kwargs["timesteps"]
+        # for backwards compatibility after implementation of DiffusionWrapper
+        if conditioning_key is None:
+            conditioning_key = "concat" if concat_mode else "crossattn"
+        if cond_stage_config == "__is_unconditional__":
+            conditioning_key = None
+        ckpt_path = kwargs.pop("ckpt_path", None)
+        ignore_keys = kwargs.pop("ignore_keys", [])
+        super().__init__(conditioning_key=conditioning_key, *args, **kwargs)
+        self.concat_mode = concat_mode
+        self.cond_stage_trainable = cond_stage_trainable
+        self.cond_stage_key = cond_stage_key
+        self.cond_stage_key_orig = cond_stage_key
+        try:
+            self.num_downs = len(first_stage_config.params.ddconfig.ch_mult) - 1
+        except:
+            self.num_downs = 0
+        if not scale_by_std:
+            self.scale_factor = scale_factor
+        else:
+            self.register_buffer("scale_factor", torch.tensor(scale_factor))
+        self.instantiate_first_stage(first_stage_config)
+        self.instantiate_cond_stage(cond_stage_config)
+        self.cond_stage_forward = cond_stage_forward
+        self.clip_denoised = False
+
+    def make_cond_schedule(
+        self,
+    ):
+        self.cond_ids = torch.full(
+            size=(self.num_timesteps,),
+            fill_value=self.num_timesteps - 1,
+            dtype=torch.long,
+        )
+        ids = torch.round(
+            torch.linspace(0, self.num_timesteps - 1, self.num_timesteps_cond)
+        ).long()
+        self.cond_ids[: self.num_timesteps_cond] = ids
+
+    def register_schedule(
+        self,
+        given_betas=None,
+        beta_schedule="linear",
+        timesteps=1000,
+        linear_start=1e-4,
+        linear_end=2e-2,
+        cosine_s=8e-3,
+    ):
+        super().register_schedule(
+            given_betas, beta_schedule, timesteps, linear_start, linear_end, cosine_s
+        )
+
+        self.shorten_cond_schedule = self.num_timesteps_cond > 1
+        if self.shorten_cond_schedule:
+            self.make_cond_schedule()
+
+    def instantiate_first_stage(self, config):
+        model = instantiate_from_config(config)
+        self.first_stage_model = model.eval()
+        self.first_stage_model.train = disabled_train
+        for param in self.first_stage_model.parameters():
+            param.requires_grad = False
+
+    def instantiate_cond_stage(self, config):
+        if not self.cond_stage_trainable:
+            if config == "__is_first_stage__":
+                print("Using first stage also as cond stage.")
+                self.cond_stage_model = self.first_stage_model
+            elif config == "__is_unconditional__":
+                print(f"Training {self.__class__.__name__} as an unconditional model.")
+                self.cond_stage_model = None
+                # self.be_unconditional = True
+            else:
+                model = instantiate_from_config(config)
+                self.cond_stage_model = model.eval()
+                self.cond_stage_model.train = disabled_train
+                for param in self.cond_stage_model.parameters():
+                    param.requires_grad = False
+        else:
+            assert config != "__is_first_stage__"
+            assert config != "__is_unconditional__"
+            model = instantiate_from_config(config)
+            self.cond_stage_model = model
+        self.cond_stage_model = self.cond_stage_model.to(self.device)
+
+    def get_first_stage_encoding(self, encoder_posterior):
+        if isinstance(encoder_posterior, DiagonalGaussianDistribution):
+            z = encoder_posterior.sample()
+        elif isinstance(encoder_posterior, torch.Tensor):
+            z = encoder_posterior
+        else:
+            raise NotImplementedError(
+                f"encoder_posterior of type '{type(encoder_posterior)}' not yet implemented"
+            )
+        return self.scale_factor * z
+
+    def get_learned_conditioning(self, c):
+        if self.cond_stage_forward is None:
+            if hasattr(self.cond_stage_model, "encode") and callable(
+                self.cond_stage_model.encode
+            ):
+                c = self.cond_stage_model.encode(c)
+                if isinstance(c, DiagonalGaussianDistribution):
+                    c = c.mode()
+            else:
+                # Text input is list
+                if type(c) == list and len(c) == 1:
+                    c = self.cond_stage_model([c[0], c[0]])
+                    c = c[0:1]
+                else:
+                    c = self.cond_stage_model(c)
+        else:
+            assert hasattr(self.cond_stage_model, self.cond_stage_forward)
+            c = getattr(self.cond_stage_model, self.cond_stage_forward)(c)
+        return c
+
+    @torch.no_grad()
+    def get_input(
+        self,
+        batch,
+        k,
+        return_first_stage_encode=True,
+        return_first_stage_outputs=False,
+        force_c_encode=False,
+        cond_key=None,
+        return_original_cond=False,
+        bs=None,
+    ):
+        x = super().get_input(batch, k)
+
+        if bs is not None:
+            x = x[:bs]
+
+        x = x.to(self.device)
+
+        if return_first_stage_encode:
+            encoder_posterior = self.encode_first_stage(x)
+            z = self.get_first_stage_encoding(encoder_posterior).detach()
+        else:
+            z = None
+
+        if self.model.conditioning_key is not None:
+            if cond_key is None:
+                cond_key = self.cond_stage_key
+            if cond_key != self.first_stage_key:
+                if cond_key in ["caption", "coordinates_bbox"]:
+                    xc = batch[cond_key]
+                elif cond_key == "class_label":
+                    xc = batch
+                else:
+                    # [bs, 1, 527]
+                    xc = super().get_input(batch, cond_key)
+                    if type(xc) == torch.Tensor:
+                        xc = xc.to(self.device)
+            else:
+                xc = x
+            if not self.cond_stage_trainable or force_c_encode:
+                if isinstance(xc, dict) or isinstance(xc, list):
+                    c = self.get_learned_conditioning(xc)
+                else:
+                    c = self.get_learned_conditioning(xc.to(self.device))
+            else:
+                c = xc
+
+            if bs is not None:
+                c = c[:bs]
+
+        else:
+            c = None
+            xc = None
+            if self.use_positional_encodings:
+                pos_x, pos_y = self.compute_latent_shifts(batch)
+                c = {"pos_x": pos_x, "pos_y": pos_y}
+        out = [z, c]
+        if return_first_stage_outputs:
+            xrec = self.decode_first_stage(z)
+            out.extend([x, xrec])
+        if return_original_cond:
+            out.append(xc)
+        return out
+
+    @torch.no_grad()
+    def decode_first_stage(self, z, predict_cids=False, force_not_quantize=False):
+        if predict_cids:
+            if z.dim() == 4:
+                z = torch.argmax(z.exp(), dim=1).long()
+            z = self.first_stage_model.quantize.get_codebook_entry(z, shape=None)
+            z = rearrange(z, "b h w c -> b c h w").contiguous()
+
+        z = 1.0 / self.scale_factor * z
+        return self.first_stage_model.decode(z)
+
+    def mel_spectrogram_to_waveform(self, mel):
+        # Mel: [bs, 1, t-steps, fbins]
+        if len(mel.size()) == 4:
+            mel = mel.squeeze(1)
+        mel = mel.permute(0, 2, 1)
+        waveform = self.first_stage_model.vocoder(mel)
+        waveform = waveform.cpu().detach().numpy()
+        return waveform
+
+    @torch.no_grad()
+    def encode_first_stage(self, x):
+        return self.first_stage_model.encode(x)
+
+    def apply_model(self, x_noisy, t, cond, return_ids=False):
+
+        if isinstance(cond, dict):
+            # hybrid case, cond is exptected to be a dict
+            pass
+        else:
+            if not isinstance(cond, list):
+                cond = [cond]
+            if self.model.conditioning_key == "concat":
+                key = "c_concat"
+            elif self.model.conditioning_key == "crossattn":
+                key = "c_crossattn"
+            else:
+                key = "c_film"
+
+            cond = {key: cond}
+
+        x_recon = self.model(x_noisy, t, **cond)
+
+        if isinstance(x_recon, tuple) and not return_ids:
+            return x_recon[0]
+        else:
+            return x_recon
+
+    def p_mean_variance(
+        self,
+        x,
+        c,
+        t,
+        clip_denoised: bool,
+        return_codebook_ids=False,
+        quantize_denoised=False,
+        return_x0=False,
+        score_corrector=None,
+        corrector_kwargs=None,
+    ):
+        t_in = t
+        model_out = self.apply_model(x, t_in, c, return_ids=return_codebook_ids)
+
+        if score_corrector is not None:
+            assert self.parameterization == "eps"
+            model_out = score_corrector.modify_score(
+                self, model_out, x, t, c, **corrector_kwargs
+            )
+
+        if return_codebook_ids:
+            model_out, logits = model_out
+
+        if self.parameterization == "eps":
+            x_recon = self.predict_start_from_noise(x, t=t, noise=model_out)
+        elif self.parameterization == "x0":
+            x_recon = model_out
+        else:
+            raise NotImplementedError()
+
+        if clip_denoised:
+            x_recon.clamp_(-1.0, 1.0)
+        if quantize_denoised:
+            x_recon, _, [_, _, indices] = self.first_stage_model.quantize(x_recon)
+        model_mean, posterior_variance, posterior_log_variance = self.q_posterior(
+            x_start=x_recon, x_t=x, t=t
+        )
+        if return_codebook_ids:
+            return model_mean, posterior_variance, posterior_log_variance, logits
+        elif return_x0:
+            return model_mean, posterior_variance, posterior_log_variance, x_recon
+        else:
+            return model_mean, posterior_variance, posterior_log_variance
+
+    @torch.no_grad()
+    def p_sample(
+        self,
+        x,
+        c,
+        t,
+        clip_denoised=False,
+        repeat_noise=False,
+        return_codebook_ids=False,
+        quantize_denoised=False,
+        return_x0=False,
+        temperature=1.0,
+        noise_dropout=0.0,
+        score_corrector=None,
+        corrector_kwargs=None,
+    ):
+        b, *_, device = *x.shape, x.device
+        outputs = self.p_mean_variance(
+            x=x,
+            c=c,
+            t=t,
+            clip_denoised=clip_denoised,
+            return_codebook_ids=return_codebook_ids,
+            quantize_denoised=quantize_denoised,
+            return_x0=return_x0,
+            score_corrector=score_corrector,
+            corrector_kwargs=corrector_kwargs,
+        )
+        if return_codebook_ids:
+            raise DeprecationWarning("Support dropped.")
+            model_mean, _, model_log_variance, logits = outputs
+        elif return_x0:
+            model_mean, _, model_log_variance, x0 = outputs
+        else:
+            model_mean, _, model_log_variance = outputs
+
+        noise = noise_like(x.shape, device, repeat_noise) * temperature
+        if noise_dropout > 0.0:
+            noise = torch.nn.functional.dropout(noise, p=noise_dropout)
+        # no noise when t == 0
+        nonzero_mask = (
+            (1 - (t == 0).float()).reshape(b, *((1,) * (len(x.shape) - 1))).contiguous()
+        )
+
+        if return_codebook_ids:
+            return model_mean + nonzero_mask * (
+                0.5 * model_log_variance
+            ).exp() * noise, logits.argmax(dim=1)
+        if return_x0:
+            return (
+                model_mean + nonzero_mask * (0.5 * model_log_variance).exp() * noise,
+                x0,
+            )
+        else:
+            return model_mean + nonzero_mask * (0.5 * model_log_variance).exp() * noise
+
+    @torch.no_grad()
+    def progressive_denoising(
+        self,
+        cond,
+        shape,
+        verbose=True,
+        callback=None,
+        quantize_denoised=False,
+        img_callback=None,
+        mask=None,
+        x0=None,
+        temperature=1.0,
+        noise_dropout=0.0,
+        score_corrector=None,
+        corrector_kwargs=None,
+        batch_size=None,
+        x_T=None,
+        start_T=None,
+        log_every_t=None,
+    ):
+        if not log_every_t:
+            log_every_t = self.log_every_t
+        timesteps = self.num_timesteps
+        if batch_size is not None:
+            b = batch_size if batch_size is not None else shape[0]
+            shape = [batch_size] + list(shape)
+        else:
+            b = batch_size = shape[0]
+        if x_T is None:
+            img = torch.randn(shape, device=self.device)
+        else:
+            img = x_T
+        intermediates = []
+        if cond is not None:
+            if isinstance(cond, dict):
+                cond = {
+                    key: cond[key][:batch_size]
+                    if not isinstance(cond[key], list)
+                    else list(map(lambda x: x[:batch_size], cond[key]))
+                    for key in cond
+                }
+            else:
+                cond = (
+                    [c[:batch_size] for c in cond]
+                    if isinstance(cond, list)
+                    else cond[:batch_size]
+                )
+
+        if start_T is not None:
+            timesteps = min(timesteps, start_T)
+        iterator = (
+            tqdm(
+                reversed(range(0, timesteps)),
+                desc="Progressive Generation",
+                total=timesteps,
+            )
+            if verbose
+            else reversed(range(0, timesteps))
+        )
+        if type(temperature) == float:
+            temperature = [temperature] * timesteps
+
+        for i in iterator:
+            ts = torch.full((b,), i, device=self.device, dtype=torch.long)
+            if self.shorten_cond_schedule:
+                assert self.model.conditioning_key != "hybrid"
+                tc = self.cond_ids[ts].to(cond.device)
+                cond = self.q_sample(x_start=cond, t=tc, noise=torch.randn_like(cond))
+
+            img, x0_partial = self.p_sample(
+                img,
+                cond,
+                ts,
+                clip_denoised=self.clip_denoised,
+                quantize_denoised=quantize_denoised,
+                return_x0=True,
+                temperature=temperature[i],
+                noise_dropout=noise_dropout,
+                score_corrector=score_corrector,
+                corrector_kwargs=corrector_kwargs,
+            )
+            if mask is not None:
+                assert x0 is not None
+                img_orig = self.q_sample(x0, ts)
+                img = img_orig * mask + (1.0 - mask) * img
+
+            if i % log_every_t == 0 or i == timesteps - 1:
+                intermediates.append(x0_partial)
+            if callback:
+                callback(i)
+            if img_callback:
+                img_callback(img, i)
+        return img, intermediates
+
+    @torch.no_grad()
+    def p_sample_loop(
+        self,
+        cond,
+        shape,
+        return_intermediates=False,
+        x_T=None,
+        verbose=True,
+        callback=None,
+        timesteps=None,
+        quantize_denoised=False,
+        mask=None,
+        x0=None,
+        img_callback=None,
+        start_T=None,
+        log_every_t=None,
+    ):
+
+        if not log_every_t:
+            log_every_t = self.log_every_t
+        device = self.betas.device
+        b = shape[0]
+        if x_T is None:
+            img = torch.randn(shape, device=device)
+        else:
+            img = x_T
+
+        intermediates = [img]
+        if timesteps is None:
+            timesteps = self.num_timesteps
+
+        if start_T is not None:
+            timesteps = min(timesteps, start_T)
+        iterator = (
+            tqdm(reversed(range(0, timesteps)), desc="Sampling t", total=timesteps)
+            if verbose
+            else reversed(range(0, timesteps))
+        )
+
+        if mask is not None:
+            assert x0 is not None
+            assert x0.shape[2:3] == mask.shape[2:3]  # spatial size has to match
+
+        for i in iterator:
+            ts = torch.full((b,), i, device=device, dtype=torch.long)
+            if self.shorten_cond_schedule:
+                assert self.model.conditioning_key != "hybrid"
+                tc = self.cond_ids[ts].to(cond.device)
+                cond = self.q_sample(x_start=cond, t=tc, noise=torch.randn_like(cond))
+
+            img = self.p_sample(
+                img,
+                cond,
+                ts,
+                clip_denoised=self.clip_denoised,
+                quantize_denoised=quantize_denoised,
+            )
+            if mask is not None:
+                img_orig = self.q_sample(x0, ts)
+                img = img_orig * mask + (1.0 - mask) * img
+
+            if i % log_every_t == 0 or i == timesteps - 1:
+                intermediates.append(img)
+            if callback:
+                callback(i)
+            if img_callback:
+                img_callback(img, i)
+
+        if return_intermediates:
+            return img, intermediates
+        return img
+
+    @torch.no_grad()
+    def sample(
+        self,
+        cond,
+        batch_size=16,
+        return_intermediates=False,
+        x_T=None,
+        verbose=True,
+        timesteps=None,
+        quantize_denoised=False,
+        mask=None,
+        x0=None,
+        shape=None,
+        **kwargs,
+    ):
+        if shape is None:
+            shape = (batch_size, self.channels, self.latent_t_size, self.latent_f_size)
+        if cond is not None:
+            if isinstance(cond, dict):
+                cond = {
+                    key: cond[key][:batch_size]
+                    if not isinstance(cond[key], list)
+                    else list(map(lambda x: x[:batch_size], cond[key]))
+                    for key in cond
+                }
+            else:
+                cond = (
+                    [c[:batch_size] for c in cond]
+                    if isinstance(cond, list)
+                    else cond[:batch_size]
+                )
+        return self.p_sample_loop(
+            cond,
+            shape,
+            return_intermediates=return_intermediates,
+            x_T=x_T,
+            verbose=verbose,
+            timesteps=timesteps,
+            quantize_denoised=quantize_denoised,
+            mask=mask,
+            x0=x0,
+            **kwargs,
+        )
+
+    @torch.no_grad()
+    def sample_log(
+        self,
+        cond,
+        batch_size,
+        ddim,
+        ddim_steps,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+        use_plms=False,
+        mask=None,
+        **kwargs,
+    ):
+
+        if mask is not None:
+            shape = (self.channels, mask.size()[-2], mask.size()[-1])
+        else:
+            shape = (self.channels, self.latent_t_size, self.latent_f_size)
+
+        intermediate = None
+        if ddim and not use_plms:
+            # print("Use ddim sampler")
+
+            ddim_sampler = DDIMSampler(self)
+            samples, intermediates = ddim_sampler.sample(
+                ddim_steps,
+                batch_size,
+                shape,
+                cond,
+                verbose=False,
+                unconditional_guidance_scale=unconditional_guidance_scale,
+                unconditional_conditioning=unconditional_conditioning,
+                mask=mask,
+                **kwargs,
+            )
+
+        else:
+            # print("Use DDPM sampler")
+            samples, intermediates = self.sample(
+                cond=cond,
+                batch_size=batch_size,
+                return_intermediates=True,
+                unconditional_guidance_scale=unconditional_guidance_scale,
+                mask=mask,
+                unconditional_conditioning=unconditional_conditioning,
+                **kwargs,
+            )
+
+        return samples, intermediate
+
+    @torch.no_grad()
+    def generate_sample(
+        self,
+        batchs,
+        ddim_steps=200,
+        ddim_eta=1.0,
+        x_T=None,
+        n_candidate_gen_per_text=1,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+        name="waveform",
+        use_plms=False,
+        save=False,
+        **kwargs,
+    ):
+        # Generate n_candidate_gen_per_text times and select the best
+        # Batch: audio, text, fnames
+        assert x_T is None
+        try:
+            batchs = iter(batchs)
+        except TypeError:
+            raise ValueError("The first input argument should be an iterable object")
+
+        if use_plms:
+            assert ddim_steps is not None
+        use_ddim = ddim_steps is not None
+        # waveform_save_path = os.path.join(self.get_log_dir(), name)
+        # os.makedirs(waveform_save_path, exist_ok=True)
+        # print("Waveform save path: ", waveform_save_path)
+
+        with self.ema_scope("Generate"):
+            for batch in batchs:
+                z, c = self.get_input(
+                    batch,
+                    self.first_stage_key,
+                    cond_key=self.cond_stage_key,
+                    return_first_stage_outputs=False,
+                    force_c_encode=True,
+                    return_original_cond=False,
+                    bs=None,
+                )
+                text = super().get_input(batch, "text")
+
+                # Generate multiple samples
+                batch_size = z.shape[0] * n_candidate_gen_per_text
+                c = torch.cat([c] * n_candidate_gen_per_text, dim=0)
+                text = text * n_candidate_gen_per_text
+
+                if unconditional_guidance_scale != 1.0:
+                    unconditional_conditioning = (
+                        self.cond_stage_model.get_unconditional_condition(batch_size)
+                    )
+
+                samples, _ = self.sample_log(
+                    cond=c,
+                    batch_size=batch_size,
+                    x_T=x_T,
+                    ddim=use_ddim,
+                    ddim_steps=ddim_steps,
+                    eta=ddim_eta,
+                    unconditional_guidance_scale=unconditional_guidance_scale,
+                    unconditional_conditioning=unconditional_conditioning,
+                    use_plms=use_plms,
+                )
+                
+                if(torch.max(torch.abs(samples)) > 1e2):
+                    samples = torch.clip(samples, min=-10, max=10)
+                    
+                mel = self.decode_first_stage(samples)
+
+                waveform = self.mel_spectrogram_to_waveform(mel)
+
+                if waveform.shape[0] > 1:
+                    similarity = self.cond_stage_model.cos_similarity(
+                        torch.FloatTensor(waveform).squeeze(1), text
+                    )
+
+                    best_index = []
+                    for i in range(z.shape[0]):
+                        candidates = similarity[i :: z.shape[0]]
+                        max_index = torch.argmax(candidates).item()
+                        best_index.append(i + max_index * z.shape[0])
+
+                    waveform = waveform[best_index]
+                    # print("Similarity between generated audio and text", similarity)
+                    # print("Choose the following indexes:", best_index)
+
+        return waveform
+
+    @torch.no_grad()
+    def generate_sample_masked(
+        self,
+        batchs,
+        ddim_steps=200,
+        ddim_eta=1.0,
+        x_T=None,
+        n_candidate_gen_per_text=1,
+        unconditional_guidance_scale=1.0,
+        unconditional_conditioning=None,
+        name="waveform",
+        use_plms=False,
+        time_mask_ratio_start_and_end=(0.25, 0.75),
+        freq_mask_ratio_start_and_end=(0.75, 1.0),
+        save=False,
+        **kwargs,
+    ):
+        # Generate n_candidate_gen_per_text times and select the best
+        # Batch: audio, text, fnames
+        assert x_T is None
+        try:
+            batchs = iter(batchs)
+        except TypeError:
+            raise ValueError("The first input argument should be an iterable object")
+
+        if use_plms:
+            assert ddim_steps is not None
+        use_ddim = ddim_steps is not None
+        # waveform_save_path = os.path.join(self.get_log_dir(), name)
+        # os.makedirs(waveform_save_path, exist_ok=True)
+        # print("Waveform save path: ", waveform_save_path)
+
+        with self.ema_scope("Generate"):
+            for batch in batchs:
+                z, c = self.get_input(
+                    batch,
+                    self.first_stage_key,
+                    cond_key=self.cond_stage_key,
+                    return_first_stage_outputs=False,
+                    force_c_encode=True,
+                    return_original_cond=False,
+                    bs=None,
+                )
+                text = super().get_input(batch, "text")
+                
+                # Generate multiple samples
+                batch_size = z.shape[0] * n_candidate_gen_per_text
+                
+                _, h, w = z.shape[0], z.shape[2], z.shape[3]
+                
+                mask = torch.ones(batch_size, h, w).to(self.device)
+                
+                mask[:, int(h * time_mask_ratio_start_and_end[0]) : int(h * time_mask_ratio_start_and_end[1]), :] = 0 
+                mask[:, :, int(w * freq_mask_ratio_start_and_end[0]) : int(w * freq_mask_ratio_start_and_end[1])] = 0 
+                mask = mask[:, None, ...]
+                
+                c = torch.cat([c] * n_candidate_gen_per_text, dim=0)
+                text = text * n_candidate_gen_per_text
+
+                if unconditional_guidance_scale != 1.0:
+                    unconditional_conditioning = (
+                        self.cond_stage_model.get_unconditional_condition(batch_size)
+                    )
+
+                samples, _ = self.sample_log(
+                    cond=c,
+                    batch_size=batch_size,
+                    x_T=x_T,
+                    ddim=use_ddim,
+                    ddim_steps=ddim_steps,
+                    eta=ddim_eta,
+                    unconditional_guidance_scale=unconditional_guidance_scale,
+                    unconditional_conditioning=unconditional_conditioning,
+                    use_plms=use_plms, mask=mask, x0=torch.cat([z] * n_candidate_gen_per_text)
+                )
+
+                mel = self.decode_first_stage(samples)
+
+                waveform = self.mel_spectrogram_to_waveform(mel)
+
+                if waveform.shape[0] > 1:
+                    similarity = self.cond_stage_model.cos_similarity(
+                        torch.FloatTensor(waveform).squeeze(1), text
+                    )
+
+                    best_index = []
+                    for i in range(z.shape[0]):
+                        candidates = similarity[i :: z.shape[0]]
+                        max_index = torch.argmax(candidates).item()
+                        best_index.append(i + max_index * z.shape[0])
+
+                    waveform = waveform[best_index]
+                    # print("Similarity between generated audio and text", similarity)
+                    # print("Choose the following indexes:", best_index)
+
+        return waveform
\ No newline at end of file
diff --git a/picoaudio/audioldm/pipeline.py b/picoaudio/audioldm/pipeline.py
new file mode 100644
index 0000000000000000000000000000000000000000..b08e1f77206483025ce027588c2dea1de78ae26c
--- /dev/null
+++ b/picoaudio/audioldm/pipeline.py
@@ -0,0 +1,301 @@
+import os
+
+import argparse
+import yaml
+import torch
+from torch import autocast
+from tqdm import tqdm, trange
+
+from audioldm import LatentDiffusion, seed_everything
+from audioldm.utils import default_audioldm_config, get_duration, get_bit_depth, get_metadata, download_checkpoint
+from audioldm.audio import wav_to_fbank, TacotronSTFT, read_wav_file
+from audioldm.latent_diffusion.ddim import DDIMSampler
+from einops import repeat
+import os
+
+def make_batch_for_text_to_audio(text, waveform=None, fbank=None, batchsize=1):
+    text = [text] * batchsize
+    if batchsize < 1:
+        print("Warning: Batchsize must be at least 1. Batchsize is set to .")
+    
+    if(fbank is None):
+        fbank = torch.zeros((batchsize, 1024, 64))  # Not used, here to keep the code format
+    else:
+        fbank = torch.FloatTensor(fbank)
+        fbank = fbank.expand(batchsize, 1024, 64)
+        assert fbank.size(0) == batchsize
+        
+    stft = torch.zeros((batchsize, 1024, 512))  # Not used
+
+    if(waveform is None):
+        waveform = torch.zeros((batchsize, 160000))  # Not used
+    else:
+        waveform = torch.FloatTensor(waveform)
+        waveform = waveform.expand(batchsize, -1)
+        assert waveform.size(0) == batchsize
+        
+    fname = [""] * batchsize  # Not used
+    
+    batch = (
+        fbank,
+        stft,
+        None,
+        fname,
+        waveform,
+        text,
+    )
+    return batch
+
+def round_up_duration(duration):
+    return int(round(duration/2.5) + 1) * 2.5
+
+def build_model(
+    ckpt_path=None,
+    config=None,
+    model_name="audioldm-s-full"
+):
+    print("Load AudioLDM: %s", model_name)
+    
+    if(ckpt_path is None):
+        ckpt_path = get_metadata()[model_name]["path"]
+    
+    if(not os.path.exists(ckpt_path)):
+        download_checkpoint(model_name)
+
+    if torch.cuda.is_available():
+        device = torch.device("cuda:0")
+    else:
+        device = torch.device("cpu")
+
+    if config is not None:
+        assert type(config) is str
+        config = yaml.load(open(config, "r"), Loader=yaml.FullLoader)
+    else:
+        config = default_audioldm_config(model_name)
+
+    # Use text as condition instead of using waveform during training
+    config["model"]["params"]["device"] = device
+    config["model"]["params"]["cond_stage_key"] = "text"
+
+    # No normalization here
+    latent_diffusion = LatentDiffusion(**config["model"]["params"])
+
+    resume_from_checkpoint = ckpt_path
+
+    checkpoint = torch.load(resume_from_checkpoint, map_location=device)
+    latent_diffusion.load_state_dict(checkpoint["state_dict"])
+
+    latent_diffusion.eval()
+    latent_diffusion = latent_diffusion.to(device)
+
+    latent_diffusion.cond_stage_model.embed_mode = "text"
+    return latent_diffusion
+
+def duration_to_latent_t_size(duration):
+    return int(duration * 25.6)
+
+def set_cond_audio(latent_diffusion):
+    latent_diffusion.cond_stage_key = "waveform"
+    latent_diffusion.cond_stage_model.embed_mode="audio"
+    return latent_diffusion
+
+def set_cond_text(latent_diffusion):
+    latent_diffusion.cond_stage_key = "text"
+    latent_diffusion.cond_stage_model.embed_mode="text"
+    return latent_diffusion
+    
+def text_to_audio(
+    latent_diffusion,
+    text,
+    original_audio_file_path = None,
+    seed=42,
+    ddim_steps=200,
+    duration=10,
+    batchsize=1,
+    guidance_scale=2.5,
+    n_candidate_gen_per_text=3,
+    config=None,
+):
+    seed_everything(int(seed))
+    waveform = None
+    if(original_audio_file_path is not None):
+        waveform = read_wav_file(original_audio_file_path, int(duration * 102.4) * 160)
+        
+    batch = make_batch_for_text_to_audio(text, waveform=waveform, batchsize=batchsize)
+
+    latent_diffusion.latent_t_size = duration_to_latent_t_size(duration)
+    
+    if(waveform is not None):
+        print("Generate audio that has similar content as %s" % original_audio_file_path)
+        latent_diffusion = set_cond_audio(latent_diffusion)
+    else:
+        print("Generate audio using text %s" % text)
+        latent_diffusion = set_cond_text(latent_diffusion)
+        
+    with torch.no_grad():
+        waveform = latent_diffusion.generate_sample(
+            [batch],
+            unconditional_guidance_scale=guidance_scale,
+            ddim_steps=ddim_steps,
+            n_candidate_gen_per_text=n_candidate_gen_per_text,
+            duration=duration,
+        )
+    return waveform
+
+def style_transfer(
+    latent_diffusion,
+    text,
+    original_audio_file_path,
+    transfer_strength,
+    seed=42,
+    duration=10,
+    batchsize=1,
+    guidance_scale=2.5,
+    ddim_steps=200,
+    config=None,
+):
+    if torch.cuda.is_available():
+        device = torch.device("cuda:0")
+    else:
+        device = torch.device("cpu")
+
+    assert original_audio_file_path is not None, "You need to provide the original audio file path"
+    
+    audio_file_duration = get_duration(original_audio_file_path)
+    
+    assert get_bit_depth(original_audio_file_path) == 16, "The bit depth of the original audio file %s must be 16" % original_audio_file_path
+    
+    # if(duration > 20):
+    #     print("Warning: The duration of the audio file %s must be less than 20 seconds. Longer duration will result in Nan in model output (we are still debugging that); Automatically set duration to 20 seconds")
+    #     duration = 20
+    
+    if(duration >= audio_file_duration):
+        print("Warning: Duration you specified %s-seconds must equal or smaller than the audio file duration %ss" % (duration, audio_file_duration))
+        duration = round_up_duration(audio_file_duration)
+        print("Set new duration as %s-seconds" % duration)
+
+    # duration = round_up_duration(duration)
+    
+    latent_diffusion = set_cond_text(latent_diffusion)
+
+    if config is not None:
+        assert type(config) is str
+        config = yaml.load(open(config, "r"), Loader=yaml.FullLoader)
+    else:
+        config = default_audioldm_config()
+
+    seed_everything(int(seed))
+    # latent_diffusion.latent_t_size = duration_to_latent_t_size(duration)
+    latent_diffusion.cond_stage_model.embed_mode = "text"
+
+    fn_STFT = TacotronSTFT(
+        config["preprocessing"]["stft"]["filter_length"],
+        config["preprocessing"]["stft"]["hop_length"],
+        config["preprocessing"]["stft"]["win_length"],
+        config["preprocessing"]["mel"]["n_mel_channels"],
+        config["preprocessing"]["audio"]["sampling_rate"],
+        config["preprocessing"]["mel"]["mel_fmin"],
+        config["preprocessing"]["mel"]["mel_fmax"],
+    )
+
+    mel, _, _ = wav_to_fbank(
+        original_audio_file_path, target_length=int(duration * 102.4), fn_STFT=fn_STFT
+    )
+    mel = mel.unsqueeze(0).unsqueeze(0).to(device)
+    mel = repeat(mel, "1 ... -> b ...", b=batchsize)
+    init_latent = latent_diffusion.get_first_stage_encoding(
+        latent_diffusion.encode_first_stage(mel)
+    )  # move to latent space, encode and sample
+    if(torch.max(torch.abs(init_latent)) > 1e2):
+        init_latent = torch.clip(init_latent, min=-10, max=10)
+    sampler = DDIMSampler(latent_diffusion)
+    sampler.make_schedule(ddim_num_steps=ddim_steps, ddim_eta=1.0, verbose=False)
+
+    t_enc = int(transfer_strength * ddim_steps)
+    prompts = text
+
+    with torch.no_grad():
+        with autocast("cuda"):
+            with latent_diffusion.ema_scope():
+                uc = None
+                if guidance_scale != 1.0:
+                    uc = latent_diffusion.cond_stage_model.get_unconditional_condition(
+                        batchsize
+                    )
+
+                c = latent_diffusion.get_learned_conditioning([prompts] * batchsize)
+                z_enc = sampler.stochastic_encode(
+                    init_latent, torch.tensor([t_enc] * batchsize).to(device)
+                )
+                samples = sampler.decode(
+                    z_enc,
+                    c,
+                    t_enc,
+                    unconditional_guidance_scale=guidance_scale,
+                    unconditional_conditioning=uc,
+                )
+                # x_samples = latent_diffusion.decode_first_stage(samples) # Will result in Nan in output
+                # print(torch.sum(torch.isnan(samples)))
+                x_samples = latent_diffusion.decode_first_stage(samples)
+                # print(x_samples)
+                x_samples = latent_diffusion.decode_first_stage(samples[:,:,:-3,:])
+                # print(x_samples)
+                waveform = latent_diffusion.first_stage_model.decode_to_waveform(
+                    x_samples
+                )
+
+    return waveform
+
+def super_resolution_and_inpainting(
+    latent_diffusion,
+    text,
+    original_audio_file_path = None,
+    seed=42,
+    ddim_steps=200,
+    duration=None,
+    batchsize=1,
+    guidance_scale=2.5,
+    n_candidate_gen_per_text=3,
+    time_mask_ratio_start_and_end=(0.10, 0.15), # regenerate the 10% to 15% of the time steps in the spectrogram
+    # time_mask_ratio_start_and_end=(1.0, 1.0), # no inpainting
+    # freq_mask_ratio_start_and_end=(0.75, 1.0), # regenerate the higher 75% to 100% mel bins
+    freq_mask_ratio_start_and_end=(1.0, 1.0), # no super-resolution
+    config=None,
+):
+    seed_everything(int(seed))
+    if config is not None:
+        assert type(config) is str
+        config = yaml.load(open(config, "r"), Loader=yaml.FullLoader)
+    else:
+        config = default_audioldm_config()
+    fn_STFT = TacotronSTFT(
+        config["preprocessing"]["stft"]["filter_length"],
+        config["preprocessing"]["stft"]["hop_length"],
+        config["preprocessing"]["stft"]["win_length"],
+        config["preprocessing"]["mel"]["n_mel_channels"],
+        config["preprocessing"]["audio"]["sampling_rate"],
+        config["preprocessing"]["mel"]["mel_fmin"],
+        config["preprocessing"]["mel"]["mel_fmax"],
+    )
+    
+    # waveform = read_wav_file(original_audio_file_path, None)
+    mel, _, _ = wav_to_fbank(
+        original_audio_file_path, target_length=int(duration * 102.4), fn_STFT=fn_STFT
+    )
+    
+    batch = make_batch_for_text_to_audio(text, fbank=mel[None,...], batchsize=batchsize)
+        
+    # latent_diffusion.latent_t_size = duration_to_latent_t_size(duration)
+    latent_diffusion = set_cond_text(latent_diffusion)
+        
+    with torch.no_grad():
+        waveform = latent_diffusion.generate_sample_masked(
+            [batch],
+            unconditional_guidance_scale=guidance_scale,
+            ddim_steps=ddim_steps,
+            n_candidate_gen_per_text=n_candidate_gen_per_text,
+            duration=duration,
+            time_mask_ratio_start_and_end=time_mask_ratio_start_and_end,
+            freq_mask_ratio_start_and_end=freq_mask_ratio_start_and_end
+        )
+    return waveform
\ No newline at end of file
diff --git a/picoaudio/audioldm/utils.py b/picoaudio/audioldm/utils.py
new file mode 100644
index 0000000000000000000000000000000000000000..5401b29d4366774233f1bf4a9e7fcb7ce214187e
--- /dev/null
+++ b/picoaudio/audioldm/utils.py
@@ -0,0 +1,281 @@
+import contextlib
+import importlib
+
+from inspect import isfunction
+import os
+import soundfile as sf
+import time
+import wave
+
+import urllib.request
+import progressbar
+
+CACHE_DIR = os.getenv(
+    "AUDIOLDM_CACHE_DIR",
+    os.path.join(os.path.expanduser("~"), ".cache/audioldm"))
+
+def get_duration(fname):
+    with contextlib.closing(wave.open(fname, 'r')) as f:
+        frames = f.getnframes()
+        rate = f.getframerate()
+        return frames / float(rate)
+    
+def get_bit_depth(fname):
+    with contextlib.closing(wave.open(fname, 'r')) as f:
+        bit_depth = f.getsampwidth() * 8
+        return bit_depth
+       
+def get_time():
+    t = time.localtime()
+    return time.strftime("%d_%m_%Y_%H_%M_%S", t)
+
+def seed_everything(seed):
+    import random, os
+    import numpy as np
+    import torch
+
+    random.seed(seed)
+    os.environ["PYTHONHASHSEED"] = str(seed)
+    np.random.seed(seed)
+    torch.manual_seed(seed)
+    torch.cuda.manual_seed(seed)
+    torch.backends.cudnn.deterministic = True
+    torch.backends.cudnn.benchmark = True
+
+
+def save_wave(waveform, savepath, name="outwav"):
+    if type(name) is not list:
+        name = [name] * waveform.shape[0]
+
+    for i in range(waveform.shape[0]):
+        path = os.path.join(
+            savepath,
+            "%s_%s.wav"
+            % (
+                os.path.basename(name[i])
+                if (not ".wav" in name[i])
+                else os.path.basename(name[i]).split(".")[0],
+                i,
+            ),
+        )
+        print("Save audio to %s" % path)
+        sf.write(path, waveform[i, 0], samplerate=16000)
+
+
+def exists(x):
+    return x is not None
+
+
+def default(val, d):
+    if exists(val):
+        return val
+    return d() if isfunction(d) else d
+
+
+def count_params(model, verbose=False):
+    total_params = sum(p.numel() for p in model.parameters())
+    if verbose:
+        print(f"{model.__class__.__name__} has {total_params * 1.e-6:.2f} M params.")
+    return total_params
+
+
+def get_obj_from_str(string, reload=False):
+    module, cls = string.rsplit(".", 1)
+    if reload:
+        module_imp = importlib.import_module(module)
+        importlib.reload(module_imp)
+    return getattr(importlib.import_module(module, package=None), cls)
+
+
+def instantiate_from_config(config):
+    if not "target" in config:
+        if config == "__is_first_stage__":
+            return None
+        elif config == "__is_unconditional__":
+            return None
+        raise KeyError("Expected key `target` to instantiate.")
+    return get_obj_from_str(config["target"])(**config.get("params", dict()))
+
+
+def default_audioldm_config(model_name="audioldm-s-full"):    
+    basic_config = {
+        "wave_file_save_path": "./output",
+        "id": {
+            "version": "v1",
+            "name": "default",
+            "root": "/mnt/fast/nobackup/users/hl01486/projects/general_audio_generation/AudioLDM-python/config/default/latent_diffusion.yaml",
+        },
+        "preprocessing": {
+            "audio": {"sampling_rate": 16000, "max_wav_value": 32768},
+            "stft": {"filter_length": 1024, "hop_length": 160, "win_length": 1024},
+            "mel": {
+                "n_mel_channels": 64,
+                "mel_fmin": 0,
+                "mel_fmax": 8000,
+                "freqm": 0,
+                "timem": 0,
+                "blur": False,
+                "mean": -4.63,
+                "std": 2.74,
+                "target_length": 1024,
+            },
+        },
+        "model": {
+            "device": "cuda",
+            "target": "audioldm.pipline.LatentDiffusion",
+            "params": {
+                "base_learning_rate": 5e-06,
+                "linear_start": 0.0015,
+                "linear_end": 0.0195,
+                "num_timesteps_cond": 1,
+                "log_every_t": 200,
+                "timesteps": 1000,
+                "first_stage_key": "fbank",
+                "cond_stage_key": "waveform",
+                "latent_t_size": 256,
+                "latent_f_size": 16,
+                "channels": 8,
+                "cond_stage_trainable": True,
+                "conditioning_key": "film",
+                "monitor": "val/loss_simple_ema",
+                "scale_by_std": True,
+                "unet_config": {
+                    "target": "audioldm.latent_diffusion.openaimodel.UNetModel",
+                    "params": {
+                        "image_size": 64,
+                        "extra_film_condition_dim": 512,
+                        "extra_film_use_concat": True,
+                        "in_channels": 8,
+                        "out_channels": 8,
+                        "model_channels": 128,
+                        "attention_resolutions": [8, 4, 2],
+                        "num_res_blocks": 2,
+                        "channel_mult": [1, 2, 3, 5],
+                        "num_head_channels": 32,
+                        "use_spatial_transformer": True,
+                    },
+                },
+                "first_stage_config": {
+                    "base_learning_rate": 4.5e-05,
+                    "target": "audioldm.variational_autoencoder.autoencoder.AutoencoderKL",
+                    "params": {
+                        "monitor": "val/rec_loss",
+                        "image_key": "fbank",
+                        "subband": 1,
+                        "embed_dim": 8,
+                        "time_shuffle": 1,
+                        "ddconfig": {
+                            "double_z": True,
+                            "z_channels": 8,
+                            "resolution": 256,
+                            "downsample_time": False,
+                            "in_channels": 1,
+                            "out_ch": 1,
+                            "ch": 128,
+                            "ch_mult": [1, 2, 4],
+                            "num_res_blocks": 2,
+                            "attn_resolutions": [],
+                            "dropout": 0.0,
+                        },
+                    },
+                },
+                "cond_stage_config": {
+                    "target": "audioldm.clap.encoders.CLAPAudioEmbeddingClassifierFreev2",
+                    "params": {
+                        "key": "waveform",
+                        "sampling_rate": 16000,
+                        "embed_mode": "audio",
+                        "unconditional_prob": 0.1,
+                    },
+                },
+            },
+        },
+    }
+    
+    if("-l-" in model_name):
+        basic_config["model"]["params"]["unet_config"]["params"]["model_channels"] = 256
+        basic_config["model"]["params"]["unet_config"]["params"]["num_head_channels"] = 64
+    elif("-m-" in model_name):
+        basic_config["model"]["params"]["unet_config"]["params"]["model_channels"] = 192
+        basic_config["model"]["params"]["cond_stage_config"]["params"]["amodel"] = "HTSAT-base" # This model use a larger HTAST
+        
+    return basic_config
+        
+def get_metadata():
+    return {
+        "audioldm-s-full": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-s-full.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7600541/files/audioldm-s-full?download=1",
+        },
+        "audioldm-l-full": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-l-full.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7698295/files/audioldm-full-l.ckpt?download=1",
+        },
+        "audioldm-s-full-v2": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-s-full-v2.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7698295/files/audioldm-full-s-v2.ckpt?download=1",
+        },
+        "audioldm-m-text-ft": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-m-text-ft.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7813012/files/audioldm-m-text-ft.ckpt?download=1",
+        },
+        "audioldm-s-text-ft": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-s-text-ft.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7813012/files/audioldm-s-text-ft.ckpt?download=1",
+        },
+        "audioldm-m-full": {
+            "path": os.path.join(
+                CACHE_DIR,
+                "audioldm-m-full.ckpt",
+            ),
+            "url": "https://zenodo.org/record/7813012/files/audioldm-m-full.ckpt?download=1",
+        },
+    }
+    
+class MyProgressBar():
+    def __init__(self):
+        self.pbar = None
+
+    def __call__(self, block_num, block_size, total_size):
+        if not self.pbar:
+            self.pbar=progressbar.ProgressBar(maxval=total_size)
+            self.pbar.start()
+
+        downloaded = block_num * block_size
+        if downloaded < total_size:
+            self.pbar.update(downloaded)
+        else:
+            self.pbar.finish()
+            
+def download_checkpoint(checkpoint_name="audioldm-s-full"):
+    meta = get_metadata()
+    if(checkpoint_name not in meta.keys()):
+        print("The model name you provided is not supported. Please use one of the following: ", meta.keys())
+
+    if not os.path.exists(meta[checkpoint_name]["path"]) or os.path.getsize(meta[checkpoint_name]["path"]) < 2*10**9:
+        os.makedirs(os.path.dirname(meta[checkpoint_name]["path"]), exist_ok=True)
+        print(f"Downloading the main structure of {checkpoint_name} into {os.path.dirname(meta[checkpoint_name]['path'])}")
+
+        urllib.request.urlretrieve(meta[checkpoint_name]["url"], meta[checkpoint_name]["path"], MyProgressBar())
+        print(
+            "Weights downloaded in: {} Size: {}".format(
+                meta[checkpoint_name]["path"],
+                os.path.getsize(meta[checkpoint_name]["path"]),
+            )
+        )
+    
\ No newline at end of file
diff --git a/picoaudio/audioldm/variational_autoencoder/__init__.py b/picoaudio/audioldm/variational_autoencoder/__init__.py
new file mode 100644
index 0000000000000000000000000000000000000000..08b2a9b9698e02918d7b0dd9fe0431b2847e5aa2
--- /dev/null
+++ b/picoaudio/audioldm/variational_autoencoder/__init__.py
@@ -0,0 +1 @@
+from .autoencoder import AutoencoderKL
\ No newline at end of file
diff --git a/picoaudio/audioldm/variational_autoencoder/autoencoder.py b/picoaudio/audioldm/variational_autoencoder/autoencoder.py
new file mode 100644
index 0000000000000000000000000000000000000000..9dadc849da65d1f9eb82dc75dc777250bf738151
--- /dev/null
+++ b/picoaudio/audioldm/variational_autoencoder/autoencoder.py
@@ -0,0 +1,135 @@
+import torch
+from audioldm.latent_diffusion.ema import *
+from audioldm.variational_autoencoder.modules import Encoder, Decoder
+from audioldm.variational_autoencoder.distributions import DiagonalGaussianDistribution
+
+from audioldm.hifigan.utilities import get_vocoder, vocoder_infer
+
+
+class AutoencoderKL(nn.Module):
+    def __init__(
+        self,
+        ddconfig=None,
+        lossconfig=None,
+        image_key="fbank",
+        embed_dim=None,
+        time_shuffle=1,
+        subband=1,
+        ckpt_path=None,
+        reload_from_ckpt=None,
+        ignore_keys=[],
+        colorize_nlabels=None,
+        monitor=None,
+        base_learning_rate=1e-5,
+        scale_factor=1
+    ):
+        super().__init__()
+
+        self.encoder = Encoder(**ddconfig)
+        self.decoder = Decoder(**ddconfig)
+
+        self.subband = int(subband)
+
+        if self.subband > 1:
+            print("Use subband decomposition %s" % self.subband)
+
+        self.quant_conv = torch.nn.Conv2d(2 * ddconfig["z_channels"], 2 * embed_dim, 1)
+        self.post_quant_conv = torch.nn.Conv2d(embed_dim, ddconfig["z_channels"], 1)
+
+        self.vocoder = get_vocoder(None, "cpu")
+        self.embed_dim = embed_dim
+
+        if monitor is not None:
+            self.monitor = monitor
+
+        self.time_shuffle = time_shuffle
+        self.reload_from_ckpt = reload_from_ckpt
+        self.reloaded = False
+        self.mean, self.std = None, None
+        
+        self.scale_factor = scale_factor
+
+    def encode(self, x):
+        # x = self.time_shuffle_operation(x)
+        x = self.freq_split_subband(x)
+        h = self.encoder(x)
+        moments = self.quant_conv(h)
+        posterior = DiagonalGaussianDistribution(moments)
+        return posterior
+
+    def decode(self, z):
+        z = self.post_quant_conv(z)
+        dec = self.decoder(z)
+        dec = self.freq_merge_subband(dec)
+        return dec
+
+    def decode_to_waveform(self, dec):
+        dec = dec.squeeze(1).permute(0, 2, 1)
+        wav_reconstruction = vocoder_infer(dec, self.vocoder)
+        return wav_reconstruction
+
+    def forward(self, input, sample_posterior=True):
+        posterior = self.encode(input)
+        if sample_posterior:
+            z = posterior.sample()
+        else:
+            z = posterior.mode()
+
+        if self.flag_first_run:
+            print("Latent size: ", z.size())
+            self.flag_first_run = False
+
+        dec = self.decode(z)
+
+        return dec, posterior
+
+    def freq_split_subband(self, fbank):
+        if self.subband == 1 or self.image_key != "stft":
+            return fbank
+
+        bs, ch, tstep, fbins = fbank.size()
+
+        assert fbank.size(-1) % self.subband == 0
+        assert ch == 1
+
+        return (
+            fbank.squeeze(1)
+            .reshape(bs, tstep, self.subband, fbins // self.subband)
+            .permute(0, 2, 1, 3)
+        )
+
+    def freq_merge_subband(self, subband_fbank):
+        if self.subband == 1 or self.image_key != "stft":
+            return subband_fbank
+        assert subband_fbank.size(1) == self.subband  # Channel dimension
+        bs, sub_ch, tstep, fbins = subband_fbank.size()
+        return subband_fbank.permute(0, 2, 1, 3).reshape(bs, tstep, -1).unsqueeze(1)
+    
+    def device(self):
+        return next(self.parameters()).device
+    
+    @torch.no_grad()
+    def encode_first_stage(self, x):
+        return self.encode(x)
+    
+    @torch.no_grad()
+    def decode_first_stage(self, z, predict_cids=False, force_not_quantize=False):
+        if predict_cids:
+            if z.dim() == 4:
+                z = torch.argmax(z.exp(), dim=1).long()
+            z = self.first_stage_model.quantize.get_codebook_entry(z, shape=None)
+            z = rearrange(z, "b h w c -> b c h w").contiguous()
+
+        z = 1.0 / self.scale_factor * z
+        return self.decode(z)
+
+    def get_first_stage_encoding(self, encoder_posterior):
+        if isinstance(encoder_posterior, DiagonalGaussianDistribution):
+            z = encoder_posterior.sample()
+        elif isinstance(encoder_posterior, torch.Tensor):
+            z = encoder_posterior
+        else:
+            raise NotImplementedError(
+                f"encoder_posterior of type '{type(encoder_posterior)}' not yet implemented"
+            )
+        return self.scale_factor * z
\ No newline at end of file
diff --git a/picoaudio/audioldm/variational_autoencoder/distributions.py b/picoaudio/audioldm/variational_autoencoder/distributions.py
new file mode 100644
index 0000000000000000000000000000000000000000..58eb535e7769f402169ddff77ee45c96ba3650d9
--- /dev/null
+++ b/picoaudio/audioldm/variational_autoencoder/distributions.py
@@ -0,0 +1,102 @@
+import torch
+import numpy as np
+
+
+class AbstractDistribution:
+    def sample(self):
+        raise NotImplementedError()
+
+    def mode(self):
+        raise NotImplementedError()
+
+
+class DiracDistribution(AbstractDistribution):
+    def __init__(self, value):
+        self.value = value
+
+    def sample(self):
+        return self.value
+
+    def mode(self):
+        return self.value
+
+
+class DiagonalGaussianDistribution(object):
+    def __init__(self, parameters, deterministic=False):
+        self.parameters = parameters
+        self.mean, self.logvar = torch.chunk(parameters, 2, dim=1)
+        self.logvar = torch.clamp(self.logvar, -30.0, 20.0)
+        self.deterministic = deterministic
+        self.std = torch.exp(0.5 * self.logvar)
+        self.var = torch.exp(self.logvar)
+        if self.deterministic:
+            self.var = self.std = torch.zeros_like(self.mean).to(
+                device=self.parameters.device
+            )
+
+    def sample(self):
+        x = self.mean + self.std * torch.randn(self.mean.shape).to(
+            device=self.parameters.device
+        )
+        return x
+
+    def kl(self, other=None):
+        if self.deterministic:
+            return torch.Tensor([0.0])
+        else:
+            if other is None:
+                return 0.5 * torch.mean(
+                    torch.pow(self.mean, 2) + self.var - 1.0 - self.logvar,
+                    dim=[1, 2, 3],
+                )
+            else:
+                return 0.5 * torch.mean(
+                    torch.pow(self.mean - other.mean, 2) / other.var
+                    + self.var / other.var
+                    - 1.0
+                    - self.logvar
+                    + other.logvar,
+                    dim=[1, 2, 3],
+                )
+
+    def nll(self, sample, dims=[1, 2, 3]):
+        if self.deterministic:
+            return torch.Tensor([0.0])
+        logtwopi = np.log(2.0 * np.pi)
+        return 0.5 * torch.sum(
+            logtwopi + self.logvar + torch.pow(sample - self.mean, 2) / self.var,
+            dim=dims,
+        )
+
+    def mode(self):
+        return self.mean
+
+
+def normal_kl(mean1, logvar1, mean2, logvar2):
+    """
+    source: https://github.com/openai/guided-diffusion/blob/27c20a8fab9cb472df5d6bdd6c8d11c8f430b924/guided_diffusion/losses.py#L12
+    Compute the KL divergence between two gaussians.
+    Shapes are automatically broadcasted, so batches can be compared to
+    scalars, among other use cases.
+    """
+    tensor = None
+    for obj in (mean1, logvar1, mean2, logvar2):
+        if isinstance(obj, torch.Tensor):
+            tensor = obj
+            break
+    assert tensor is not None, "at least one argument must be a Tensor"
+
+    # Force variances to be Tensors. Broadcasting helps convert scalars to
+    # Tensors, but it does not work for torch.exp().
+    logvar1, logvar2 = [
+        x if isinstance(x, torch.Tensor) else torch.tensor(x).to(tensor)
+        for x in (logvar1, logvar2)
+    ]
+
+    return 0.5 * (
+        -1.0
+        + logvar2
+        - logvar1
+        + torch.exp(logvar1 - logvar2)
+        + ((mean1 - mean2) ** 2) * torch.exp(-logvar2)
+    )
diff --git a/picoaudio/audioldm/variational_autoencoder/modules.py b/picoaudio/audioldm/variational_autoencoder/modules.py
new file mode 100644
index 0000000000000000000000000000000000000000..e48386d045c1d0e159de33db02af1035159c3447
--- /dev/null
+++ b/picoaudio/audioldm/variational_autoencoder/modules.py
@@ -0,0 +1,1066 @@
+# pytorch_diffusion + derived encoder decoder
+import math
+import torch
+import torch.nn as nn
+import numpy as np
+from einops import rearrange
+
+from audioldm.utils import instantiate_from_config
+from audioldm.latent_diffusion.attention import LinearAttention
+
+
+def get_timestep_embedding(timesteps, embedding_dim):
+    """
+    This matches the implementation in Denoising Diffusion Probabilistic Models:
+    From Fairseq.
+    Build sinusoidal embeddings.
+    This matches the implementation in tensor2tensor, but differs slightly
+    from the description in Section 3.5 of "Attention Is All You Need".
+    """
+    assert len(timesteps.shape) == 1
+
+    half_dim = embedding_dim // 2
+    emb = math.log(10000) / (half_dim - 1)
+    emb = torch.exp(torch.arange(half_dim, dtype=torch.float32) * -emb)
+    emb = emb.to(device=timesteps.device)
+    emb = timesteps.float()[:, None] * emb[None, :]
+    emb = torch.cat([torch.sin(emb), torch.cos(emb)], dim=1)
+    if embedding_dim % 2 == 1:  # zero pad
+        emb = torch.nn.functional.pad(emb, (0, 1, 0, 0))
+    return emb
+
+
+def nonlinearity(x):
+    # swish
+    return x * torch.sigmoid(x)
+
+
+def Normalize(in_channels, num_groups=32):
+    return torch.nn.GroupNorm(
+        num_groups=num_groups, num_channels=in_channels, eps=1e-6, affine=True
+    )
+
+
+class Upsample(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            self.conv = torch.nn.Conv2d(
+                in_channels, in_channels, kernel_size=3, stride=1, padding=1
+            )
+
+    def forward(self, x):
+        x = torch.nn.functional.interpolate(x, scale_factor=2.0, mode="nearest")
+        if self.with_conv:
+            x = self.conv(x)
+        return x
+
+
+class UpsampleTimeStride4(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            self.conv = torch.nn.Conv2d(
+                in_channels, in_channels, kernel_size=5, stride=1, padding=2
+            )
+
+    def forward(self, x):
+        x = torch.nn.functional.interpolate(x, scale_factor=(4.0, 2.0), mode="nearest")
+        if self.with_conv:
+            x = self.conv(x)
+        return x
+
+
+class Downsample(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            # Do time downsampling here
+            # no asymmetric padding in torch conv, must do it ourselves
+            self.conv = torch.nn.Conv2d(
+                in_channels, in_channels, kernel_size=3, stride=2, padding=0
+            )
+
+    def forward(self, x):
+        if self.with_conv:
+            pad = (0, 1, 0, 1)
+            x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
+            x = self.conv(x)
+        else:
+            x = torch.nn.functional.avg_pool2d(x, kernel_size=2, stride=2)
+        return x
+
+
+class DownsampleTimeStride4(nn.Module):
+    def __init__(self, in_channels, with_conv):
+        super().__init__()
+        self.with_conv = with_conv
+        if self.with_conv:
+            # Do time downsampling here
+            # no asymmetric padding in torch conv, must do it ourselves
+            self.conv = torch.nn.Conv2d(
+                in_channels, in_channels, kernel_size=5, stride=(4, 2), padding=1
+            )
+
+    def forward(self, x):
+        if self.with_conv:
+            pad = (0, 1, 0, 1)
+            x = torch.nn.functional.pad(x, pad, mode="constant", value=0)
+            x = self.conv(x)
+        else:
+            x = torch.nn.functional.avg_pool2d(x, kernel_size=(4, 2), stride=(4, 2))
+        return x
+
+
+class ResnetBlock(nn.Module):
+    def __init__(
+        self,
+        *,
+        in_channels,
+        out_channels=None,
+        conv_shortcut=False,
+        dropout,
+        temb_channels=512,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        out_channels = in_channels if out_channels is None else out_channels
+        self.out_channels = out_channels
+        self.use_conv_shortcut = conv_shortcut
+
+        self.norm1 = Normalize(in_channels)
+        self.conv1 = torch.nn.Conv2d(
+            in_channels, out_channels, kernel_size=3, stride=1, padding=1
+        )
+        if temb_channels > 0:
+            self.temb_proj = torch.nn.Linear(temb_channels, out_channels)
+        self.norm2 = Normalize(out_channels)
+        self.dropout = torch.nn.Dropout(dropout)
+        self.conv2 = torch.nn.Conv2d(
+            out_channels, out_channels, kernel_size=3, stride=1, padding=1
+        )
+        if self.in_channels != self.out_channels:
+            if self.use_conv_shortcut:
+                self.conv_shortcut = torch.nn.Conv2d(
+                    in_channels, out_channels, kernel_size=3, stride=1, padding=1
+                )
+            else:
+                self.nin_shortcut = torch.nn.Conv2d(
+                    in_channels, out_channels, kernel_size=1, stride=1, padding=0
+                )
+
+    def forward(self, x, temb):
+        h = x
+        h = self.norm1(h)
+        h = nonlinearity(h)
+        h = self.conv1(h)
+
+        if temb is not None:
+            h = h + self.temb_proj(nonlinearity(temb))[:, :, None, None]
+
+        h = self.norm2(h)
+        h = nonlinearity(h)
+        h = self.dropout(h)
+        h = self.conv2(h)
+
+        if self.in_channels != self.out_channels:
+            if self.use_conv_shortcut:
+                x = self.conv_shortcut(x)
+            else:
+                x = self.nin_shortcut(x)
+
+        return x + h
+
+
+class LinAttnBlock(LinearAttention):
+    """to match AttnBlock usage"""
+
+    def __init__(self, in_channels):
+        super().__init__(dim=in_channels, heads=1, dim_head=in_channels)
+
+
+class AttnBlock(nn.Module):
+    def __init__(self, in_channels):
+        super().__init__()
+        self.in_channels = in_channels
+
+        self.norm = Normalize(in_channels)
+        self.q = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.k = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.v = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+        self.proj_out = torch.nn.Conv2d(
+            in_channels, in_channels, kernel_size=1, stride=1, padding=0
+        )
+
+    def forward(self, x):
+        h_ = x
+        h_ = self.norm(h_)
+        q = self.q(h_)
+        k = self.k(h_)
+        v = self.v(h_)
+
+        # compute attention
+        b, c, h, w = q.shape
+        q = q.reshape(b, c, h * w).contiguous()
+        q = q.permute(0, 2, 1).contiguous()  # b,hw,c
+        k = k.reshape(b, c, h * w).contiguous()  # b,c,hw
+        w_ = torch.bmm(q, k).contiguous()  # b,hw,hw    w[b,i,j]=sum_c q[b,i,c]k[b,c,j]
+        w_ = w_ * (int(c) ** (-0.5))
+        w_ = torch.nn.functional.softmax(w_, dim=2)
+
+        # attend to values
+        v = v.reshape(b, c, h * w).contiguous()
+        w_ = w_.permute(0, 2, 1).contiguous()  # b,hw,hw (first hw of k, second of q)
+        h_ = torch.bmm(
+            v, w_
+        ).contiguous()  # b, c,hw (hw of q) h_[b,c,j] = sum_i v[b,c,i] w_[b,i,j]
+        h_ = h_.reshape(b, c, h, w).contiguous()
+
+        h_ = self.proj_out(h_)
+
+        return x + h_
+
+
+def make_attn(in_channels, attn_type="vanilla"):
+    assert attn_type in ["vanilla", "linear", "none"], f"attn_type {attn_type} unknown"
+    # print(f"making attention of type '{attn_type}' with {in_channels} in_channels")
+    if attn_type == "vanilla":
+        return AttnBlock(in_channels)
+    elif attn_type == "none":
+        return nn.Identity(in_channels)
+    else:
+        return LinAttnBlock(in_channels)
+
+
+class Model(nn.Module):
+    def __init__(
+        self,
+        *,
+        ch,
+        out_ch,
+        ch_mult=(1, 2, 4, 8),
+        num_res_blocks,
+        attn_resolutions,
+        dropout=0.0,
+        resamp_with_conv=True,
+        in_channels,
+        resolution,
+        use_timestep=True,
+        use_linear_attn=False,
+        attn_type="vanilla",
+    ):
+        super().__init__()
+        if use_linear_attn:
+            attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = self.ch * 4
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+
+        self.use_timestep = use_timestep
+        if self.use_timestep:
+            # timestep embedding
+            self.temb = nn.Module()
+            self.temb.dense = nn.ModuleList(
+                [
+                    torch.nn.Linear(self.ch, self.temb_ch),
+                    torch.nn.Linear(self.temb_ch, self.temb_ch),
+                ]
+            )
+
+        # downsampling
+        self.conv_in = torch.nn.Conv2d(
+            in_channels, self.ch, kernel_size=3, stride=1, padding=1
+        )
+
+        curr_res = resolution
+        in_ch_mult = (1,) + tuple(ch_mult)
+        self.down = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_in = ch * in_ch_mult[i_level]
+            block_out = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks):
+                block.append(
+                    ResnetBlock(
+                        in_channels=block_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            down = nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions - 1:
+                down.downsample = Downsample(block_in, resamp_with_conv)
+                curr_res = curr_res // 2
+            self.down.append(down)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+        self.mid.block_2 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+
+        # upsampling
+        self.up = nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_out = ch * ch_mult[i_level]
+            skip_in = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks + 1):
+                if i_block == self.num_res_blocks:
+                    skip_in = ch * in_ch_mult[i_level]
+                block.append(
+                    ResnetBlock(
+                        in_channels=block_in + skip_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            up = nn.Module()
+            up.block = block
+            up.attn = attn
+            if i_level != 0:
+                up.upsample = Upsample(block_in, resamp_with_conv)
+                curr_res = curr_res * 2
+            self.up.insert(0, up)  # prepend to get consistent order
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = torch.nn.Conv2d(
+            block_in, out_ch, kernel_size=3, stride=1, padding=1
+        )
+
+    def forward(self, x, t=None, context=None):
+        # assert x.shape[2] == x.shape[3] == self.resolution
+        if context is not None:
+            # assume aligned context, cat along channel axis
+            x = torch.cat((x, context), dim=1)
+        if self.use_timestep:
+            # timestep embedding
+            assert t is not None
+            temb = get_timestep_embedding(t, self.ch)
+            temb = self.temb.dense[0](temb)
+            temb = nonlinearity(temb)
+            temb = self.temb.dense[1](temb)
+        else:
+            temb = None
+
+        # downsampling
+        hs = [self.conv_in(x)]
+        for i_level in range(self.num_resolutions):
+            for i_block in range(self.num_res_blocks):
+                h = self.down[i_level].block[i_block](hs[-1], temb)
+                if len(self.down[i_level].attn) > 0:
+                    h = self.down[i_level].attn[i_block](h)
+                hs.append(h)
+            if i_level != self.num_resolutions - 1:
+                hs.append(self.down[i_level].downsample(hs[-1]))
+
+        # middle
+        h = hs[-1]
+        h = self.mid.block_1(h, temb)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h, temb)
+
+        # upsampling
+        for i_level in reversed(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks + 1):
+                h = self.up[i_level].block[i_block](
+                    torch.cat([h, hs.pop()], dim=1), temb
+                )
+                if len(self.up[i_level].attn) > 0:
+                    h = self.up[i_level].attn[i_block](h)
+            if i_level != 0:
+                h = self.up[i_level].upsample(h)
+
+        # end
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        return h
+
+    def get_last_layer(self):
+        return self.conv_out.weight
+
+
+class Encoder(nn.Module):
+    def __init__(
+        self,
+        *,
+        ch,
+        out_ch,
+        ch_mult=(1, 2, 4, 8),
+        num_res_blocks,
+        attn_resolutions,
+        dropout=0.0,
+        resamp_with_conv=True,
+        in_channels,
+        resolution,
+        z_channels,
+        double_z=True,
+        use_linear_attn=False,
+        attn_type="vanilla",
+        downsample_time_stride4_levels=[],
+        **ignore_kwargs,
+    ):
+        super().__init__()
+        if use_linear_attn:
+            attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = 0
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+        self.downsample_time_stride4_levels = downsample_time_stride4_levels
+
+        if len(self.downsample_time_stride4_levels) > 0:
+            assert max(self.downsample_time_stride4_levels) < self.num_resolutions, (
+                "The level to perform downsample 4 operation need to be smaller than the total resolution number %s"
+                % str(self.num_resolutions)
+            )
+
+        # downsampling
+        self.conv_in = torch.nn.Conv2d(
+            in_channels, self.ch, kernel_size=3, stride=1, padding=1
+        )
+
+        curr_res = resolution
+        in_ch_mult = (1,) + tuple(ch_mult)
+        self.in_ch_mult = in_ch_mult
+        self.down = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_in = ch * in_ch_mult[i_level]
+            block_out = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks):
+                block.append(
+                    ResnetBlock(
+                        in_channels=block_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            down = nn.Module()
+            down.block = block
+            down.attn = attn
+            if i_level != self.num_resolutions - 1:
+                if i_level in self.downsample_time_stride4_levels:
+                    down.downsample = DownsampleTimeStride4(block_in, resamp_with_conv)
+                else:
+                    down.downsample = Downsample(block_in, resamp_with_conv)
+                curr_res = curr_res // 2
+            self.down.append(down)
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+        self.mid.block_2 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = torch.nn.Conv2d(
+            block_in,
+            2 * z_channels if double_z else z_channels,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+        )
+
+    def forward(self, x):
+        # timestep embedding
+        temb = None
+        # downsampling
+        hs = [self.conv_in(x)]
+        for i_level in range(self.num_resolutions):
+            for i_block in range(self.num_res_blocks):
+                h = self.down[i_level].block[i_block](hs[-1], temb)
+                if len(self.down[i_level].attn) > 0:
+                    h = self.down[i_level].attn[i_block](h)
+                hs.append(h)
+            if i_level != self.num_resolutions - 1:
+                hs.append(self.down[i_level].downsample(hs[-1]))
+
+        # middle
+        h = hs[-1]
+        h = self.mid.block_1(h, temb)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h, temb)
+
+        # end
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        return h
+
+
+class Decoder(nn.Module):
+    def __init__(
+        self,
+        *,
+        ch,
+        out_ch,
+        ch_mult=(1, 2, 4, 8),
+        num_res_blocks,
+        attn_resolutions,
+        dropout=0.0,
+        resamp_with_conv=True,
+        in_channels,
+        resolution,
+        z_channels,
+        give_pre_end=False,
+        tanh_out=False,
+        use_linear_attn=False,
+        downsample_time_stride4_levels=[],
+        attn_type="vanilla",
+        **ignorekwargs,
+    ):
+        super().__init__()
+        if use_linear_attn:
+            attn_type = "linear"
+        self.ch = ch
+        self.temb_ch = 0
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        self.resolution = resolution
+        self.in_channels = in_channels
+        self.give_pre_end = give_pre_end
+        self.tanh_out = tanh_out
+        self.downsample_time_stride4_levels = downsample_time_stride4_levels
+
+        if len(self.downsample_time_stride4_levels) > 0:
+            assert max(self.downsample_time_stride4_levels) < self.num_resolutions, (
+                "The level to perform downsample 4 operation need to be smaller than the total resolution number %s"
+                % str(self.num_resolutions)
+            )
+
+        # compute in_ch_mult, block_in and curr_res at lowest res
+        in_ch_mult = (1,) + tuple(ch_mult)
+        block_in = ch * ch_mult[self.num_resolutions - 1]
+        curr_res = resolution // 2 ** (self.num_resolutions - 1)
+        self.z_shape = (1, z_channels, curr_res, curr_res)
+        # print("Working with z of shape {} = {} dimensions.".format(
+        # self.z_shape, np.prod(self.z_shape)))
+
+        # z to block_in
+        self.conv_in = torch.nn.Conv2d(
+            z_channels, block_in, kernel_size=3, stride=1, padding=1
+        )
+
+        # middle
+        self.mid = nn.Module()
+        self.mid.block_1 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+        self.mid.attn_1 = make_attn(block_in, attn_type=attn_type)
+        self.mid.block_2 = ResnetBlock(
+            in_channels=block_in,
+            out_channels=block_in,
+            temb_channels=self.temb_ch,
+            dropout=dropout,
+        )
+
+        # upsampling
+        self.up = nn.ModuleList()
+        for i_level in reversed(range(self.num_resolutions)):
+            block = nn.ModuleList()
+            attn = nn.ModuleList()
+            block_out = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks + 1):
+                block.append(
+                    ResnetBlock(
+                        in_channels=block_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+                if curr_res in attn_resolutions:
+                    attn.append(make_attn(block_in, attn_type=attn_type))
+            up = nn.Module()
+            up.block = block
+            up.attn = attn
+            if i_level != 0:
+                if i_level - 1 in self.downsample_time_stride4_levels:
+                    up.upsample = UpsampleTimeStride4(block_in, resamp_with_conv)
+                else:
+                    up.upsample = Upsample(block_in, resamp_with_conv)
+                curr_res = curr_res * 2
+            self.up.insert(0, up)  # prepend to get consistent order
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = torch.nn.Conv2d(
+            block_in, out_ch, kernel_size=3, stride=1, padding=1
+        )
+
+    def forward(self, z):
+        # assert z.shape[1:] == self.z_shape[1:]
+        self.last_z_shape = z.shape
+
+        # timestep embedding
+        temb = None
+
+        # z to block_in
+        h = self.conv_in(z)
+
+        # middle
+        h = self.mid.block_1(h, temb)
+        h = self.mid.attn_1(h)
+        h = self.mid.block_2(h, temb)
+
+        # upsampling
+        for i_level in reversed(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks + 1):
+                h = self.up[i_level].block[i_block](h, temb)
+                if len(self.up[i_level].attn) > 0:
+                    h = self.up[i_level].attn[i_block](h)
+            if i_level != 0:
+                h = self.up[i_level].upsample(h)
+
+        # end
+        if self.give_pre_end:
+            return h
+
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        if self.tanh_out:
+            h = torch.tanh(h)
+        return h
+
+
+class SimpleDecoder(nn.Module):
+    def __init__(self, in_channels, out_channels, *args, **kwargs):
+        super().__init__()
+        self.model = nn.ModuleList(
+            [
+                nn.Conv2d(in_channels, in_channels, 1),
+                ResnetBlock(
+                    in_channels=in_channels,
+                    out_channels=2 * in_channels,
+                    temb_channels=0,
+                    dropout=0.0,
+                ),
+                ResnetBlock(
+                    in_channels=2 * in_channels,
+                    out_channels=4 * in_channels,
+                    temb_channels=0,
+                    dropout=0.0,
+                ),
+                ResnetBlock(
+                    in_channels=4 * in_channels,
+                    out_channels=2 * in_channels,
+                    temb_channels=0,
+                    dropout=0.0,
+                ),
+                nn.Conv2d(2 * in_channels, in_channels, 1),
+                Upsample(in_channels, with_conv=True),
+            ]
+        )
+        # end
+        self.norm_out = Normalize(in_channels)
+        self.conv_out = torch.nn.Conv2d(
+            in_channels, out_channels, kernel_size=3, stride=1, padding=1
+        )
+
+    def forward(self, x):
+        for i, layer in enumerate(self.model):
+            if i in [1, 2, 3]:
+                x = layer(x, None)
+            else:
+                x = layer(x)
+
+        h = self.norm_out(x)
+        h = nonlinearity(h)
+        x = self.conv_out(h)
+        return x
+
+
+class UpsampleDecoder(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        out_channels,
+        ch,
+        num_res_blocks,
+        resolution,
+        ch_mult=(2, 2),
+        dropout=0.0,
+    ):
+        super().__init__()
+        # upsampling
+        self.temb_ch = 0
+        self.num_resolutions = len(ch_mult)
+        self.num_res_blocks = num_res_blocks
+        block_in = in_channels
+        curr_res = resolution // 2 ** (self.num_resolutions - 1)
+        self.res_blocks = nn.ModuleList()
+        self.upsample_blocks = nn.ModuleList()
+        for i_level in range(self.num_resolutions):
+            res_block = []
+            block_out = ch * ch_mult[i_level]
+            for i_block in range(self.num_res_blocks + 1):
+                res_block.append(
+                    ResnetBlock(
+                        in_channels=block_in,
+                        out_channels=block_out,
+                        temb_channels=self.temb_ch,
+                        dropout=dropout,
+                    )
+                )
+                block_in = block_out
+            self.res_blocks.append(nn.ModuleList(res_block))
+            if i_level != self.num_resolutions - 1:
+                self.upsample_blocks.append(Upsample(block_in, True))
+                curr_res = curr_res * 2
+
+        # end
+        self.norm_out = Normalize(block_in)
+        self.conv_out = torch.nn.Conv2d(
+            block_in, out_channels, kernel_size=3, stride=1, padding=1
+        )
+
+    def forward(self, x):
+        # upsampling
+        h = x
+        for k, i_level in enumerate(range(self.num_resolutions)):
+            for i_block in range(self.num_res_blocks + 1):
+                h = self.res_blocks[i_level][i_block](h, None)
+            if i_level != self.num_resolutions - 1:
+                h = self.upsample_blocks[k](h)
+        h = self.norm_out(h)
+        h = nonlinearity(h)
+        h = self.conv_out(h)
+        return h
+
+
+class LatentRescaler(nn.Module):
+    def __init__(self, factor, in_channels, mid_channels, out_channels, depth=2):
+        super().__init__()
+        # residual block, interpolate, residual block
+        self.factor = factor
+        self.conv_in = nn.Conv2d(
+            in_channels, mid_channels, kernel_size=3, stride=1, padding=1
+        )
+        self.res_block1 = nn.ModuleList(
+            [
+                ResnetBlock(
+                    in_channels=mid_channels,
+                    out_channels=mid_channels,
+                    temb_channels=0,
+                    dropout=0.0,
+                )
+                for _ in range(depth)
+            ]
+        )
+        self.attn = AttnBlock(mid_channels)
+        self.res_block2 = nn.ModuleList(
+            [
+                ResnetBlock(
+                    in_channels=mid_channels,
+                    out_channels=mid_channels,
+                    temb_channels=0,
+                    dropout=0.0,
+                )
+                for _ in range(depth)
+            ]
+        )
+
+        self.conv_out = nn.Conv2d(
+            mid_channels,
+            out_channels,
+            kernel_size=1,
+        )
+
+    def forward(self, x):
+        x = self.conv_in(x)
+        for block in self.res_block1:
+            x = block(x, None)
+        x = torch.nn.functional.interpolate(
+            x,
+            size=(
+                int(round(x.shape[2] * self.factor)),
+                int(round(x.shape[3] * self.factor)),
+            ),
+        )
+        x = self.attn(x).contiguous()
+        for block in self.res_block2:
+            x = block(x, None)
+        x = self.conv_out(x)
+        return x
+
+
+class MergedRescaleEncoder(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        ch,
+        resolution,
+        out_ch,
+        num_res_blocks,
+        attn_resolutions,
+        dropout=0.0,
+        resamp_with_conv=True,
+        ch_mult=(1, 2, 4, 8),
+        rescale_factor=1.0,
+        rescale_module_depth=1,
+    ):
+        super().__init__()
+        intermediate_chn = ch * ch_mult[-1]
+        self.encoder = Encoder(
+            in_channels=in_channels,
+            num_res_blocks=num_res_blocks,
+            ch=ch,
+            ch_mult=ch_mult,
+            z_channels=intermediate_chn,
+            double_z=False,
+            resolution=resolution,
+            attn_resolutions=attn_resolutions,
+            dropout=dropout,
+            resamp_with_conv=resamp_with_conv,
+            out_ch=None,
+        )
+        self.rescaler = LatentRescaler(
+            factor=rescale_factor,
+            in_channels=intermediate_chn,
+            mid_channels=intermediate_chn,
+            out_channels=out_ch,
+            depth=rescale_module_depth,
+        )
+
+    def forward(self, x):
+        x = self.encoder(x)
+        x = self.rescaler(x)
+        return x
+
+
+class MergedRescaleDecoder(nn.Module):
+    def __init__(
+        self,
+        z_channels,
+        out_ch,
+        resolution,
+        num_res_blocks,
+        attn_resolutions,
+        ch,
+        ch_mult=(1, 2, 4, 8),
+        dropout=0.0,
+        resamp_with_conv=True,
+        rescale_factor=1.0,
+        rescale_module_depth=1,
+    ):
+        super().__init__()
+        tmp_chn = z_channels * ch_mult[-1]
+        self.decoder = Decoder(
+            out_ch=out_ch,
+            z_channels=tmp_chn,
+            attn_resolutions=attn_resolutions,
+            dropout=dropout,
+            resamp_with_conv=resamp_with_conv,
+            in_channels=None,
+            num_res_blocks=num_res_blocks,
+            ch_mult=ch_mult,
+            resolution=resolution,
+            ch=ch,
+        )
+        self.rescaler = LatentRescaler(
+            factor=rescale_factor,
+            in_channels=z_channels,
+            mid_channels=tmp_chn,
+            out_channels=tmp_chn,
+            depth=rescale_module_depth,
+        )
+
+    def forward(self, x):
+        x = self.rescaler(x)
+        x = self.decoder(x)
+        return x
+
+
+class Upsampler(nn.Module):
+    def __init__(self, in_size, out_size, in_channels, out_channels, ch_mult=2):
+        super().__init__()
+        assert out_size >= in_size
+        num_blocks = int(np.log2(out_size // in_size)) + 1
+        factor_up = 1.0 + (out_size % in_size)
+        print(
+            f"Building {self.__class__.__name__} with in_size: {in_size} --> out_size {out_size} and factor {factor_up}"
+        )
+        self.rescaler = LatentRescaler(
+            factor=factor_up,
+            in_channels=in_channels,
+            mid_channels=2 * in_channels,
+            out_channels=in_channels,
+        )
+        self.decoder = Decoder(
+            out_ch=out_channels,
+            resolution=out_size,
+            z_channels=in_channels,
+            num_res_blocks=2,
+            attn_resolutions=[],
+            in_channels=None,
+            ch=in_channels,
+            ch_mult=[ch_mult for _ in range(num_blocks)],
+        )
+
+    def forward(self, x):
+        x = self.rescaler(x)
+        x = self.decoder(x)
+        return x
+
+
+class Resize(nn.Module):
+    def __init__(self, in_channels=None, learned=False, mode="bilinear"):
+        super().__init__()
+        self.with_conv = learned
+        self.mode = mode
+        if self.with_conv:
+            print(
+                f"Note: {self.__class__.__name} uses learned downsampling and will ignore the fixed {mode} mode"
+            )
+            raise NotImplementedError()
+            assert in_channels is not None
+            # no asymmetric padding in torch conv, must do it ourselves
+            self.conv = torch.nn.Conv2d(
+                in_channels, in_channels, kernel_size=4, stride=2, padding=1
+            )
+
+    def forward(self, x, scale_factor=1.0):
+        if scale_factor == 1.0:
+            return x
+        else:
+            x = torch.nn.functional.interpolate(
+                x, mode=self.mode, align_corners=False, scale_factor=scale_factor
+            )
+        return x
+
+
+class FirstStagePostProcessor(nn.Module):
+    def __init__(
+        self,
+        ch_mult: list,
+        in_channels,
+        pretrained_model: nn.Module = None,
+        reshape=False,
+        n_channels=None,
+        dropout=0.0,
+        pretrained_config=None,
+    ):
+        super().__init__()
+        if pretrained_config is None:
+            assert (
+                pretrained_model is not None
+            ), 'Either "pretrained_model" or "pretrained_config" must not be None'
+            self.pretrained_model = pretrained_model
+        else:
+            assert (
+                pretrained_config is not None
+            ), 'Either "pretrained_model" or "pretrained_config" must not be None'
+            self.instantiate_pretrained(pretrained_config)
+
+        self.do_reshape = reshape
+
+        if n_channels is None:
+            n_channels = self.pretrained_model.encoder.ch
+
+        self.proj_norm = Normalize(in_channels, num_groups=in_channels // 2)
+        self.proj = nn.Conv2d(
+            in_channels, n_channels, kernel_size=3, stride=1, padding=1
+        )
+
+        blocks = []
+        downs = []
+        ch_in = n_channels
+        for m in ch_mult:
+            blocks.append(
+                ResnetBlock(
+                    in_channels=ch_in, out_channels=m * n_channels, dropout=dropout
+                )
+            )
+            ch_in = m * n_channels
+            downs.append(Downsample(ch_in, with_conv=False))
+
+        self.model = nn.ModuleList(blocks)
+        self.downsampler = nn.ModuleList(downs)
+
+    def instantiate_pretrained(self, config):
+        model = instantiate_from_config(config)
+        self.pretrained_model = model.eval()
+        # self.pretrained_model.train = False
+        for param in self.pretrained_model.parameters():
+            param.requires_grad = False
+
+    @torch.no_grad()
+    def encode_with_pretrained(self, x):
+        c = self.pretrained_model.encode(x)
+        if isinstance(c, DiagonalGaussianDistribution):
+            c = c.mode()
+        return c
+
+    def forward(self, x):
+        z_fs = self.encode_with_pretrained(x)
+        z = self.proj_norm(z_fs)
+        z = self.proj(z)
+        z = nonlinearity(z)
+
+        for submodel, downmodel in zip(self.model, self.downsampler):
+            z = submodel(z, temb=None)
+            z = downmodel(z)
+
+        if self.do_reshape:
+            z = rearrange(z, "b c h w -> b (h w) c")
+        return z
diff --git a/picoaudio/data/filter_data.py b/picoaudio/data/filter_data.py
new file mode 100644
index 0000000000000000000000000000000000000000..0eb226cace5e44234b3ae4f2108a3ec817d1a38c
--- /dev/null
+++ b/picoaudio/data/filter_data.py
@@ -0,0 +1,24 @@
+def get_event_list():
+    event_list = [
+        "burping_belching",             # 0
+        "car_horn_honking",             #
+        "cat_meowing",                  #    
+        "cow_mooing",                   #
+        "dog_barking",                  #  
+        "door_knocking",                #
+        "door_slamming",                #
+        "explosion",                    #  
+        "gunshot",                      # 8
+        "sheep_goat_bleating",          #
+        "sneeze",                       #
+        "spraying",                     # 
+        "thump_thud",                   #   
+        "train_horn",                   #
+        "tapping_clicking_clanking",    #
+        "woman_laughing",               #         
+        "duck_quacking",                # 16   
+        "whistling",                    #    
+
+    ]
+    return event_list
+    
diff --git a/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json b/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..de9e98acac39a0935f7baee5baa9a6b98eb8cd8d
--- /dev/null
+++ b/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_multi-event.json
@@ -0,0 +1,200 @@
+{"filepath": "data/multi_event_test/syn_1.wav", "onoffCaption": "cat meowing at 0.5-2.0, 3.0-4.5 and whistling at 5.0-6.5 and explosion at 7.0-8.0, 8.5-9.5", "frequencyCaption": "cat meowing two times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_6.wav", "onoffCaption": "whistling at 2.0-6.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_8.wav", "onoffCaption": "cow mooing at 1.954-4.954, 6.219-9.219", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_11.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5 and dog barking at 5.0-7.0", "frequencyCaption": "burping belching two times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_16.wav", "onoffCaption": "duck quacking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_18.wav", "onoffCaption": "door knocking at 0.138-2.518, 3.708-6.088 and door slamming at 2.798-4.798", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_21.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_26.wav", "onoffCaption": "whistling at 0.2-4.2", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_28.wav", "onoffCaption": "cow mooing at 0.0-1.0 and spraying at 1.0-2.0", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_32.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_35.wav", "onoffCaption": "car horn honking at 0.5-2.5, 3.0-5.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_43.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5 and burping belching at 5.0-7.0, 7.5-9.5 and explosion at 4.8-7.8", "frequencyCaption": "dog barking two times and burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_44.wav", "onoffCaption": "sneeze at 0.5-1.5", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_50.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0 and sneeze at 6.0-7.0 and train horn at 8.0-10.0", "frequencyCaption": "car horn honking two times and sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_57.wav", "onoffCaption": "dog barking at 0.0-2.0, 3.0-5.0 and cow mooing at 6.0-9.0", "frequencyCaption": "dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_59.wav", "onoffCaption": "door slamming at 0.0-1.0 and explosion at 1.5-4.5, 5.0-8.0", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_60.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_67.wav", "onoffCaption": "whistling at 0.204-5.379", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_69.wav", "onoffCaption": "door knocking at 0-1, 2-3, 4-5", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_73.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3 and sneeze at 3-4, 4-5", "frequencyCaption": "door knocking three times and sneeze two times"}
+{"filepath": "data/multi_event_test/syn_74.wav", "onoffCaption": "spraying at 0.5-1.0, 1.5-2.0 and gunshot at 3.0-4.0, 5.0-6.0, 7.0-8.0", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_test/syn_82.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_91.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_96.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_98.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_101.wav", "onoffCaption": "dog barking at 0.464-2.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_106.wav", "onoffCaption": "burping belching at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_test/syn_108.wav", "onoffCaption": "sneeze at 0.5-1.5", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_112.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_115.wav", "onoffCaption": "duck quacking at 0.3-2.3 and tapping clicking clanking at 2.5-5.5, 6.0-9.0", "frequencyCaption": "duck quacking one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_122.wav", "onoffCaption": "door knocking at 0-1, 3-4, 6-7", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_125.wav", "onoffCaption": "cow mooing at 1.5-4.5, 5.5-8.5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_131.wav", "onoffCaption": "whistling at 0-1, 2-3 and cat meowing at 1-2", "frequencyCaption": "whistling two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_136.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.5-5.5 and whistling at 6.0-8.0, 8.5-9.5 and woman laughing at 2.0-4.0", "frequencyCaption": "sheep goat bleating two times and whistling two times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_138.wav", "onoffCaption": "gunshot at 0.0-1.0 and tapping clicking clanking at 1.5-5.0", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_140.wav", "onoffCaption": "door knocking at 0.00-2.00, 3.00-5.00, 6.00-8.00", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_147.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_149.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0 and spraying at 5.5-6.0, 7.0-7.5", "frequencyCaption": "car horn honking two times and spraying two times"}
+{"filepath": "data/multi_event_test/syn_153.wav", "onoffCaption": "cat meowing at 0-1.0", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_154.wav", "onoffCaption": "cat meowing at 0.5-1.5 and door knocking at 2-3.5", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_163.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_164.wav", "onoffCaption": "whistling at 0.204-5.379, 7.724-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_test/syn_170.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_177.wav", "onoffCaption": "thump thud at 0-1 and cow mooing at 1-2", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_179.wav", "onoffCaption": "cow mooing at 1.954-4.602, 5.719-8.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_181.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_186.wav", "onoffCaption": "gunshot at 0.0-1.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_188.wav", "onoffCaption": "gunshot at 0-1, 1-2 and duck quacking at 2-3", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_192.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_195.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_3.wav", "onoffCaption": "tapping clicking clanking at 0.5-3.0, 4.0-7.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_4.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_13.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_14.wav", "onoffCaption": "sneeze at 0.38-1.38", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_23.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.5-3.5", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_24.wav", "onoffCaption": "woman laughing at 2.782-5.368", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_30.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.871-7.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_37.wav", "onoffCaption": "thump thud at 0.0-1.5, 5.0-6.5 and door knocking at 1.5-3.5, 6.5-8.5 and burping belching at 3.5-4.5", "frequencyCaption": "thump thud two times and door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_39.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_41.wav", "onoffCaption": "thump thud at 0.0-1.0", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_test/syn_48.wav", "onoffCaption": "cat meowing at 0-1, 1-2, 2-3", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_52.wav", "onoffCaption": "gunshot at 0.0-1.0 and duck quacking at 1.5-2.5 and tapping clicking clanking at 3.0-4.0", "frequencyCaption": "gunshot one times and duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_55.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_62.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_65.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_71.wav", "onoffCaption": "door slamming at 0.0-1.0, 2.0-3.0 and whistling at 4.0-8.0", "frequencyCaption": "door slamming two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_76.wav", "onoffCaption": "dog barking at 0.464-2.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_78.wav", "onoffCaption": "explosion at 0.0-2.0, 2.5-4.5 and duck quacking at 5.0-7.0, 7.5-9.5", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_80.wav", "onoffCaption": "door slamming at 0.0-1.0 and sheep goat bleating at 2.0-4.0", "frequencyCaption": "door slamming one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_85.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_87.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_89.wav", "onoffCaption": "car horn honking at 0.0-2.0 and cat meowing at 2.5-4.0", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_93.wav", "onoffCaption": "dog barking at 0-2, 2-4", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_94.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0 and thump thud at 1.0-3.0, 4.0-6.0 and sheep goat bleating at 2.0-4.0, 7.0-9.0", "frequencyCaption": "gunshot three times and thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_103.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_104.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_110.wav", "onoffCaption": "train horn at 0-1 and duck quacking at 1-2 and cow mooing at 2-3", "frequencyCaption": "train horn one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_117.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.5-6.5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_119.wav", "onoffCaption": "train horn at 0.0-2.0 and door knocking at 2.5-4.5, 5.0-7.0", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_120.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.871-5.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_127.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_129.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_133.wav", "onoffCaption": "duck quacking at 2.203-4.203, 5.361-7.361", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_134.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_142.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.0-3.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_145.wav", "onoffCaption": "door knocking at 0.002-2.092, 2.842-5.601 and whistling at 1.9-10.0", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_151.wav", "onoffCaption": "dog barking at 0.121-2.121, 3.824-5.824, 7.767-9.767", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_156.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_158.wav", "onoffCaption": "tapping clicking clanking at 1.5-4.5, 5.5-8.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_161.wav", "onoffCaption": "spraying at 0-1, 2-3", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_166.wav", "onoffCaption": "woman laughing at 1.672-3.955", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_168.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_172.wav", "onoffCaption": "door knocking at 0-1, 1-2, 2-3", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_175.wav", "onoffCaption": "cow mooing at 0-3 and spraying at 3-6", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_183.wav", "onoffCaption": "explosion at 0.0-2.0, 2.1-4.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_184.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_190.wav", "onoffCaption": "whistling at 0.0-1.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_197.wav", "onoffCaption": "tapping clicking clanking at 0.032-2.032, 2.532-4.532", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_199.wav", "onoffCaption": "duck quacking at 0.0-2.0 and cat meowing at 2.5-4.5", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_200.wav", "onoffCaption": "explosion at 1.0-3.0, 4.0-6.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_2.wav", "onoffCaption": "door knocking at 0.0-1.0", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_5.wav", "onoffCaption": "burping belching at 0.359-2.774", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_test/syn_12.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0 and sneeze at 2.5-3.5", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_test/syn_15.wav", "onoffCaption": "tapping clicking clanking at 2.992-6.432", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_22.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_25.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.391-7.391", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_31.wav", "onoffCaption": "woman laughing at 0-1, 2-3", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_36.wav", "onoffCaption": "door slamming at 0.5-1.5, 2.0-3.0, 3.5-4.5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_38.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_40.wav", "onoffCaption": "door knocking at 0.138-2.518, 3.708-6.088 and cow mooing at 6.91-9.447", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_46.wav", "onoffCaption": "door slamming at 1.145-2.085, 3.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_47.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0 and cow mooing at 6.0-8.0, 8.5-10.0", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_49.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0 and tapping clicking clanking at 0.0-4.0, 5.5-9.5", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_51.wav", "onoffCaption": "train horn at 0.873-4.633, 5.147-8.907", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_53.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_54.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_63.wav", "onoffCaption": "train horn at 0-1 and cat meowing at 2-3 and dog barking at 4-5", "frequencyCaption": "train horn one times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_70.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_77.wav", "onoffCaption": "cow mooing at 0.0-3.0", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_79.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_81.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_86.wav", "onoffCaption": "whistling at 0-1 and woman laughing at 1-3, 3-5", "frequencyCaption": "whistling one times and woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_88.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.0-6.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_92.wav", "onoffCaption": "door slamming at 0.0-1.0, 2.0-3.0, 4.0-5.0 and tapping clicking clanking at 6.0-7.0", "frequencyCaption": "door slamming three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_95.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_102.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_105.wav", "onoffCaption": "train horn at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_test/syn_111.wav", "onoffCaption": "whistling at 0.204-2.79, 4.0-6.586 and door slamming at 7.0-8.0", "frequencyCaption": "whistling two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_116.wav", "onoffCaption": "burping belching at 1.0-3.0, 4.0-6.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_118.wav", "onoffCaption": "sneeze at 0.0-1.0", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_121.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.65-2.65, 3.65-5.65", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_126.wav", "onoffCaption": "sneeze at 0.373-2.332 and car horn honking at 1.03-5.542, 6.081-10.0", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_128.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0 and door knocking at 3.5-5.5, 6.0-8.0", "frequencyCaption": "sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_132.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0 and spraying at 2.5-3.0, 4.0-4.5, 5.5-6.0 and duck quacking at 6.5-7.5, 8.0-9.0, 9.5-10.0", "frequencyCaption": "sheep goat bleating one times and spraying three times and duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_135.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_139.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_143.wav", "onoffCaption": "spraying at 0.0-1.0 and explosion at 1.5-4.5", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_144.wav", "onoffCaption": "duck quacking at 0-1, 2-3, 4-5", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_150.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_157.wav", "onoffCaption": "train horn at 0-3.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_159.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_160.wav", "onoffCaption": "spraying at 0.0-1.0 and whistling at 1.0-3.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_167.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5 and gunshot at 5.0-7.0", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_169.wav", "onoffCaption": "sneeze at 0.373-2.332, 3.255-5.716", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_173.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_174.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_176.wav", "onoffCaption": "woman laughing at 1.625-3.98, 4.735-6.981", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_182.wav", "onoffCaption": "cow mooing at 0.0-3.0 and gunshot at 4.0-5.0", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_185.wav", "onoffCaption": "spraying at 0.22-0.847 and door knocking at 2.797-5.334", "frequencyCaption": "spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_189.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_191.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.5-4.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_193.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_196.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_test/syn_198.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_7.wav", "onoffCaption": "spraying at 0.0-1.0 and burping belching at 1.5-2.5", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_9.wav", "onoffCaption": "cow mooing at 0.0-3.0", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_10.wav", "onoffCaption": "door knocking at 2-4, 5-7", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_17.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_19.wav", "onoffCaption": "gunshot at 0.0-1.0 and spraying at 1.5-2.5", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_27.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_33.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5 and car horn honking at 5.0-7.0", "frequencyCaption": "dog barking two times and car horn honking one times"}
+{"filepath": "data/multi_event_test/syn_34.wav", "onoffCaption": "sheep goat bleating at 1.575-3.575", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_42.wav", "onoffCaption": "tapping clicking clanking at 0.0-2.0, 2.5-4.5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_45.wav", "onoffCaption": "cat meowing at 0.5-1.5 and train horn at 2.0-6.0", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_56.wav", "onoffCaption": "tapping clicking clanking at 0.961-4.401, 6.37-9.81", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_58.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_test/syn_61.wav", "onoffCaption": "explosion at 0.5-3.5, 4.0-7.0 and train horn at 7.5-10.0 and woman laughing at 1.0-4.0", "frequencyCaption": "explosion two times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_66.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_68.wav", "onoffCaption": "car horn honking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_72.wav", "onoffCaption": "spraying at 0.0-0.6, 1.0-1.6 and thump thud at 2.0-3.6 and dog barking at 4.0-6.0", "frequencyCaption": "spraying two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_75.wav", "onoffCaption": "explosion at 0.5-2.5, 2.501-4.501", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_83.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_84.wav", "onoffCaption": "burping belching at 0.871-3.871, 4.642-7.642", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_90.wav", "onoffCaption": "gunshot at 0.2-1.2", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_97.wav", "onoffCaption": "cat meowing at 0.5-1.5", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_99.wav", "onoffCaption": "duck quacking at 0.0-2.0, 2.0-4.0", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_100.wav", "onoffCaption": "cat meowing at 0.0-2.0 and sheep goat bleating at 3.0-5.0, 6.0-8.0, 9.0-10.0", "frequencyCaption": "cat meowing one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_107.wav", "onoffCaption": "spraying at 0.0-1.5, 2.0-3.5 and dog barking at 4.0-6.0, 7.0-9.0 and tapping clicking clanking at 1.6-3.1, 3.6-5.1", "frequencyCaption": "spraying two times and dog barking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_109.wav", "onoffCaption": "cow mooing at 0.0-3.0 and gunshot at 3.5-4.5", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_113.wav", "onoffCaption": "whistling at 0.742-5.917 and tapping clicking clanking at 2.992-6.432", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_114.wav", "onoffCaption": "car horn honking at 0-2 and door knocking at 2-4", "frequencyCaption": "car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_124.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_130.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_137.wav", "onoffCaption": "door knocking at 0-1 and cow mooing at 2-3 and gunshot at 4-5", "frequencyCaption": "door knocking one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_141.wav", "onoffCaption": "sneeze at 0.33-1.403, 2.759-3.832", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_146.wav", "onoffCaption": "sneeze at 0.0-1.0, 2.0-3.0 and cat meowing at 4.0-5.0", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_148.wav", "onoffCaption": "duck quacking at 0-1, 2-3 and cow mooing at 4-5", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_152.wav", "onoffCaption": "tapping clicking clanking at 0.0-1.0, 1.5-2.5 and train horn at 3.0-7.0", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_155.wav", "onoffCaption": "tapping clicking clanking at 0.0-1.0 and gunshot at 2.0-3.0 and cat meowing at 4.0-5.0", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_162.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_165.wav", "onoffCaption": "thump thud at 0.0-1.5, 2.0-3.5 and whistling at 4.0-7.0", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_171.wav", "onoffCaption": "spraying at 0.0-0.5, 1.5-2.0, 3.0-3.5 and thump thud at 4.0-5.0 and sheep goat bleating at 5.5-6.5, 7.0-8.0", "frequencyCaption": "spraying three times and thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_178.wav", "onoffCaption": "door slamming at 0.355-2.581 and woman laughing at 0.964-3.319", "frequencyCaption": "door slamming one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_180.wav", "onoffCaption": "spraying at 0.0-1.0 and cow mooing at 2.0-5.0", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_187.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_194.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
diff --git a/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json b/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..9310b20b6657974a4e9a39f0ddf9c0ab7c252c6f
--- /dev/null
+++ b/picoaudio/data/meta_data/test-frequency-control_onoffFromGpt_single-event.json
@@ -0,0 +1,400 @@
+{"filepath": "data/single_event_multi_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 1.674-5.019", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_6.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_8.wav", "onoffCaption": "door slamming at 0-1", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_11.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_16.wav", "onoffCaption": "thump thud at 0-1", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_18.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_21.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_26.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_28.wav", "onoffCaption": "sneeze at 0-1, 2-3", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_32.wav", "onoffCaption": "cow mooing at 0-3.309", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_35.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_43.wav", "onoffCaption": "thump thud at 0-1, 2-3", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_44.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_50.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_57.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_59.wav", "onoffCaption": "woman laughing at 2.0-4.5, 5.0-7.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_60.wav", "onoffCaption": "cat meowing at 1-2, 3-4", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_69.wav", "onoffCaption": "burping belching at 2.0-3.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_73.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_74.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_82.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-8.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_91.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_96.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_98.wav", "onoffCaption": "woman laughing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_101.wav", "onoffCaption": "burping belching at 0.871-1.871, 2.871-3.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_106.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_108.wav", "onoffCaption": "cat meowing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_112.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_115.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.0-4.0, 5.0-8.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_125.wav", "onoffCaption": "car horn honking at 0.0-2.0, 3.0-5.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_131.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_136.wav", "onoffCaption": "tapping clicking clanking at 1-3, 6-8", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_138.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_140.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_147.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.5-5.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_149.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_153.wav", "onoffCaption": "cow mooing at 0-1, 2-3", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_154.wav", "onoffCaption": "train horn at 0-1, 2-3", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_163.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_164.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_170.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_177.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_179.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_181.wav", "onoffCaption": "door knocking at 0-1.0", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_186.wav", "onoffCaption": "sneeze at 0.3-1.3, 2.3-3.3", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_188.wav", "onoffCaption": "explosion at 0-2", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_192.wav", "onoffCaption": "cat meowing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_195.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_4.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.032-1.032, 2.032-3.032, 4.032-5.032", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_14.wav", "onoffCaption": "tapping clicking clanking at 0-1, 2-3", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_23.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_24.wav", "onoffCaption": "thump thud at 1-2, 3-4", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_30.wav", "onoffCaption": "explosion at 0-1, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_37.wav", "onoffCaption": "tapping clicking clanking at 0.5-2.5, 3-5", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_39.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_41.wav", "onoffCaption": "car horn honking at 1-2, 3-4", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_48.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_55.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_62.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_65.wav", "onoffCaption": "tapping clicking clanking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_71.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_76.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_78.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_80.wav", "onoffCaption": "car horn honking at 0-1", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_85.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_87.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_89.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_93.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_94.wav", "onoffCaption": "burping belching at 0.871-2.871, 5.218-7.218", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_103.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_104.wav", "onoffCaption": "thump thud at 0-2, 3-5", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_110.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_117.wav", "onoffCaption": "tapping clicking clanking at 0-2", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_119.wav", "onoffCaption": "duck quacking at 0.235-2.235, 3.085-5.085", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_120.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_127.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_129.wav", "onoffCaption": "burping belching at 0.5-1.5, 2.5-3.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_133.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_134.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_142.wav", "onoffCaption": "cow mooing at 1-2, 3-4", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_145.wav", "onoffCaption": "cow mooing at 0.0-2.0, 3.0-5.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_151.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_156.wav", "onoffCaption": "thump thud at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_158.wav", "onoffCaption": "burping belching at 1.5-3.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_161.wav", "onoffCaption": "car horn honking at 1.5-3.5, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_166.wav", "onoffCaption": "burping belching at 0-1, 1-2, 2-3", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_168.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_172.wav", "onoffCaption": "woman laughing at 0.0-2.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_175.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_183.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_184.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_190.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_197.wav", "onoffCaption": "car horn honking at 1.817-4.404, 5.85-8.437", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_199.wav", "onoffCaption": "car horn honking at 0.664-3.129", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_200.wav", "onoffCaption": "train horn at 0-2", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 0.5-2.5", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_12.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_15.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_22.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_25.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_31.wav", "onoffCaption": "sneeze at 0.0-1.0", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_36.wav", "onoffCaption": "sheep goat bleating at 0-1, 1-2", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_38.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_40.wav", "onoffCaption": "woman laughing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_46.wav", "onoffCaption": "tapping clicking clanking at 1.5-5.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_47.wav", "onoffCaption": "cow mooing at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_49.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_51.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_53.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_54.wav", "onoffCaption": "cow mooing at 1.954-6.383, 7.52-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_63.wav", "onoffCaption": "explosion at 0.0-3.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_64.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_70.wav", "onoffCaption": "dog barking at 1-2", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_77.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_79.wav", "onoffCaption": "train horn at 0-2.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_81.wav", "onoffCaption": "tapping clicking clanking at 0-1", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_86.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.1-4.1", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_88.wav", "onoffCaption": "car horn honking at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_92.wav", "onoffCaption": "door slamming at 0-1, 2-4", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_95.wav", "onoffCaption": "woman laughing at 1.5-3.5, 4.0-6.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_102.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_111.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_116.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_118.wav", "onoffCaption": "sheep goat bleating at 0.5-1.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_121.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_126.wav", "onoffCaption": "burping belching at 0.5-2.5, 3.0-5.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_128.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_132.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_135.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_139.wav", "onoffCaption": "spraying at 0-1, 1-2", "frequencyCaption": "spraying two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_143.wav", "onoffCaption": "door knocking at 0.645-2.772, 3.875-6.782, 7.405-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_144.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_150.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_157.wav", "onoffCaption": "explosion at 0.5-1.5, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_159.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.5-3.5", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_160.wav", "onoffCaption": "woman laughing at 0.0-2.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_167.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_169.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_173.wav", "onoffCaption": "explosion at 0-3", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_174.wav", "onoffCaption": "duck quacking at 0.2-2.2, 3.2-5.2", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_176.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_182.wav", "onoffCaption": "car horn honking at 0.653-3.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_185.wav", "onoffCaption": "dog barking at 0-2, 3-5", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_189.wav", "onoffCaption": "burping belching at 0-1, 2-3", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_191.wav", "onoffCaption": "tapping clicking clanking at 0.0-4.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_193.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_196.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_198.wav", "onoffCaption": "sheep goat bleating at 0.56-2.56", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_7.wav", "onoffCaption": "door slamming at 0-2, 2-4", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_9.wav", "onoffCaption": "sneeze at 0.5-1.5, 2.0-3.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_10.wav", "onoffCaption": "door slamming at 0.0-1.0", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_17.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_19.wav", "onoffCaption": "thump thud at 1.9-4.5, 5.5-8.1", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_20.wav", "onoffCaption": "dog barking at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_27.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_29.wav", "onoffCaption": "woman laughing at 0-1", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_33.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_34.wav", "onoffCaption": "dog barking at 0-1, 1.5-2.5, 3-4", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_42.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_45.wav", "onoffCaption": "woman laughing at 0-1", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_56.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_58.wav", "onoffCaption": "spraying at 0.5-1.5", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_61.wav", "onoffCaption": "sheep goat bleating at 0.8-2.8, 3.8-5.8", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_66.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_68.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_72.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_75.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_83.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_84.wav", "onoffCaption": "burping belching at 0-3", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_90.wav", "onoffCaption": "whistling at 1-2", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_97.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_99.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_100.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_107.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_113.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_114.wav", "onoffCaption": "explosion at 0.0-3.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_124.wav", "onoffCaption": "woman laughing at 0-1, 1-2", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_130.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_137.wav", "onoffCaption": "train horn at 0-1, 2-3, 4-5", "frequencyCaption": "train horn three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 0.5-3.5", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_146.wav", "onoffCaption": "sneeze at 0.8-1.8", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_148.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_152.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_155.wav", "onoffCaption": "spraying at 0.033-1.519", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_162.wav", "onoffCaption": "explosion at 0-1, 2-3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_165.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_171.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_178.wav", "onoffCaption": "tapping clicking clanking at 1-3, 4-6", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_187.wav", "onoffCaption": "explosion at 1.5-4.5", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_194.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_11.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_16.wav", "onoffCaption": "burping belching at 0.5-3.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_18.wav", "onoffCaption": "burping belching at 1.0-2.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_21.wav", "onoffCaption": "burping belching at 0.5-1.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_26.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_28.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_32.wav", "onoffCaption": "gunshot at 0-1, 2-3, 4-5", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_35.wav", "onoffCaption": "woman laughing at 0-2, 3-5", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_43.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_44.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_50.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_57.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_59.wav", "onoffCaption": "gunshot at 0-1, 2-3, 4-5", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_60.wav", "onoffCaption": "cow mooing at 0.0-2.0, 2.5-4.5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_69.wav", "onoffCaption": "burping belching at 0.0-2.0, 2.1-4.1", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_73.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_74.wav", "onoffCaption": "cow mooing at 0-2, 3-5", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_82.wav", "onoffCaption": "woman laughing at 2.0-6.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_85.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_91.wav", "onoffCaption": "tapping clicking clanking at 1.0-4.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_96.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_98.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_101.wav", "onoffCaption": "spraying at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_106.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_108.wav", "onoffCaption": "gunshot at 0.2-2.2, 3.2-5.2", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_112.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.871-5.871", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_115.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.0-5.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_125.wav", "onoffCaption": "explosion at 1.0-3.0, 3.5-5.5", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_131.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_136.wav", "onoffCaption": "car horn honking at 0-1", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_138.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_140.wav", "onoffCaption": "train horn at 0-1, 1-2", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_147.wav", "onoffCaption": "explosion at 0-1, 2-3, 4-5", "frequencyCaption": "explosion three times"}
+{"filepath": "data/single_event_single_identity_test/syn_149.wav", "onoffCaption": "spraying at 0.1-1.1, 1.2-2.2, 3.3-4.3", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_153.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_154.wav", "onoffCaption": "explosion at 0.0-1.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_163.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_164.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_170.wav", "onoffCaption": "burping belching at 0.5-2.5", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_175.wav", "onoffCaption": "explosion at 2.941-5.813", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_177.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_179.wav", "onoffCaption": "explosion at 0.0-4.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_181.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_186.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 2.5-4.5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_188.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_190.wav", "onoffCaption": "woman laughing at 0-1, 2-3", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_192.wav", "onoffCaption": "door knocking at 1-2, 3-4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_195.wav", "onoffCaption": "cow mooing at 1.5-3.5, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_14.wav", "onoffCaption": "woman laughing at 0-2", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_23.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_24.wav", "onoffCaption": "dog barking at 0.311-2.711", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_30.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_37.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_39.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_41.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 2.75-4.75", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_46.wav", "onoffCaption": "car horn honking at 1.0-3.0, 4.0-6.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_48.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_55.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_62.wav", "onoffCaption": "burping belching at 1.5-3.5, 4.0-6.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_65.wav", "onoffCaption": "sheep goat bleating at 1.0-3.0, 4.0-6.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_71.wav", "onoffCaption": "tapping clicking clanking at 0.5-3.0", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_76.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_78.wav", "onoffCaption": "train horn at 0-1, 2-3", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_80.wav", "onoffCaption": "whistling at 0-1, 2-3", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_87.wav", "onoffCaption": "car horn honking at 1.817-4.12, 6.106-8.453", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_89.wav", "onoffCaption": "train horn at 0-1", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_93.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_94.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_103.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_104.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_110.wav", "onoffCaption": "whistling at 0-1", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_117.wav", "onoffCaption": "cat meowing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_119.wav", "onoffCaption": "car horn honking at 0.0-2.0", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_120.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_127.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3-5", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_129.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_133.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_134.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_142.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_145.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_151.wav", "onoffCaption": "burping belching at 1-3", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_156.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_158.wav", "onoffCaption": "door knocking at 0-1", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_161.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_166.wav", "onoffCaption": "tapping clicking clanking at 0.032-3.472, 4.758-7.489", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_168.wav", "onoffCaption": "explosion at 2.941-5.813", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_172.wav", "onoffCaption": "gunshot at 0.0-2.0", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_183.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_184.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_197.wav", "onoffCaption": "sheep goat bleating at 0.0-2.0, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_199.wav", "onoffCaption": "dog barking at 0-2, 2-4", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_200.wav", "onoffCaption": "thump thud at 2.224-5.891, 7.389-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_12.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_15.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_22.wav", "onoffCaption": "whistling at 2.603-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_25.wav", "onoffCaption": "explosion at 0.0-2.0, 2.5-4.5", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_31.wav", "onoffCaption": "gunshot at 0.0-2.0, 2.5-4.5, 5.0-7.0", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_36.wav", "onoffCaption": "dog barking at 0.5-1.5", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_38.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_40.wav", "onoffCaption": "sheep goat bleating at 0-1", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_single_identity_test/syn_47.wav", "onoffCaption": "door slamming at 0-1, 2-3, 4-5", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_49.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_51.wav", "onoffCaption": "cat meowing at 0.0-1.0, 2.0-3.0, 4.0-5.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_53.wav", "onoffCaption": "cat meowing at 0-2", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_54.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_63.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3, 4-5", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_70.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_77.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_79.wav", "onoffCaption": "tapping clicking clanking at 0.536-3.976", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_81.wav", "onoffCaption": "spraying at 0-1", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_86.wav", "onoffCaption": "door knocking at 1-2", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_88.wav", "onoffCaption": "cow mooing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_92.wav", "onoffCaption": "train horn at 0.0-2.0, 2.5-4.5", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_95.wav", "onoffCaption": "thump thud at 0.0-1.0", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_102.wav", "onoffCaption": "thump thud at 0-1, 2-3", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_111.wav", "onoffCaption": "door knocking at 0-1, 2-3", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_116.wav", "onoffCaption": "gunshot at 0-1, 2-3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_118.wav", "onoffCaption": "cat meowing at 0-3", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_121.wav", "onoffCaption": "door knocking at 1.155-5.305", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_126.wav", "onoffCaption": "sheep goat bleating at 0.5-2.5, 3.0-5.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_128.wav", "onoffCaption": "tapping clicking clanking at 0-1", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_132.wav", "onoffCaption": "cat meowing at 0-1, 2-3, 4-5", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_135.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_139.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_143.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_144.wav", "onoffCaption": "cow mooing at 1.954-4.602, 6.719-9.729", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_150.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_157.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_159.wav", "onoffCaption": "sneeze at 0-1", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_160.wav", "onoffCaption": "tapping clicking clanking at 1.5-3.5, 5-7", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_167.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_169.wav", "onoffCaption": "train horn at 0-3.5", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_173.wav", "onoffCaption": "thump thud at 0-1", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_174.wav", "onoffCaption": "cat meowing at 0-1.2, 2-3.2, 4-5.2", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_182.wav", "onoffCaption": "door knocking at 1-3, 4-6", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_185.wav", "onoffCaption": "gunshot at 0.0-2.0, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_189.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_191.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_196.wav", "onoffCaption": "door knocking at 0-1, 2-3, 4-5", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_198.wav", "onoffCaption": "explosion at 0.0-1.0", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_10.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_17.wav", "onoffCaption": "burping belching at 0.871-5.871, 7.218-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_19.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_27.wav", "onoffCaption": "spraying at 0-1, 2-3, 4-5", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 1-3, 4-6", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_33.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_34.wav", "onoffCaption": "burping belching at 0-1, 2-3", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_42.wav", "onoffCaption": "dog barking at 2.579-4.579", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_45.wav", "onoffCaption": "cat meowing at 0-1", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_56.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_58.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_61.wav", "onoffCaption": "spraying at 0.0-1.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_66.wav", "onoffCaption": "cat meowing at 0-1, 2-3", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_68.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_72.wav", "onoffCaption": "sneeze at 1.3-2.403, 4.759-6.442", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_75.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_83.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_84.wav", "onoffCaption": "woman laughing at 0-2, 2-4", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_90.wav", "onoffCaption": "dog barking at 0-1, 2-3, 4-5", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_97.wav", "onoffCaption": "gunshot at 0-1", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_99.wav", "onoffCaption": "door knocking at 1-3, 4-6", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_100.wav", "onoffCaption": "burping belching at 0.871-2.871, 3.891-5.891", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_107.wav", "onoffCaption": "woman laughing at 0-2, 5-7", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_113.wav", "onoffCaption": "duck quacking at 0-1", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_114.wav", "onoffCaption": "duck quacking at 0-1, 2-3", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_123.wav", "onoffCaption": "woman laughing at 2.777-6.165", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_124.wav", "onoffCaption": "door slamming at 0.145-1.085, 2.545-4.463", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_130.wav", "onoffCaption": "duck quacking at 0.0-2.0, 3.0-5.0, 6.0-8.0", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_137.wav", "onoffCaption": "dog barking at 0.5-2.5, 3.0-5.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 2.782-5.368", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_146.wav", "onoffCaption": "dog barking at 0-1, 2-3", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_148.wav", "onoffCaption": "thump thud at 2-3", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_152.wav", "onoffCaption": "sheep goat bleating at 0-1, 2-3", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_155.wav", "onoffCaption": "woman laughing at 0.5-2.5, 3.0-5.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_162.wav", "onoffCaption": "door knocking at 0-1, 1-2", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_165.wav", "onoffCaption": "door slamming at 0.355-2.581", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_171.wav", "onoffCaption": "woman laughing at 2.672-5.672", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_176.wav", "onoffCaption": "burping belching at 0.5-3.5, 4.5-7.5", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_178.wav", "onoffCaption": "sheep goat bleating at 1-3, 4-7", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0-3", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_187.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.0-5.0", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_193.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_194.wav", "onoffCaption": "train horn at 0-3", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 1.0-3.0, 4.0-6.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 0.5-1.5, 2.5-3.5", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0-1, 2-3, 4-5", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_single_identity_test/syn_4.wav", "onoffCaption": "car horn honking at 0.664-3.129, 4.357-7.014", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.0-2.0", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_6.wav", "onoffCaption": "explosion at 0-1", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_7.wav", "onoffCaption": "dog barking at 0-1", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_8.wav", "onoffCaption": "burping belching at 2.861-8.462", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_9.wav", "onoffCaption": "burping belching at 0.569-4.438", "frequencyCaption": "burping belching one times"}
diff --git a/picoaudio/data/meta_data/test-onoff-control_multi-event.json b/picoaudio/data/meta_data/test-onoff-control_multi-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..7c57e688ed2805ae7ac07be7b61dc10e52d3b82e
--- /dev/null
+++ b/picoaudio/data/meta_data/test-onoff-control_multi-event.json
@@ -0,0 +1,200 @@
+{"filepath": "data/multi_event_test/syn_1.wav", "onoffCaption": "cat meowing at 0.393-1.783, 3.975-5.365 and whistling at 0.861-5.455 and explosion at 2.089-4.841, 5.738-8.538", "frequencyCaption": "cat meowing two times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_6.wav", "onoffCaption": "whistling at 2.093-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_8.wav", "onoffCaption": "cow mooing at 1.177-3.977, 5.15-7.774", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_11.wav", "onoffCaption": "burping belching at 1.039-3.198, 4.085-6.244 and dog barking at 3.119-5.119", "frequencyCaption": "burping belching two times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_16.wav", "onoffCaption": "duck quacking at 0.799-2.799, 3.634-5.634, 6.976-8.976", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_18.wav", "onoffCaption": "door knocking at 1.225-3.352, 5.173-7.3 and door slamming at 5.439-7.678", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_21.wav", "onoffCaption": "dog barking at 2.947-4.947, 6.186-8.186", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_26.wav", "onoffCaption": "whistling at 2.848-7.442", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_28.wav", "onoffCaption": "cow mooing at 2.639-5.263 and spraying at 8.565-9.667", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_32.wav", "onoffCaption": "duck quacking at 0.039-2.039, 3.171-5.171, 5.938-7.938", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_35.wav", "onoffCaption": "car horn honking at 2.31-5.271, 5.91-8.871", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_43.wav", "onoffCaption": "dog barking at 2.157-4.157, 5.953-7.953 and burping belching at 2.431-5.388, 6.452-8.611 and explosion at 4.8-7.552", "frequencyCaption": "dog barking two times and burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_44.wav", "onoffCaption": "sneeze at 2.638-6.791", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_50.wav", "onoffCaption": "car horn honking at 0.874-3.835, 4.429-7.39 and sneeze at 1.814-5.167 and train horn at 2.818-7.898", "frequencyCaption": "car horn honking two times and sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_57.wav", "onoffCaption": "dog barking at 3.007-5.007, 6.103-8.103 and cow mooing at 3.017-5.641", "frequencyCaption": "dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_59.wav", "onoffCaption": "door slamming at 0.035-2.274 and explosion at 3.857-6.609, 7.377-10.0", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_test/syn_60.wav", "onoffCaption": "train horn at 0.062-3.062", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_67.wav", "onoffCaption": "whistling at 1.616-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_69.wav", "onoffCaption": "door knocking at 0.237-2.801, 4.117-6.681, 7.378-9.942", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_73.wav", "onoffCaption": "door knocking at 0.045-2.172, 2.718-5.282, 6.027-8.591 and sneeze at 2.92-6.273, 6.847-9.032", "frequencyCaption": "door knocking three times and sneeze two times"}
+{"filepath": "data/multi_event_test/syn_74.wav", "onoffCaption": "spraying at 0.38-1.176, 3.06-3.856 and gunshot at 1.729-3.729, 4.367-6.367, 7.031-9.031", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_test/syn_82.wav", "onoffCaption": "dog barking at 0.497-2.497, 4.187-6.187", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_91.wav", "onoffCaption": "gunshot at 0.501-2.501, 3.148-5.148", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_96.wav", "onoffCaption": "door slamming at 0.154-2.393, 3.23-4.641, 5.232-7.471", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_98.wav", "onoffCaption": "thump thud at 1.835-4.135, 6.505-9.18", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_101.wav", "onoffCaption": "dog barking at 0.681-2.681", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_106.wav", "onoffCaption": "burping belching at 0.093-3.05, 3.962-6.121, 7.309-9.468", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_test/syn_108.wav", "onoffCaption": "sneeze at 3.287-7.44", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_112.wav", "onoffCaption": "woman laughing at 1.823-4.587, 6.243-9.007", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_115.wav", "onoffCaption": "duck quacking at 0.044-1.862 and tapping clicking clanking at 0.436-3.876, 5.547-7.6", "frequencyCaption": "duck quacking one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_122.wav", "onoffCaption": "door knocking at 1.266-3.83, 4.854-7.418, 7.929-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_125.wav", "onoffCaption": "cow mooing at 2.954-5.754, 6.384-9.008", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_131.wav", "onoffCaption": "whistling at 0.666-5.26, 5.984-8.335 and cat meowing at 0.904-2.294", "frequencyCaption": "whistling two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_136.wav", "onoffCaption": "sheep goat bleating at 0.226-2.226, 3.707-5.707 and whistling at 1.058-5.652, 6.943-10.0 and woman laughing at 2.749-7.207", "frequencyCaption": "sheep goat bleating two times and whistling two times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_138.wav", "onoffCaption": "gunshot at 0.785-2.785 and tapping clicking clanking at 5.685-9.125", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_140.wav", "onoffCaption": "door knocking at 0.341-2.468, 3.382-5.946, 7.206-9.77", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_147.wav", "onoffCaption": "door slamming at 0.305-1.716, 2.95-4.361, 5.691-7.102", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_149.wav", "onoffCaption": "car horn honking at 0.666-3.35, 5.748-8.432 and spraying at 7.494-8.29, 8.904-9.7", "frequencyCaption": "car horn honking two times and spraying two times"}
+{"filepath": "data/multi_event_test/syn_153.wav", "onoffCaption": "cat meowing at 3.029-4.355", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_154.wav", "onoffCaption": "cat meowing at 2.044-3.37 and door knocking at 2.866-5.43", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_163.wav", "onoffCaption": "sheep goat bleating at 0.139-2.139, 3.188-5.188, 6.077-8.077", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_164.wav", "onoffCaption": "whistling at 0.15-4.744, 6.868-8.971", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_test/syn_170.wav", "onoffCaption": "dog barking at 0.286-2.286, 3.801-5.801", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_177.wav", "onoffCaption": "thump thud at 0.593-2.893 and cow mooing at 4.617-7.241", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_179.wav", "onoffCaption": "cow mooing at 2.754-5.378, 6.145-8.769", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_181.wav", "onoffCaption": "cow mooing at 3.381-6.181, 7.936-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_186.wav", "onoffCaption": "gunshot at 0.131-2.131", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_188.wav", "onoffCaption": "gunshot at 0.785-2.785, 3.847-5.847 and duck quacking at 2.99-4.99", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_192.wav", "onoffCaption": "spraying at 1.763-2.865, 5.335-6.437", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_195.wav", "onoffCaption": "thump thud at 2.422-5.097, 5.945-8.245", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_3.wav", "onoffCaption": "tapping clicking clanking at 2.711-6.151, 7.783-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_4.wav", "onoffCaption": "door slamming at 3.076-4.487, 6.877-8.288", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_13.wav", "onoffCaption": "duck quacking at 0.012-2.012, 3.202-5.202, 7.582-9.582", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_14.wav", "onoffCaption": "sneeze at 1.853-6.006", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_23.wav", "onoffCaption": "sneeze at 0.109-4.262, 6.151-8.608", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_24.wav", "onoffCaption": "woman laughing at 3.051-7.509", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_30.wav", "onoffCaption": "burping belching at 3.234-6.191, 7.597-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_37.wav", "onoffCaption": "thump thud at 1.883-4.558, 6.153-8.453 and door knocking at 2.227-4.791, 5.771-8.335 and burping belching at 6.746-8.905", "frequencyCaption": "thump thud two times and door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_39.wav", "onoffCaption": "train horn at 2.197-5.197, 5.755-8.755", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_41.wav", "onoffCaption": "thump thud at 1.465-3.765", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_test/syn_48.wav", "onoffCaption": "cat meowing at 0.07-1.396, 3.738-5.064, 6.912-8.238", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_52.wav", "onoffCaption": "gunshot at 0.761-2.761 and duck quacking at 0.994-2.994 and tapping clicking clanking at 5.144-8.584", "frequencyCaption": "gunshot one times and duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_55.wav", "onoffCaption": "sneeze at 2.529-6.682, 7.206-9.677", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_62.wav", "onoffCaption": "woman laughing at 0.152-2.916, 5.112-7.934", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_65.wav", "onoffCaption": "gunshot at 3.755-5.755, 6.54-8.54", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_71.wav", "onoffCaption": "door slamming at 0.023-2.262, 4.712-6.123 and whistling at 1.979-6.573", "frequencyCaption": "door slamming two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_76.wav", "onoffCaption": "dog barking at 0.741-2.741", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_test/syn_78.wav", "onoffCaption": "explosion at 0.11-2.862, 4.292-7.044 and duck quacking at 2.338-4.156, 5.898-7.716", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_80.wav", "onoffCaption": "door slamming at 0.695-2.106 and sheep goat bleating at 0.985-2.985", "frequencyCaption": "door slamming one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_85.wav", "onoffCaption": "door knocking at 4.074-6.201", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_87.wav", "onoffCaption": "explosion at 0.371-3.123, 5.335-8.087", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_89.wav", "onoffCaption": "car horn honking at 2.099-5.06 and cat meowing at 5.989-7.315", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_93.wav", "onoffCaption": "dog barking at 0.988-2.988, 5.289-7.289", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_94.wav", "onoffCaption": "gunshot at 1.463-3.463, 4.41-6.41, 7.226-9.226 and thump thud at 1.729-4.404, 6.318-8.993 and sheep goat bleating at 1.895-3.895, 5.909-7.909", "frequencyCaption": "gunshot three times and thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_103.wav", "onoffCaption": "whistling at 2.759-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_104.wav", "onoffCaption": "duck quacking at 4.149-5.967", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_test/syn_110.wav", "onoffCaption": "train horn at 0.111-5.191 and duck quacking at 0.894-2.894 and cow mooing at 5.062-7.862", "frequencyCaption": "train horn one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_117.wav", "onoffCaption": "sheep goat bleating at 3.487-5.487, 7.705-9.705", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_119.wav", "onoffCaption": "train horn at 2.056-5.056 and door knocking at 2.912-5.039, 5.997-8.124", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_120.wav", "onoffCaption": "burping belching at 2.114-5.071, 5.723-8.68", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_127.wav", "onoffCaption": "whistling at 1.653-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_129.wav", "onoffCaption": "door knocking at 0.592-2.719, 3.326-5.453, 6.255-8.382", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_133.wav", "onoffCaption": "duck quacking at 1.444-3.262, 4.595-6.413", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_134.wav", "onoffCaption": "car horn honking at 0.439-3.123, 5.193-7.877", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_142.wav", "onoffCaption": "sneeze at 0.338-4.491, 5.776-7.91", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_145.wav", "onoffCaption": "door knocking at 0.308-2.872, 4.395-6.959 and whistling at 0.583-9.383", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_151.wav", "onoffCaption": "dog barking at 0.368-2.368, 3.112-5.112, 5.983-7.983", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_156.wav", "onoffCaption": "car horn honking at 0.03-2.714, 3.401-6.085, 6.775-9.459", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_158.wav", "onoffCaption": "tapping clicking clanking at 3.057-6.497, 7.876-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_161.wav", "onoffCaption": "spraying at 0.049-1.151, 2.004-2.8", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_test/syn_166.wav", "onoffCaption": "woman laughing at 1.442-5.9", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_168.wav", "onoffCaption": "sheep goat bleating at 0.016-2.016", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_172.wav", "onoffCaption": "door knocking at 0.153-2.28, 3.142-5.706, 6.305-8.869", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_test/syn_175.wav", "onoffCaption": "cow mooing at 0.61-3.41 and spraying at 3.012-4.114", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_183.wav", "onoffCaption": "explosion at 0.192-5.114, 5.844-8.596", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_184.wav", "onoffCaption": "sheep goat bleating at 0.322-2.322", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_190.wav", "onoffCaption": "whistling at 2.571-7.165", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_197.wav", "onoffCaption": "tapping clicking clanking at 1.043-4.483, 5.786-9.226", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_199.wav", "onoffCaption": "duck quacking at 3.246-5.246 and cat meowing at 7.245-8.635", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_200.wav", "onoffCaption": "explosion at 3.045-5.797, 7.133-9.196", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_2.wav", "onoffCaption": "door knocking at 2.42-4.984", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_test/syn_5.wav", "onoffCaption": "burping belching at 3.676-5.835", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_test/syn_12.wav", "onoffCaption": "sheep goat bleating at 1.611-3.611 and sneeze at 5.808-9.161", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_test/syn_15.wav", "onoffCaption": "tapping clicking clanking at 0.807-4.247", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_22.wav", "onoffCaption": "whistling at 3.354-7.948", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_25.wav", "onoffCaption": "burping belching at 2.316-5.273, 6.42-9.377", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_31.wav", "onoffCaption": "woman laughing at 0.674-5.132, 6.464-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_36.wav", "onoffCaption": "door slamming at 0.106-2.345, 2.885-5.124, 5.997-8.236", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_38.wav", "onoffCaption": "cat meowing at 0.245-1.571, 3.125-4.451, 5.016-6.342", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_test/syn_40.wav", "onoffCaption": "door knocking at 2.051-4.178, 4.942-7.506 and cow mooing at 2.928-5.728", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_46.wav", "onoffCaption": "door slamming at 0.382-1.793, 2.674-4.913", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_47.wav", "onoffCaption": "spraying at 0.719-1.515, 2.813-3.915, 4.469-5.265 and cow mooing at 1.592-4.392, 4.998-7.798", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_49.wav", "onoffCaption": "sheep goat bleating at 0.44-2.44, 3.141-5.141 and tapping clicking clanking at 1.283-4.723, 6.144-8.215", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_51.wav", "onoffCaption": "train horn at 0.258-3.258, 4.737-7.277", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_53.wav", "onoffCaption": "dog barking at 0.072-2.072, 3.076-5.076, 6.003-8.003", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_54.wav", "onoffCaption": "train horn at 0.347-3.347, 4.652-7.652", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_test/syn_63.wav", "onoffCaption": "train horn at 0.507-3.507 and cat meowing at 7.463-8.789 and dog barking at 7.612-9.612", "frequencyCaption": "train horn one times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 1.521-3.521", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_70.wav", "onoffCaption": "whistling at 2.267-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_77.wav", "onoffCaption": "cow mooing at 0.75-3.55", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_79.wav", "onoffCaption": "dog barking at 1.282-3.282, 4.117-6.117, 6.789-8.789", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_81.wav", "onoffCaption": "gunshot at 0.019-2.019, 2.851-4.851, 5.918-7.918", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_86.wav", "onoffCaption": "whistling at 1.438-6.032 and woman laughing at 2.351-5.115, 6.601-9.365", "frequencyCaption": "whistling one times and woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_88.wav", "onoffCaption": "sheep goat bleating at 3.021-5.021, 6.26-8.26", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_92.wav", "onoffCaption": "door slamming at 0.346-1.757, 2.569-3.98, 5.839-7.25 and tapping clicking clanking at 2.508-5.948", "frequencyCaption": "door slamming three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_95.wav", "onoffCaption": "door slamming at 2.522-3.933, 5.673-7.084, 8.486-9.897", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_test/syn_102.wav", "onoffCaption": "door knocking at 2.145-4.272, 4.881-7.008", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_105.wav", "onoffCaption": "train horn at 0.682-3.682, 4.465-6.698, 7.809-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_test/syn_111.wav", "onoffCaption": "whistling at 0.032-4.626, 6.182-10.0 and door slamming at 0.753-2.164", "frequencyCaption": "whistling two times and door slamming one times"}
+{"filepath": "data/multi_event_test/syn_116.wav", "onoffCaption": "burping belching at 3.577-5.736, 6.261-9.218", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_118.wav", "onoffCaption": "sneeze at 3.124-6.477", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_test/syn_121.wav", "onoffCaption": "car horn honking at 0.782-3.743, 4.51-7.194, 7.76-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 2.222-4.222, 6.493-8.493", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_126.wav", "onoffCaption": "sneeze at 2.136-6.289 and car horn honking at 2.473-5.434, 7.027-9.711", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_test/syn_128.wav", "onoffCaption": "sheep goat bleating at 0.291-2.291 and door knocking at 0.293-2.42, 3.227-5.791", "frequencyCaption": "sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_test/syn_132.wav", "onoffCaption": "sheep goat bleating at 0.295-2.295 and spraying at 0.328-1.124, 2.065-3.167, 4.421-5.217 and duck quacking at 0.387-2.387, 2.967-4.785, 5.384-7.384", "frequencyCaption": "sheep goat bleating one times and spraying three times and duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_135.wav", "onoffCaption": "tapping clicking clanking at 0.458-3.898, 5.425-8.865", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_139.wav", "onoffCaption": "thump thud at 2.477-4.777, 6.095-8.77", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_test/syn_143.wav", "onoffCaption": "spraying at 2.679-3.475 and explosion at 5.945-10.0", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_test/syn_144.wav", "onoffCaption": "duck quacking at 1.162-2.98, 3.994-5.994, 8.158-9.976", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_test/syn_150.wav", "onoffCaption": "gunshot at 1.946-3.946, 4.6-6.6, 7.322-9.322", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_157.wav", "onoffCaption": "train horn at 1.991-7.071", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_test/syn_159.wav", "onoffCaption": "door slamming at 3.182-5.421, 7.675-9.086", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_160.wav", "onoffCaption": "spraying at 0.179-0.975 and whistling at 3.947-10.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_167.wav", "onoffCaption": "burping belching at 0.386-3.343, 4.105-6.264 and gunshot at 4.772-6.772", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_169.wav", "onoffCaption": "sneeze at 0.56-4.713, 5.69-7.783", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_173.wav", "onoffCaption": "sheep goat bleating at 0.834-2.834, 3.932-5.932, 6.656-8.656", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_174.wav", "onoffCaption": "dog barking at 0.021-2.021, 2.529-4.529, 5.505-7.505", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_176.wav", "onoffCaption": "woman laughing at 2.645-5.409, 7.198-9.435", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_test/syn_182.wav", "onoffCaption": "cow mooing at 0.007-2.807 and gunshot at 1.124-3.124", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_185.wav", "onoffCaption": "spraying at 2.564-3.666 and door knocking at 6.756-9.32", "frequencyCaption": "spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_189.wav", "onoffCaption": "door slamming at 2.717-4.956, 5.586-6.997", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_test/syn_191.wav", "onoffCaption": "burping belching at 2.833-4.992, 6.271-8.43", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_193.wav", "onoffCaption": "cow mooing at 0.942-3.742, 4.83-7.454", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_test/syn_196.wav", "onoffCaption": "spraying at 3.461-4.563", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_test/syn_198.wav", "onoffCaption": "gunshot at 1.546-3.546, 4.501-6.501, 7.428-9.428", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_7.wav", "onoffCaption": "spraying at 0.113-0.909 and burping belching at 0.623-3.58", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_test/syn_9.wav", "onoffCaption": "cow mooing at 1.06-3.86", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_10.wav", "onoffCaption": "door knocking at 0.3-2.864, 5.022-7.586", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_test/syn_17.wav", "onoffCaption": "dog barking at 3.791-5.791, 7.757-9.757", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_19.wav", "onoffCaption": "gunshot at 0.007-2.007 and spraying at 4.251-5.047", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 0.017-3.457, 5.475-7.882", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_27.wav", "onoffCaption": "dog barking at 2.012-4.012, 4.76-6.76", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 2.18-5.62, 6.49-9.93", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_33.wav", "onoffCaption": "dog barking at 2.805-4.805, 5.866-7.866 and car horn honking at 5.136-8.097", "frequencyCaption": "dog barking two times and car horn honking one times"}
+{"filepath": "data/multi_event_test/syn_34.wav", "onoffCaption": "sheep goat bleating at 1.113-3.113", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_42.wav", "onoffCaption": "tapping clicking clanking at 2.443-5.883, 7.179-9.684", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_45.wav", "onoffCaption": "cat meowing at 0.324-1.65 and train horn at 4.186-9.266", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_56.wav", "onoffCaption": "tapping clicking clanking at 1.696-5.136, 6.886-9.533", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_58.wav", "onoffCaption": "door slamming at 2.48-3.891", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_test/syn_61.wav", "onoffCaption": "explosion at 2.489-5.241, 5.792-8.521 and train horn at 2.512-7.592 and woman laughing at 6.424-9.188", "frequencyCaption": "explosion two times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_66.wav", "onoffCaption": "sheep goat bleating at 1.634-3.634", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_test/syn_68.wav", "onoffCaption": "car horn honking at 0.051-3.012, 4.062-6.746, 7.319-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_test/syn_72.wav", "onoffCaption": "spraying at 0.013-0.809, 1.742-2.844 and thump thud at 1.117-3.792 and dog barking at 6.065-8.065", "frequencyCaption": "spraying two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_test/syn_75.wav", "onoffCaption": "explosion at 0.266-5.188, 6.431-9.183", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_test/syn_83.wav", "onoffCaption": "whistling at 2.863-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_test/syn_84.wav", "onoffCaption": "burping belching at 2.009-4.966, 6.768-8.927", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_test/syn_90.wav", "onoffCaption": "gunshot at 0.175-2.175", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_test/syn_97.wav", "onoffCaption": "cat meowing at 3.666-5.056", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_99.wav", "onoffCaption": "duck quacking at 0.697-2.515, 3.677-5.677", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_test/syn_100.wav", "onoffCaption": "cat meowing at 0.122-1.512 and sheep goat bleating at 0.564-2.564, 3.078-5.078, 5.762-7.762", "frequencyCaption": "cat meowing one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_test/syn_107.wav", "onoffCaption": "spraying at 0.005-1.107, 3.385-4.487 and dog barking at 1.269-3.269, 4.85-6.85 and tapping clicking clanking at 1.455-4.895, 5.47-8.91", "frequencyCaption": "spraying two times and dog barking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_test/syn_109.wav", "onoffCaption": "cow mooing at 1.573-4.373 and gunshot at 7.482-9.482", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_113.wav", "onoffCaption": "whistling at 0.12-4.714 and tapping clicking clanking at 0.731-4.171", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_test/syn_114.wav", "onoffCaption": "car horn honking at 3.216-5.9 and door knocking at 3.814-6.378", "frequencyCaption": "car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_test/syn_124.wav", "onoffCaption": "gunshot at 2.794-4.794, 5.712-7.712", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_test/syn_130.wav", "onoffCaption": "dog barking at 0.835-2.835, 3.911-5.911, 6.459-8.459", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_test/syn_137.wav", "onoffCaption": "door knocking at 0.152-2.716 and cow mooing at 1.559-4.183 and gunshot at 5.826-7.826", "frequencyCaption": "door knocking one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_test/syn_141.wav", "onoffCaption": "sneeze at 0.816-4.969, 5.643-9.796", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_146.wav", "onoffCaption": "sneeze at 0.145-4.298, 5.107-8.031 and cat meowing at 1.128-2.454", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_148.wav", "onoffCaption": "duck quacking at 3.185-5.003, 5.701-7.701 and cow mooing at 3.469-6.093", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_152.wav", "onoffCaption": "tapping clicking clanking at 0.851-4.291, 4.863-7.054 and train horn at 5.524-8.524", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_test/syn_155.wav", "onoffCaption": "tapping clicking clanking at 0.869-4.309 and gunshot at 1.402-3.402 and cat meowing at 6.9-8.226", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_test/syn_162.wav", "onoffCaption": "gunshot at 0.5-2.5, 3.074-5.074, 5.829-7.829", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_test/syn_165.wav", "onoffCaption": "thump thud at 0.322-2.622, 4.239-6.914 and whistling at 0.361-9.161", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_test/syn_171.wav", "onoffCaption": "spraying at 1.23-2.332, 3.511-4.613, 5.79-6.892 and thump thud at 1.604-3.904 and sheep goat bleating at 1.985-3.985, 4.796-6.796", "frequencyCaption": "spraying three times and thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_test/syn_178.wav", "onoffCaption": "door slamming at 0.233-2.472 and woman laughing at 6.658-10.0", "frequencyCaption": "door slamming one times and woman laughing one times"}
+{"filepath": "data/multi_event_test/syn_180.wav", "onoffCaption": "spraying at 2.203-3.305 and cow mooing at 4.398-7.198", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_test/syn_187.wav", "onoffCaption": "sneeze at 2.13-6.283, 6.866-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_test/syn_194.wav", "onoffCaption": "duck quacking at 2.028-3.846, 5.612-7.43", "frequencyCaption": "duck quacking two times"}
diff --git a/picoaudio/data/meta_data/test-onoff-control_single-event.json b/picoaudio/data/meta_data/test-onoff-control_single-event.json
new file mode 100644
index 0000000000000000000000000000000000000000..6071ff006d0d2a2618a7aaf573812f9a058ca4ac
--- /dev/null
+++ b/picoaudio/data/meta_data/test-onoff-control_single-event.json
@@ -0,0 +1,400 @@
+{"filepath": "data/single_event_multi_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 0.258-1.584", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_6.wav", "onoffCaption": "tapping clicking clanking at 1.246-4.686", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_8.wav", "onoffCaption": "door slamming at 2.564-4.803", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_11.wav", "onoffCaption": "dog barking at 0.084-2.084, 2.908-4.908", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_16.wav", "onoffCaption": "thump thud at 0.776-3.451", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_18.wav", "onoffCaption": "sheep goat bleating at 3.833-5.833", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_21.wav", "onoffCaption": "sheep goat bleating at 2.491-4.491", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_26.wav", "onoffCaption": "tapping clicking clanking at 0.89-4.33", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_28.wav", "onoffCaption": "sneeze at 0.109-4.262, 6.151-8.608", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_32.wav", "onoffCaption": "cow mooing at 1.486-4.11", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_35.wav", "onoffCaption": "door slamming at 0.085-2.324, 4.153-5.564", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_43.wav", "onoffCaption": "thump thud at 2.551-4.851, 5.601-8.276", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_44.wav", "onoffCaption": "burping belching at 0.979-3.138, 4.115-7.072, 7.609-9.768", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_50.wav", "onoffCaption": "car horn honking at 1.566-4.25, 6.473-9.434", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_57.wav", "onoffCaption": "train horn at 3.341-8.421", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_59.wav", "onoffCaption": "woman laughing at 2.439-5.203, 6.08-8.827", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_60.wav", "onoffCaption": "cat meowing at 0.074-1.464, 3.742-5.068", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 3.535-6.159", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_69.wav", "onoffCaption": "burping belching at 0.799-2.958", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_73.wav", "onoffCaption": "whistling at 2.868-7.462", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_74.wav", "onoffCaption": "cat meowing at 1.655-3.045", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_82.wav", "onoffCaption": "thump thud at 1.925-4.6, 5.398-7.698", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_91.wav", "onoffCaption": "duck quacking at 0.497-2.497", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_96.wav", "onoffCaption": "cat meowing at 0.044-1.37, 3.201-4.591, 5.458-6.848", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_98.wav", "onoffCaption": "woman laughing at 2.458-6.916, 7.905-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_101.wav", "onoffCaption": "burping belching at 1.697-4.654, 5.403-7.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_106.wav", "onoffCaption": "gunshot at 0.047-2.047", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_108.wav", "onoffCaption": "cat meowing at 1.96-3.35, 4.662-5.988", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_112.wav", "onoffCaption": "train horn at 3.416-8.496", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_115.wav", "onoffCaption": "sheep goat bleating at 3.021-5.021, 6.26-8.26", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 1.126-4.566, 6.974-9.783", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_125.wav", "onoffCaption": "car horn honking at 3.106-5.79, 6.31-9.271", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_131.wav", "onoffCaption": "cow mooing at 2.423-5.047, 6.252-9.052", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_136.wav", "onoffCaption": "tapping clicking clanking at 0.672-4.112, 5.733-7.916", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_138.wav", "onoffCaption": "burping belching at 0.093-3.05, 3.962-6.121, 7.309-9.468", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_140.wav", "onoffCaption": "duck quacking at 1.928-3.928, 5.108-6.926", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_147.wav", "onoffCaption": "burping belching at 2.269-4.428, 5.085-8.042", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_149.wav", "onoffCaption": "gunshot at 0.434-2.434", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_153.wav", "onoffCaption": "cow mooing at 3.209-5.833, 6.681-9.481", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_154.wav", "onoffCaption": "train horn at 3.48-6.48, 7.121-9.68", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_163.wav", "onoffCaption": "cow mooing at 1.335-3.959, 6.377-9.177", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_164.wav", "onoffCaption": "door slamming at 3.391-4.802, 5.918-8.157", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_170.wav", "onoffCaption": "whistling at 0.053-8.853", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_177.wav", "onoffCaption": "door knocking at 0.585-2.712, 4.192-6.756", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_179.wav", "onoffCaption": "gunshot at 2.477-4.477", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_181.wav", "onoffCaption": "door knocking at 2.753-5.317", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_186.wav", "onoffCaption": "sneeze at 2.336-6.489, 7.757-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_188.wav", "onoffCaption": "explosion at 1.933-6.855", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_192.wav", "onoffCaption": "cat meowing at 0.139-1.465, 2.845-4.235", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_195.wav", "onoffCaption": "duck quacking at 3.185-5.003, 5.701-7.701", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.203-3.16, 3.696-5.855", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_4.wav", "onoffCaption": "cat meowing at 1.562-2.888", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 0.838-4.278, 4.839-6.935, 7.732-9.827", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_14.wav", "onoffCaption": "tapping clicking clanking at 0.51-3.95, 5.245-8.17", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_23.wav", "onoffCaption": "cow mooing at 0.467-3.267, 4.388-7.012", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_24.wav", "onoffCaption": "thump thud at 3.239-5.539, 6.108-8.783", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_30.wav", "onoffCaption": "explosion at 2.75-5.502, 7.44-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_37.wav", "onoffCaption": "tapping clicking clanking at 2.357-5.797, 7.176-9.79", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_39.wav", "onoffCaption": "burping belching at 1.038-3.197, 4.613-7.57", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_41.wav", "onoffCaption": "car horn honking at 2.524-5.485, 6.594-9.278", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_48.wav", "onoffCaption": "train horn at 2.211-7.291", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_52.wav", "onoffCaption": "dog barking at 2.157-4.157, 5.953-7.953", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_55.wav", "onoffCaption": "spraying at 1.616-2.718, 3.653-4.449, 5.396-6.498", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_62.wav", "onoffCaption": "woman laughing at 0.881-5.339, 6.657-9.421", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_65.wav", "onoffCaption": "tapping clicking clanking at 1.976-5.416, 6.573-9.12", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_71.wav", "onoffCaption": "train horn at 2.442-7.522", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_76.wav", "onoffCaption": "door knocking at 0.618-3.182", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_78.wav", "onoffCaption": "door knocking at 0.065-2.192, 3.439-6.003", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_80.wav", "onoffCaption": "car horn honking at 3.533-6.494", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_85.wav", "onoffCaption": "gunshot at 1.931-3.931, 4.716-6.716, 7.891-9.891", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_87.wav", "onoffCaption": "thump thud at 1.759-4.059, 6.133-8.808", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_89.wav", "onoffCaption": "door knocking at 0.065-2.192, 3.164-5.728", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_93.wav", "onoffCaption": "whistling at 0.042-8.842", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_94.wav", "onoffCaption": "burping belching at 0.167-2.326, 3.873-6.83", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_103.wav", "onoffCaption": "dog barking at 1.282-3.282, 4.117-6.117, 6.789-8.789", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_104.wav", "onoffCaption": "thump thud at 1.988-4.663, 7.028-9.328", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_110.wav", "onoffCaption": "whistling at 1.555-6.149", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_117.wav", "onoffCaption": "tapping clicking clanking at 0.487-3.927", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_119.wav", "onoffCaption": "duck quacking at 2.537-4.355, 5.889-7.889", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_120.wav", "onoffCaption": "dog barking at 0.013-2.013, 3.064-5.064, 5.694-7.694", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_127.wav", "onoffCaption": "duck quacking at 0.78-2.78, 5.24-7.058", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_129.wav", "onoffCaption": "burping belching at 1.965-4.922, 6.696-8.855", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_133.wav", "onoffCaption": "train horn at 3.059-8.139", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_134.wav", "onoffCaption": "spraying at 0.184-0.98, 2.498-3.6, 4.402-5.198", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_142.wav", "onoffCaption": "cow mooing at 2.715-5.339, 6.568-9.368", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_145.wav", "onoffCaption": "cow mooing at 0.071-2.695, 4.586-7.386", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_151.wav", "onoffCaption": "duck quacking at 0.425-2.425, 4.73-6.548", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_156.wav", "onoffCaption": "thump thud at 0.071-2.746, 3.838-6.138, 7.435-9.735", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_158.wav", "onoffCaption": "burping belching at 0.027-2.186", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_161.wav", "onoffCaption": "car horn honking at 0.937-3.898, 5.036-7.72", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_166.wav", "onoffCaption": "burping belching at 0.323-3.28, 4.07-6.229, 7.049-9.208", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_168.wav", "onoffCaption": "door slamming at 0.115-1.526, 2.595-4.834, 5.389-7.628", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_172.wav", "onoffCaption": "woman laughing at 3.125-5.889", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_175.wav", "onoffCaption": "spraying at 0.007-0.803", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_183.wav", "onoffCaption": "woman laughing at 2.259-6.717, 7.786-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_184.wav", "onoffCaption": "door slamming at 3.397-4.808, 6.096-8.335", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_190.wav", "onoffCaption": "explosion at 0.228-5.15, 6.074-8.826", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_197.wav", "onoffCaption": "car horn honking at 3.732-6.416, 7.567-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_199.wav", "onoffCaption": "car horn honking at 1.911-4.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_200.wav", "onoffCaption": "train horn at 0.413-3.413", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 1.299-2.689", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_5.wav", "onoffCaption": "dog barking at 3.791-5.791, 6.571-8.571", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_12.wav", "onoffCaption": "tapping clicking clanking at 1.245-4.685", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_15.wav", "onoffCaption": "explosion at 3.815-6.567, 7.214-9.546", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_22.wav", "onoffCaption": "sheep goat bleating at 0.26-2.26, 3.592-5.592, 7.325-9.325", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_25.wav", "onoffCaption": "gunshot at 0.166-2.166, 3.749-5.749", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_31.wav", "onoffCaption": "sneeze at 3.917-8.07", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_36.wav", "onoffCaption": "sheep goat bleating at 2.86-4.86, 7.119-9.119", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_38.wav", "onoffCaption": "whistling at 2.996-7.59", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_40.wav", "onoffCaption": "woman laughing at 0.024-4.482, 5.882-8.646", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_46.wav", "onoffCaption": "tapping clicking clanking at 2.067-5.507", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_47.wav", "onoffCaption": "cow mooing at 0.008-2.808, 3.956-6.58, 7.995-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_49.wav", "onoffCaption": "dog barking at 3.464-5.464", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_51.wav", "onoffCaption": "whistling at 0.26-9.06", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_53.wav", "onoffCaption": "whistling at 0.748-5.342, 6.45-8.456", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_54.wav", "onoffCaption": "cow mooing at 0.48-3.28, 4.237-6.861", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_63.wav", "onoffCaption": "explosion at 1.214-6.136", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_64.wav", "onoffCaption": "whistling at 0.012-4.606, 5.649-8.052", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_70.wav", "onoffCaption": "dog barking at 4.239-6.239", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_77.wav", "onoffCaption": "train horn at 2.336-7.416", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_79.wav", "onoffCaption": "train horn at 2.15-7.23", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_81.wav", "onoffCaption": "tapping clicking clanking at 3.241-6.681", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_86.wav", "onoffCaption": "gunshot at 0.406-2.406, 4.136-6.136", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_88.wav", "onoffCaption": "car horn honking at 0.051-3.012, 4.062-6.746, 7.319-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_92.wav", "onoffCaption": "door slamming at 1.032-2.443, 4.422-6.661", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_95.wav", "onoffCaption": "woman laughing at 0.147-4.605, 5.939-8.703", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_102.wav", "onoffCaption": "duck quacking at 0.363-2.363, 2.979-4.797", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.253-1.664", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_111.wav", "onoffCaption": "dog barking at 0.562-2.562, 4.25-6.25", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_116.wav", "onoffCaption": "sheep goat bleating at 2.658-4.658", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_118.wav", "onoffCaption": "sheep goat bleating at 2.634-4.634", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_121.wav", "onoffCaption": "cat meowing at 2.182-3.508", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_123.wav", "onoffCaption": "sheep goat bleating at 2.042-4.042, 5.044-7.044", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_126.wav", "onoffCaption": "burping belching at 0.139-3.096, 4.403-6.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_128.wav", "onoffCaption": "train horn at 1.814-4.814", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_132.wav", "onoffCaption": "duck quacking at 1.582-3.582, 4.673-6.491", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_135.wav", "onoffCaption": "whistling at 1.414-6.008, 7.012-9.463", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_139.wav", "onoffCaption": "spraying at 1.819-2.615, 3.181-4.283", "frequencyCaption": "spraying two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_143.wav", "onoffCaption": "door knocking at 0.495-3.059, 4.039-6.166, 7.128-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_144.wav", "onoffCaption": "spraying at 0.584-1.686, 2.49-3.286, 3.892-4.688", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_150.wav", "onoffCaption": "duck quacking at 2.654-4.654", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_157.wav", "onoffCaption": "explosion at 2.478-5.23, 6.261-9.209", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_159.wav", "onoffCaption": "sneeze at 1.342-4.695, 6.662-9.384", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_160.wav", "onoffCaption": "woman laughing at 0.352-3.116", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_167.wav", "onoffCaption": "thump thud at 0.177-2.852, 4.459-6.759", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_169.wav", "onoffCaption": "gunshot at 0.088-2.088", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_173.wav", "onoffCaption": "explosion at 0.195-5.117", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_174.wav", "onoffCaption": "duck quacking at 0.089-2.089, 4.166-5.984", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_176.wav", "onoffCaption": "gunshot at 3.54-5.54, 7.238-9.238", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_182.wav", "onoffCaption": "car horn honking at 0.14-2.824", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_185.wav", "onoffCaption": "dog barking at 3.434-5.434, 6.333-8.333", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_189.wav", "onoffCaption": "burping belching at 0.432-3.389, 4.403-6.562", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_191.wav", "onoffCaption": "tapping clicking clanking at 2.168-5.608", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_193.wav", "onoffCaption": "dog barking at 3.219-5.219", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_196.wav", "onoffCaption": "duck quacking at 1.8-3.618", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_198.wav", "onoffCaption": "sheep goat bleating at 0.073-2.073", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_7.wav", "onoffCaption": "door slamming at 2.809-4.22, 6.263-8.502", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_9.wav", "onoffCaption": "sneeze at 0.07-4.223, 4.927-7.216", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_10.wav", "onoffCaption": "door slamming at 2.191-3.602", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_17.wav", "onoffCaption": "gunshot at 0.033-2.033", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_19.wav", "onoffCaption": "thump thud at 2.571-4.871, 6.726-9.401", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_20.wav", "onoffCaption": "dog barking at 2.557-4.557, 5.093-7.093, 7.963-9.963", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_27.wav", "onoffCaption": "whistling at 2.141-6.735, 7.84-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_29.wav", "onoffCaption": "woman laughing at 3.051-7.509", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_33.wav", "onoffCaption": "dog barking at 1.397-3.397, 5.014-7.014", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_34.wav", "onoffCaption": "dog barking at 0.691-2.691, 4.339-6.339, 7.597-9.597", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_42.wav", "onoffCaption": "door slamming at 0.111-1.522, 2.919-5.158", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_45.wav", "onoffCaption": "woman laughing at 0.913-5.371", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_56.wav", "onoffCaption": "cat meowing at 3.25-4.576", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_58.wav", "onoffCaption": "spraying at 2.012-2.808", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_61.wav", "onoffCaption": "sheep goat bleating at 0.44-2.44, 3.141-5.141", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_66.wav", "onoffCaption": "duck quacking at 0.199-2.199", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_68.wav", "onoffCaption": "door slamming at 0.555-1.966", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_72.wav", "onoffCaption": "duck quacking at 3.008-5.008", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_75.wav", "onoffCaption": "door slamming at 2.007-4.246, 5.403-6.814, 7.324-9.563", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_83.wav", "onoffCaption": "spraying at 0.42-1.522, 2.179-2.975, 4.216-5.012", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_84.wav", "onoffCaption": "burping belching at 1.998-4.955", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_90.wav", "onoffCaption": "whistling at 0.292-9.092", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_97.wav", "onoffCaption": "dog barking at 1.995-3.995", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_99.wav", "onoffCaption": "gunshot at 1.846-3.846, 5.067-7.067", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_100.wav", "onoffCaption": "gunshot at 2.965-4.965, 5.836-7.836", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_107.wav", "onoffCaption": "cat meowing at 0.382-1.772, 4.195-5.521, 7.481-8.871", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 1.827-3.217, 5.396-6.722, 8.387-9.777", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_113.wav", "onoffCaption": "door slamming at 1.281-3.52, 4.645-6.056", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_114.wav", "onoffCaption": "explosion at 2.267-7.189", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_124.wav", "onoffCaption": "woman laughing at 0.666-5.124, 7.521-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_130.wav", "onoffCaption": "gunshot at 2.672-4.672", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_137.wav", "onoffCaption": "train horn at 0.682-3.682, 4.465-6.698, 7.809-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 0.105-2.869", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_146.wav", "onoffCaption": "sneeze at 1.102-4.455", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_148.wav", "onoffCaption": "dog barking at 0.061-2.061, 3.265-5.265, 6.197-8.197", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_multi_identity_test/syn_152.wav", "onoffCaption": "dog barking at 0.127-2.127", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_155.wav", "onoffCaption": "spraying at 3.549-4.651", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_162.wav", "onoffCaption": "explosion at 0.391-3.143, 3.673-5.706", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_165.wav", "onoffCaption": "whistling at 3.448-8.042", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_171.wav", "onoffCaption": "duck quacking at 2.752-4.752", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_178.wav", "onoffCaption": "tapping clicking clanking at 1.713-5.153, 6.827-9.222", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_multi_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0.181-2.981", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_187.wav", "onoffCaption": "explosion at 2.424-5.176", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_multi_identity_test/syn_194.wav", "onoffCaption": "gunshot at 2.339-4.339", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_11.wav", "onoffCaption": "door knocking at 3.808-5.935, 6.708-8.835", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_16.wav", "onoffCaption": "burping belching at 2.569-5.526", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_18.wav", "onoffCaption": "burping belching at 2.907-5.066", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_21.wav", "onoffCaption": "burping belching at 0.64-2.799", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_26.wav", "onoffCaption": "spraying at 0.127-0.923", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_28.wav", "onoffCaption": "train horn at 3.589-8.669", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_32.wav", "onoffCaption": "gunshot at 1.173-3.173, 3.96-5.96, 6.617-8.617", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_35.wav", "onoffCaption": "woman laughing at 0.948-5.406, 7.602-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_43.wav", "onoffCaption": "door slamming at 3.246-4.657, 6.312-7.723", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_44.wav", "onoffCaption": "dog barking at 1.211-3.211, 4.206-6.206, 6.728-8.728", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_50.wav", "onoffCaption": "door knocking at 0.488-3.052, 5.244-7.808", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_57.wav", "onoffCaption": "train horn at 0.177-5.257", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_59.wav", "onoffCaption": "gunshot at 0.24-2.24, 3.277-5.277, 7.394-9.394", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_60.wav", "onoffCaption": "cow mooing at 1.847-4.471, 6.336-8.96", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_67.wav", "onoffCaption": "cow mooing at 2.819-5.443, 6.06-8.684", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_69.wav", "onoffCaption": "burping belching at 1.971-4.928, 6.428-9.385", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_73.wav", "onoffCaption": "dog barking at 0.094-2.094, 3.294-5.294, 6.771-8.771", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_74.wav", "onoffCaption": "cow mooing at 2.351-4.975, 5.558-8.182", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_82.wav", "onoffCaption": "woman laughing at 2.876-7.334", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_85.wav", "onoffCaption": "dog barking at 2.785-4.785", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_91.wav", "onoffCaption": "tapping clicking clanking at 1.295-4.735", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_96.wav", "onoffCaption": "door knocking at 0.452-2.579", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_98.wav", "onoffCaption": "door slamming at 2.339-4.578", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_101.wav", "onoffCaption": "spraying at 0.013-1.115, 1.805-2.907, 5.09-6.192", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_106.wav", "onoffCaption": "spraying at 2.518-3.314", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_108.wav", "onoffCaption": "gunshot at 3.946-5.946, 6.959-8.959", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_112.wav", "onoffCaption": "burping belching at 3.346-6.303, 7.74-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_115.wav", "onoffCaption": "explosion at 0.084-5.006", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_122.wav", "onoffCaption": "tapping clicking clanking at 0.407-3.847", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_125.wav", "onoffCaption": "explosion at 0.371-3.123, 5.335-8.087", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_131.wav", "onoffCaption": "door slamming at 0.346-1.757, 2.569-3.98, 5.839-7.25", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_136.wav", "onoffCaption": "car horn honking at 0.066-2.75", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_138.wav", "onoffCaption": "explosion at 2.129-4.881", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_140.wav", "onoffCaption": "train horn at 1.872-6.952, 7.829-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_147.wav", "onoffCaption": "explosion at 0.38-3.132, 4.352-7.104, 7.977-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/single_event_single_identity_test/syn_149.wav", "onoffCaption": "spraying at 0.031-1.133, 1.86-2.962, 3.961-5.063", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_153.wav", "onoffCaption": "dog barking at 0.435-2.435, 4.016-6.016", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_154.wav", "onoffCaption": "explosion at 1.704-6.626", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_163.wav", "onoffCaption": "sneeze at 2.736-6.889", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_164.wav", "onoffCaption": "sneeze at 2.624-6.777", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_170.wav", "onoffCaption": "burping belching at 3.451-6.408", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_175.wav", "onoffCaption": "explosion at 1.902-6.824", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_177.wav", "onoffCaption": "door knocking at 3.219-5.346, 7.058-9.185", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_179.wav", "onoffCaption": "explosion at 2.521-5.273", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_181.wav", "onoffCaption": "train horn at 0.212-3.212", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_186.wav", "onoffCaption": "sheep goat bleating at 0.651-2.651, 3.512-5.512", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_188.wav", "onoffCaption": "whistling at 0.87-9.67", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_190.wav", "onoffCaption": "woman laughing at 0.484-3.248, 4.163-6.927", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_192.wav", "onoffCaption": "door knocking at 1.863-3.99, 5.187-7.314", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_195.wav", "onoffCaption": "cow mooing at 0.958-3.582, 5.272-7.896", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_13.wav", "onoffCaption": "tapping clicking clanking at 3.109-6.549", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_14.wav", "onoffCaption": "woman laughing at 0.127-2.891", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_23.wav", "onoffCaption": "whistling at 0.074-8.874", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_24.wav", "onoffCaption": "dog barking at 0.91-2.91", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_30.wav", "onoffCaption": "whistling at 0.978-5.572", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_37.wav", "onoffCaption": "whistling at 2.107-6.701", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_39.wav", "onoffCaption": "whistling at 0.165-4.759, 5.362-9.956", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_41.wav", "onoffCaption": "sheep goat bleating at 0.023-2.023, 3.507-5.507", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_46.wav", "onoffCaption": "car horn honking at 1.978-4.939, 5.578-8.539", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_48.wav", "onoffCaption": "thump thud at 1.392-4.067, 5.357-8.032", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_52.wav", "onoffCaption": "dog barking at 0.25-2.25, 3.486-5.486, 6.439-8.439", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_55.wav", "onoffCaption": "gunshot at 2.722-4.722, 6.936-8.936", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_62.wav", "onoffCaption": "burping belching at 0.459-3.416, 4.188-7.145", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_65.wav", "onoffCaption": "sheep goat bleating at 0.55-2.55, 4.457-6.457", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_71.wav", "onoffCaption": "tapping clicking clanking at 3.396-6.836", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_76.wav", "onoffCaption": "sheep goat bleating at 0.056-2.056, 3.47-5.47", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_78.wav", "onoffCaption": "train horn at 0.083-5.163, 6.748-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_80.wav", "onoffCaption": "whistling at 1.269-5.863, 6.498-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/single_event_single_identity_test/syn_87.wav", "onoffCaption": "car horn honking at 1.885-4.569, 5.797-8.481", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_89.wav", "onoffCaption": "train horn at 0.507-3.507", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_93.wav", "onoffCaption": "dog barking at 3.063-5.063, 6.381-8.381", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_94.wav", "onoffCaption": "duck quacking at 2.332-4.15", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_103.wav", "onoffCaption": "gunshot at 1.066-3.066", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_104.wav", "onoffCaption": "cat meowing at 0.488-1.878, 4.297-5.687, 6.263-7.653", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_110.wav", "onoffCaption": "whistling at 0.407-5.001", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_117.wav", "onoffCaption": "cat meowing at 1.091-2.481, 3.509-4.899", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_119.wav", "onoffCaption": "car horn honking at 0.202-2.886", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_120.wav", "onoffCaption": "door knocking at 2.729-5.293", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_127.wav", "onoffCaption": "sheep goat bleating at 2.262-4.262, 5.801-7.801", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_129.wav", "onoffCaption": "sheep goat bleating at 0.602-2.602, 4.548-6.548, 7.151-9.151", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_133.wav", "onoffCaption": "gunshot at 1.679-3.679, 5.98-7.98", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_134.wav", "onoffCaption": "sheep goat bleating at 0.091-2.091, 3.322-5.322", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_142.wav", "onoffCaption": "dog barking at 0.622-2.622, 5.087-7.087", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_145.wav", "onoffCaption": "train horn at 2.269-5.269", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_151.wav", "onoffCaption": "burping belching at 0.193-2.352", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_156.wav", "onoffCaption": "cow mooing at 1.573-4.373", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_158.wav", "onoffCaption": "door knocking at 1.174-3.301", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_161.wav", "onoffCaption": "spraying at 0.159-1.261, 2.033-3.135, 4.44-5.542", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_166.wav", "onoffCaption": "tapping clicking clanking at 1.641-5.081, 6.146-9.586", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_168.wav", "onoffCaption": "explosion at 3.277-8.199", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_172.wav", "onoffCaption": "gunshot at 1.58-3.58", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_183.wav", "onoffCaption": "duck quacking at 0.511-2.511", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_184.wav", "onoffCaption": "spraying at 0.044-0.84", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_197.wav", "onoffCaption": "sheep goat bleating at 2.317-4.317, 6.052-8.052", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_199.wav", "onoffCaption": "dog barking at 3.728-5.728, 6.93-8.93", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_200.wav", "onoffCaption": "thump thud at 1.717-4.017, 5.949-8.249", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_12.wav", "onoffCaption": "dog barking at 2.048-4.048", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_15.wav", "onoffCaption": "dog barking at 0.046-2.046, 4.09-6.09", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_22.wav", "onoffCaption": "whistling at 2.136-6.73", "frequencyCaption": "whistling one times"}
+{"filepath": "data/single_event_single_identity_test/syn_25.wav", "onoffCaption": "explosion at 1.944-4.696, 6.227-8.979", "frequencyCaption": "explosion two times"}
+{"filepath": "data/single_event_single_identity_test/syn_31.wav", "onoffCaption": "gunshot at 0.269-2.269, 3.559-5.559, 6.243-8.243", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/single_event_single_identity_test/syn_36.wav", "onoffCaption": "dog barking at 0.991-2.991", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_38.wav", "onoffCaption": "dog barking at 3.368-5.368, 6.043-8.043", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_40.wav", "onoffCaption": "sheep goat bleating at 0.185-2.185", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/single_event_single_identity_test/syn_47.wav", "onoffCaption": "door slamming at 0.106-2.345, 2.885-5.124, 5.997-8.236", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/single_event_single_identity_test/syn_49.wav", "onoffCaption": "duck quacking at 0.37-2.37", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_51.wav", "onoffCaption": "cat meowing at 0.245-1.571, 3.125-4.451, 5.016-6.342", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_53.wav", "onoffCaption": "cat meowing at 0.277-1.603", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_54.wav", "onoffCaption": "gunshot at 0.17-2.17, 4.644-6.644", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_63.wav", "onoffCaption": "door slamming at 1.788-4.027", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_64.wav", "onoffCaption": "sheep goat bleating at 1.736-3.736, 4.735-6.735, 7.944-9.944", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/single_event_single_identity_test/syn_70.wav", "onoffCaption": "sneeze at 0.231-4.384, 5.433-9.586", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_77.wav", "onoffCaption": "dog barking at 3.416-5.416, 5.973-7.973", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_79.wav", "onoffCaption": "tapping clicking clanking at 0.931-4.371", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_81.wav", "onoffCaption": "spraying at 2.201-2.997", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_86.wav", "onoffCaption": "door knocking at 0.221-2.785", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_88.wav", "onoffCaption": "cow mooing at 2.087-4.887, 6.12-8.92", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_92.wav", "onoffCaption": "train horn at 0.429-5.509, 6.408-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/single_event_single_identity_test/syn_95.wav", "onoffCaption": "thump thud at 2.906-5.581", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_102.wav", "onoffCaption": "thump thud at 2.581-4.881, 6.222-8.522", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/single_event_single_identity_test/syn_105.wav", "onoffCaption": "door slamming at 0.833-3.072, 4.449-6.688", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_111.wav", "onoffCaption": "door knocking at 1.124-3.688, 6.152-8.716", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_116.wav", "onoffCaption": "gunshot at 0.875-2.875, 4.735-6.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_118.wav", "onoffCaption": "cat meowing at 0.483-1.809", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_121.wav", "onoffCaption": "door knocking at 1.619-4.183", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_126.wav", "onoffCaption": "sheep goat bleating at 3.885-5.885, 7.836-9.836", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_128.wav", "onoffCaption": "tapping clicking clanking at 2.571-6.011", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_132.wav", "onoffCaption": "cat meowing at 2.927-4.317, 5.007-6.397, 6.922-8.312", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_135.wav", "onoffCaption": "door slamming at 3.195-5.434, 6.893-9.132", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_139.wav", "onoffCaption": "duck quacking at 2.765-4.583, 6.906-8.724", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_143.wav", "onoffCaption": "cat meowing at 2.231-3.621", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_144.wav", "onoffCaption": "cow mooing at 0.562-3.186, 4.31-6.934", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_150.wav", "onoffCaption": "dog barking at 0.436-2.436", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_157.wav", "onoffCaption": "sneeze at 3.222-7.375", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_159.wav", "onoffCaption": "sneeze at 2.417-6.57", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/single_event_single_identity_test/syn_160.wav", "onoffCaption": "tapping clicking clanking at 0.262-3.702, 5.703-9.143", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_167.wav", "onoffCaption": "cat meowing at 0.205-1.595, 2.703-4.093", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_169.wav", "onoffCaption": "train horn at 3.293-8.373", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_173.wav", "onoffCaption": "thump thud at 3.392-5.692", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_174.wav", "onoffCaption": "cat meowing at 2.478-3.804, 4.701-6.027, 7.098-8.424", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_182.wav", "onoffCaption": "door knocking at 2.598-4.725, 5.428-7.555", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_185.wav", "onoffCaption": "gunshot at 3.329-5.329, 6.811-8.811", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_189.wav", "onoffCaption": "door knocking at 2.566-4.693", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_191.wav", "onoffCaption": "cow mooing at 2.094-4.894", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_196.wav", "onoffCaption": "door knocking at 0.398-2.525, 3.558-5.685, 6.802-8.929", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_198.wav", "onoffCaption": "explosion at 3.575-6.327", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_10.wav", "onoffCaption": "duck quacking at 0.107-1.925", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_17.wav", "onoffCaption": "burping belching at 0.839-2.998, 4.442-6.601", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_19.wav", "onoffCaption": "cat meowing at 2.357-3.683, 5.023-6.349", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_20.wav", "onoffCaption": "tapping clicking clanking at 2.446-5.886, 7.886-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_27.wav", "onoffCaption": "spraying at 0.301-1.403, 2.423-3.525, 4.539-5.641", "frequencyCaption": "spraying three times"}
+{"filepath": "data/single_event_single_identity_test/syn_29.wav", "onoffCaption": "tapping clicking clanking at 0.69-4.13, 5.59-9.03", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_33.wav", "onoffCaption": "train horn at 2.016-7.096", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_34.wav", "onoffCaption": "burping belching at 3.636-5.795, 7.726-9.885", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_42.wav", "onoffCaption": "dog barking at 2.092-4.092", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_45.wav", "onoffCaption": "cat meowing at 2.902-4.228", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_56.wav", "onoffCaption": "train horn at 0.125-3.125", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_58.wav", "onoffCaption": "duck quacking at 0.179-2.179, 4.629-6.629", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_61.wav", "onoffCaption": "spraying at 2.685-3.787", "frequencyCaption": "spraying one times"}
+{"filepath": "data/single_event_single_identity_test/syn_66.wav", "onoffCaption": "cat meowing at 0.1-1.426, 2.691-4.017", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_68.wav", "onoffCaption": "duck quacking at 0.259-2.077", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_72.wav", "onoffCaption": "sneeze at 0.32-3.673, 4.809-8.162", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/single_event_single_identity_test/syn_75.wav", "onoffCaption": "door slamming at 3.048-4.459, 6.382-7.793", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_83.wav", "onoffCaption": "dog barking at 1.005-3.005, 5.367-7.367", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_84.wav", "onoffCaption": "woman laughing at 0.34-4.798, 6.685-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_90.wav", "onoffCaption": "dog barking at 0.965-2.965, 3.842-5.842, 7.713-9.713", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_97.wav", "onoffCaption": "gunshot at 1.924-3.924", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/single_event_single_identity_test/syn_99.wav", "onoffCaption": "door knocking at 3.167-5.294, 6.941-9.068", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_100.wav", "onoffCaption": "burping belching at 2.361-4.52, 5.23-7.389", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_107.wav", "onoffCaption": "woman laughing at 2.849-5.613, 6.83-9.594", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_109.wav", "onoffCaption": "cat meowing at 0.321-1.647, 2.314-3.64, 4.695-6.021", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/single_event_single_identity_test/syn_113.wav", "onoffCaption": "duck quacking at 1.194-3.012", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_114.wav", "onoffCaption": "duck quacking at 0.737-2.737, 3.972-5.972", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_123.wav", "onoffCaption": "woman laughing at 3.064-7.522", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_124.wav", "onoffCaption": "door slamming at 0.317-2.556, 3.904-6.143", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/single_event_single_identity_test/syn_130.wav", "onoffCaption": "duck quacking at 1.714-3.532, 4.074-5.892, 6.517-8.335", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/single_event_single_identity_test/syn_137.wav", "onoffCaption": "dog barking at 0.126-2.126, 2.714-4.714", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_141.wav", "onoffCaption": "woman laughing at 3.098-7.556", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_146.wav", "onoffCaption": "dog barking at 0.087-2.087, 4.127-6.127", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_148.wav", "onoffCaption": "thump thud at 2.712-5.387", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/single_event_single_identity_test/syn_152.wav", "onoffCaption": "sheep goat bleating at 1.645-3.645, 5.29-7.29", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_155.wav", "onoffCaption": "woman laughing at 0.079-4.537, 5.539-9.997", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_162.wav", "onoffCaption": "door knocking at 0.465-2.592, 4.247-6.374", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_165.wav", "onoffCaption": "door slamming at 0.439-2.678", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/single_event_single_identity_test/syn_171.wav", "onoffCaption": "woman laughing at 0.467-3.231", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_176.wav", "onoffCaption": "burping belching at 0.432-2.591, 5.061-7.22", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/single_event_single_identity_test/syn_178.wav", "onoffCaption": "sheep goat bleating at 4.036-6.036, 6.704-8.704", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/single_event_single_identity_test/syn_180.wav", "onoffCaption": "cow mooing at 0.178-2.802", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/single_event_single_identity_test/syn_187.wav", "onoffCaption": "gunshot at 0.523-2.523, 3.427-5.427", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/single_event_single_identity_test/syn_193.wav", "onoffCaption": "tapping clicking clanking at 1.074-4.514, 6.811-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_194.wav", "onoffCaption": "train horn at 1.729-6.809", "frequencyCaption": "train horn one times"}
+{"filepath": "data/single_event_single_identity_test/syn_1.wav", "onoffCaption": "cat meowing at 0.393-1.783, 3.975-5.365", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_2.wav", "onoffCaption": "cat meowing at 2.278-3.668, 5.204-6.594", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/single_event_single_identity_test/syn_3.wav", "onoffCaption": "burping belching at 0.042-2.999, 4.324-7.281, 7.849-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/single_event_single_identity_test/syn_4.wav", "onoffCaption": "car horn honking at 0.38-3.341, 4.605-7.566", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/single_event_single_identity_test/syn_5.wav", "onoffCaption": "dog barking at 0.088-2.088", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_6.wav", "onoffCaption": "explosion at 2.796-7.718", "frequencyCaption": "explosion one times"}
+{"filepath": "data/single_event_single_identity_test/syn_7.wav", "onoffCaption": "dog barking at 2.565-4.565", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/single_event_single_identity_test/syn_8.wav", "onoffCaption": "burping belching at 0.45-3.407", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/single_event_single_identity_test/syn_9.wav", "onoffCaption": "burping belching at 0.775-3.732", "frequencyCaption": "burping belching one times"}
diff --git a/picoaudio/data/meta_data/train.json b/picoaudio/data/meta_data/train.json
new file mode 100644
index 0000000000000000000000000000000000000000..c14872a174e0c402586c8572b0104c8929c5df1e
--- /dev/null
+++ b/picoaudio/data/meta_data/train.json
@@ -0,0 +1,5000 @@
+{"filepath": "data/multi_event_train/syn_21.wav", "onoffCaption": "door slamming at 0.45-1.991, 3.019-5.8, 6.623-8.102", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_74.wav", "onoffCaption": "train horn at 2.817-5.697, 6.941-9.151", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_91.wav", "onoffCaption": "door knocking at 1.155-5.305", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_106.wav", "onoffCaption": "duck quacking at 0.309-2.309, 3.42-5.42, 6.714-8.714 and cow mooing at 2.038-5.007, 6.542-8.58", "frequencyCaption": "duck quacking three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_153.wav", "onoffCaption": "cow mooing at 1.592-4.602, 6.719-9.729 and explosion at 3.329-6.882", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_320.wav", "onoffCaption": "train horn at 3.325-5.48, 6.561-9.201", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_375.wav", "onoffCaption": "whistling at 0.48-4.964, 7.059-9.934", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_390.wav", "onoffCaption": "dog barking at 3.038-5.038, 5.585-7.585", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_505.wav", "onoffCaption": "thump thud at 1.59-6.04, 6.614-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_550.wav", "onoffCaption": "explosion at 0.707-5.707 and woman laughing at 3.44-5.677, 6.23-8.467", "frequencyCaption": "explosion one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_638.wav", "onoffCaption": "door knocking at 1.973-5.029, 6.285-9.132", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_688.wav", "onoffCaption": "burping belching at 0.159-4.028, 6.032-8.977 and door knocking at 0.525-2.902, 4.436-6.813", "frequencyCaption": "burping belching two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_723.wav", "onoffCaption": "burping belching at 0.211-2.336, 2.942-5.466, 6.496-8.59 and dog barking at 0.78-2.78", "frequencyCaption": "burping belching three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_776.wav", "onoffCaption": "woman laughing at 2.782-5.368, 6.831-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_793.wav", "onoffCaption": "explosion at 2.941-5.813 and spraying at 4.494-5.575, 6.097-7.161", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_801.wav", "onoffCaption": "whistling at 2.268-8.349 and door slamming at 2.584-4.78, 5.618-7.814", "frequencyCaption": "whistling one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_854.wav", "onoffCaption": "woman laughing at 0.638-3.925, 5.136-7.364 and dog barking at 1.269-3.269", "frequencyCaption": "woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1035.wav", "onoffCaption": "tapping clicking clanking at 0.961-4.401, 6.37-9.81 and sheep goat bleating at 1.484-4.78 and door knocking at 1.608-4.664, 5.492-7.703", "frequencyCaption": "tapping clicking clanking two times and sheep goat bleating one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1060.wav", "onoffCaption": "spraying at 0.669-1.238, 2.488-3.057, 4.351-4.92 and whistling at 0.918-3.793, 5.909-8.784", "frequencyCaption": "spraying three times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1085.wav", "onoffCaption": "whistling at 2.158-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1209.wav", "onoffCaption": "spraying at 1.431-2.181, 2.756-3.34, 4.475-5.722", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1213.wav", "onoffCaption": "gunshot at 3.324-5.324, 6.563-8.563", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1246.wav", "onoffCaption": "cow mooing at 1.954-6.383, 7.52-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1308.wav", "onoffCaption": "duck quacking at 0.295-2.295, 3.085-5.085, 5.734-7.734 and cat meowing at 3.868-4.879, 5.645-7.2 and cow mooing at 5.281-8.579", "frequencyCaption": "duck quacking three times and cat meowing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1312.wav", "onoffCaption": "dog barking at 3.089-5.089, 6.258-8.258", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1436.wav", "onoffCaption": "train horn at 2.446-5.806, 7.457-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1463.wav", "onoffCaption": "sneeze at 0.309-2.016 and car horn honking at 1.819-4.637 and cow mooing at 7.987-10.0", "frequencyCaption": "sneeze one times and car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1486.wav", "onoffCaption": "sheep goat bleating at 0.657-4.297, 4.952-8.592 and door slamming at 1.974-4.403, 5.259-7.688", "frequencyCaption": "sheep goat bleating two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1578.wav", "onoffCaption": "dog barking at 0.121-2.121, 3.824-5.824, 7.767-9.767", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1610.wav", "onoffCaption": "tapping clicking clanking at 1.851-5.291, 7.569-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1645.wav", "onoffCaption": "door knocking at 0.645-2.772, 3.875-6.782, 7.405-9.692", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1744.wav", "onoffCaption": "spraying at 0.033-1.519, 2.5-3.986, 4.812-6.298 and burping belching at 0.275-7.443", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1829.wav", "onoffCaption": "duck quacking at 0.235-2.235, 4.037-6.037 and burping belching at 1.845-5.024, 6.206-8.841", "frequencyCaption": "duck quacking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1899.wav", "onoffCaption": "woman laughing at 2.777-6.165, 7.557-10.0 and gunshot at 2.778-4.778", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1932.wav", "onoffCaption": "sheep goat bleating at 3.113-5.113, 5.767-7.767 and car horn honking at 3.453-5.453, 7.291-9.291 and spraying at 6.748-7.375", "frequencyCaption": "sheep goat bleating two times and car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1967.wav", "onoffCaption": "train horn at 0.91-4.39 and tapping clicking clanking at 1.016-4.456, 6.672-10.0", "frequencyCaption": "train horn one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1982.wav", "onoffCaption": "tapping clicking clanking at 2.58-6.02, 7.81-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3002.wav", "onoffCaption": "gunshot at 0.025-2.025, 3.214-5.214, 6.486-8.486", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3057.wav", "onoffCaption": "dog barking at 0.464-2.464, 4.19-6.19", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3119.wav", "onoffCaption": "sheep goat bleating at 0.524-2.524, 3.378-5.378, 6.204-8.892", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3224.wav", "onoffCaption": "burping belching at 0.304-2.719 and woman laughing at 6.258-10.0", "frequencyCaption": "burping belching one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3271.wav", "onoffCaption": "door knocking at 2.047-4.422", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3294.wav", "onoffCaption": "cat meowing at 1.674-5.019 and spraying at 2.493-4.188", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3395.wav", "onoffCaption": "whistling at 0.204-5.379, 7.724-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3401.wav", "onoffCaption": "sneeze at 0.283-4.783", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3454.wav", "onoffCaption": "cow mooing at 0.34-3.309", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3627.wav", "onoffCaption": "sneeze at 2.845-4.091", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3672.wav", "onoffCaption": "duck quacking at 2.203-4.203, 5.361-7.361", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3697.wav", "onoffCaption": "cow mooing at 1.882-5.18", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3769.wav", "onoffCaption": "dog barking at 2.579-4.579", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3905.wav", "onoffCaption": "door knocking at 0.141-2.362, 2.877-5.098, 5.687-7.908", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3950.wav", "onoffCaption": "whistling at 2.603-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_149.wav", "onoffCaption": "door knocking at 0.413-2.54, 3.538-5.665, 6.803-8.93 and burping belching at 6.659-9.282", "frequencyCaption": "door knocking three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_221.wav", "onoffCaption": "cow mooing at 2.969-6.267, 7.423-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_250.wav", "onoffCaption": "thump thud at 1.017-4.684, 5.695-9.362", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_274.wav", "onoffCaption": "tapping clicking clanking at 0.404-3.844, 5.186-8.626", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_291.wav", "onoffCaption": "dog barking at 1.244-3.244, 4.164-6.164 and door knocking at 3.598-6.168", "frequencyCaption": "dog barking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_404.wav", "onoffCaption": "gunshot at 0.415-2.545, 4.002-6.132", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_451.wav", "onoffCaption": "car horn honking at 0.154-3.067, 3.794-6.707, 7.909-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_606.wav", "onoffCaption": "cow mooing at 2.907-7.887", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_622.wav", "onoffCaption": "thump thud at 3.172-7.622", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_677.wav", "onoffCaption": "car horn honking at 0.664-3.129, 4.357-7.014", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_692.wav", "onoffCaption": "door slamming at 0.005-1.396 and burping belching at 0.843-4.349, 4.994-7.088 and cat meowing at 6.136-7.136, 8.336-9.336", "frequencyCaption": "door slamming one times and burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_739.wav", "onoffCaption": "car horn honking at 1.773-4.12, 6.106-8.453 and gunshot at 2.478-4.478", "frequencyCaption": "car horn honking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_789.wav", "onoffCaption": "train horn at 0.264-3.064, 3.64-6.44 and door slamming at 4.141-6.361, 7.801-10.0", "frequencyCaption": "train horn two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_900.wav", "onoffCaption": "door slamming at 0.098-1.098, 1.787-3.787, 4.856-7.735", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_955.wav", "onoffCaption": "thump thud at 2.224-5.891, 7.389-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_994.wav", "onoffCaption": "sneeze at 0.59-2.297", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1134.wav", "onoffCaption": "cow mooing at 3.267-6.236", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1145.wav", "onoffCaption": "train horn at 3.352-6.832", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1161.wav", "onoffCaption": "cow mooing at 2.373-5.671, 7.577-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1184.wav", "onoffCaption": "burping belching at 2.861-8.462", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1336.wav", "onoffCaption": "cow mooing at 4.099-7.109, 7.72-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1347.wav", "onoffCaption": "woman laughing at 1.672-3.955", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1408.wav", "onoffCaption": "spraying at 0.161-0.736, 2.006-4.59, 6.215-7.296 and burping belching at 1.503-5.503 and sheep goat bleating at 6.745-8.745", "frequencyCaption": "spraying three times and burping belching one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1479.wav", "onoffCaption": "car horn honking at 2.125-5.779, 7.871-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1513.wav", "onoffCaption": "door knocking at 3.332-5.635, 6.643-9.518", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1537.wav", "onoffCaption": "explosion at 1.773-4.034, 5.15-7.411", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1562.wav", "onoffCaption": "whistling at 3.153-8.328", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1587.wav", "onoffCaption": "cat meowing at 1.159-2.743, 4.625-6.753", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1711.wav", "onoffCaption": "gunshot at 0.349-2.349, 3.25-5.25 and whistling at 7.288-10.0", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1760.wav", "onoffCaption": "explosion at 2.27-5.138, 5.861-8.729", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1785.wav", "onoffCaption": "train horn at 1.794-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1833.wav", "onoffCaption": "thump thud at 0.115-2.454, 3.071-5.41", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1866.wav", "onoffCaption": "door knocking at 3.895-6.625", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1883.wav", "onoffCaption": "door knocking at 0.406-3.253 and woman laughing at 0.635-7.369 and explosion at 1.903-6.903", "frequencyCaption": "door knocking one times and woman laughing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1928.wav", "onoffCaption": "woman laughing at 0.004-2.372, 3.672-6.653", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1998.wav", "onoffCaption": "cat meowing at 0.728-2.283, 3.385-4.94, 5.621-7.176", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3018.wav", "onoffCaption": "burping belching at 2.502-4.625, 6.121-8.244", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3103.wav", "onoffCaption": "thump thud at 1.211-4.878, 7.154-9.492", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3156.wav", "onoffCaption": "cat meowing at 1.931-3.071, 4.724-5.864 and door knocking at 2.857-7.559", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3325.wav", "onoffCaption": "thump thud at 2.22-4.72 and dog barking at 3.589-5.589, 7.361-9.361", "frequencyCaption": "thump thud one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3354.wav", "onoffCaption": "gunshot at 0.26-2.26, 4.679-6.679", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3370.wav", "onoffCaption": "gunshot at 1.812-3.812", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3500.wav", "onoffCaption": "woman laughing at 0.391-3.091, 4.456-7.156", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3555.wav", "onoffCaption": "cow mooing at 1.331-4.3, 5.203-8.172", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3594.wav", "onoffCaption": "train horn at 0.651-4.051, 4.742-8.142", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3619.wav", "onoffCaption": "gunshot at 0.367-2.367", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3668.wav", "onoffCaption": "whistling at 2.959-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3702.wav", "onoffCaption": "gunshot at 1.811-3.811 and duck quacking at 7.641-9.641", "frequencyCaption": "gunshot one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3726.wav", "onoffCaption": "woman laughing at 1.491-3.728 and whistling at 2.496-6.98", "frequencyCaption": "woman laughing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3773.wav", "onoffCaption": "dog barking at 1.003-3.003 and door knocking at 5.52-9.288", "frequencyCaption": "dog barking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3796.wav", "onoffCaption": "cat meowing at 0.365-1.781, 2.545-3.82", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3804.wav", "onoffCaption": "explosion at 3.305-8.305 and spraying at 6.953-7.557", "frequencyCaption": "explosion one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3851.wav", "onoffCaption": "spraying at 0.101-0.952 and whistling at 1.043-6.218, 7.564-10.0", "frequencyCaption": "spraying one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_138.wav", "onoffCaption": "train horn at 3.379-5.819, 6.721-8.914", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_177.wav", "onoffCaption": "car horn honking at 0.74-3.24, 5.316-7.829 and thump thud at 1.929-4.429", "frequencyCaption": "car horn honking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_188.wav", "onoffCaption": "woman laughing at 0.562-2.799 and sheep goat bleating at 2.676-4.676, 6.672-8.672", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_192.wav", "onoffCaption": "thump thud at 3.782-6.829", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_205.wav", "onoffCaption": "cow mooing at 0.72-3.702, 5.102-8.084", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_304.wav", "onoffCaption": "tapping clicking clanking at 3.198-6.638", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_420.wav", "onoffCaption": "door slamming at 0.344-1.597, 3.048-4.026 and thump thud at 7.971-10.0", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_475.wav", "onoffCaption": "gunshot at 0.17-2.17, 3.09-5.09, 6.183-8.183", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_490.wav", "onoffCaption": "explosion at 0.558-5.558, 6.884-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_521.wav", "onoffCaption": "door knocking at 0.623-3.193 and door slamming at 6.292-6.792, 8.204-8.704", "frequencyCaption": "door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_649.wav", "onoffCaption": "door slamming at 3.421-4.594", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_653.wav", "onoffCaption": "explosion at 0.055-5.055, 7.24-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_748.wav", "onoffCaption": "whistling at 0.244-4.728 and explosion at 0.417-3.008 and door slamming at 1.002-2.393, 3.236-4.627", "frequencyCaption": "whistling one times and explosion one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_752.wav", "onoffCaption": "whistling at 1.609-9.994", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_825.wav", "onoffCaption": "thump thud at 3.622-6.122, 7.47-9.97", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_870.wav", "onoffCaption": "explosion at 0.987-4.856, 5.954-8.826", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_924.wav", "onoffCaption": "sneeze at 1.0-3.085, 4.232-6.317", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_971.wav", "onoffCaption": "car horn honking at 0.235-5.142", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1011.wav", "onoffCaption": "cow mooing at 1.805-4.774, 5.439-7.469", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1110.wav", "onoffCaption": "gunshot at 2.921-4.921, 7.411-9.411", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1262.wav", "onoffCaption": "train horn at 0.21-4.65, 6.639-8.799", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1278.wav", "onoffCaption": "gunshot at 3.554-5.554, 7.567-9.567", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1287.wav", "onoffCaption": "dog barking at 3.277-5.277, 6.629-8.629", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1363.wav", "onoffCaption": "spraying at 0.161-1.065, 3.194-4.369, 6.556-7.064", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1379.wav", "onoffCaption": "cat meowing at 3.426-4.692", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1386.wav", "onoffCaption": "door knocking at 1.508-3.668 and cow mooing at 5.206-8.216", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1447.wav", "onoffCaption": "sneeze at 0.818-3.135, 3.983-6.3 and door knocking at 1.957-4.527", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1546.wav", "onoffCaption": "spraying at 1.401-2.252, 3.819-4.67 and whistling at 1.789-6.273, 7.051-10.0", "frequencyCaption": "spraying two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1634.wav", "onoffCaption": "burping belching at 0.231-2.266 and spraying at 5.82-6.57, 8.991-9.858", "frequencyCaption": "burping belching one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1735.wav", "onoffCaption": "tapping clicking clanking at 2.463-5.903", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1817.wav", "onoffCaption": "sneeze at 1.669-3.665, 5.139-8.104", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1842.wav", "onoffCaption": "door slamming at 2.786-3.786", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1858.wav", "onoffCaption": "burping belching at 1.238-3.361", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1916.wav", "onoffCaption": "thump thud at 0.025-4.475", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1943.wav", "onoffCaption": "car horn honking at 0.017-3.192, 3.948-7.123, 7.76-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1959.wav", "onoffCaption": "door slamming at 3.143-5.369, 6.461-7.48", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3026.wav", "onoffCaption": "explosion at 0.682-5.682 and gunshot at 5.466-7.466", "frequencyCaption": "explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3069.wav", "onoffCaption": "car horn honking at 3.111-6.765 and tapping clicking clanking at 4.73-8.17", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3073.wav", "onoffCaption": "train horn at 1.238-5.438", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3096.wav", "onoffCaption": "burping belching at 0.356-3.535", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3127.wav", "onoffCaption": "gunshot at 2.644-5.15, 5.815-7.908", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3168.wav", "onoffCaption": "cow mooing at 3.298-6.28", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3172.wav", "onoffCaption": "sheep goat bleating at 0.162-2.162 and spraying at 4.646-5.497", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3197.wav", "onoffCaption": "cow mooing at 0.258-3.268, 4.919-7.929", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3200.wav", "onoffCaption": "thump thud at 3.356-5.695, 6.815-9.154", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3301.wav", "onoffCaption": "car horn honking at 0.679-3.192 and sheep goat bleating at 0.684-2.684, 5.024-7.024 and gunshot at 0.815-2.815, 3.554-5.554", "frequencyCaption": "car horn honking one times and sheep goat bleating two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3425.wav", "onoffCaption": "sneeze at 3.563-5.88", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3470.wav", "onoffCaption": "duck quacking at 2.5-4.5, 6.729-8.729", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3524.wav", "onoffCaption": "door knocking at 4.11-8.485", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3571.wav", "onoffCaption": "train horn at 2.494-5.134, 7.013-9.653", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3656.wav", "onoffCaption": "door knocking at 0.171-3.724 and door slamming at 7.536-8.685", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3757.wav", "onoffCaption": "woman laughing at 1.808-4.508 and door slamming at 7.615-9.094", "frequencyCaption": "woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3820.wav", "onoffCaption": "gunshot at 0.478-2.478, 4.009-6.009", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3875.wav", "onoffCaption": "sheep goat bleating at 1.313-3.313, 5.182-7.182", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3890.wav", "onoffCaption": "gunshot at 3.266-5.266, 5.827-7.827", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3921.wav", "onoffCaption": "sheep goat bleating at 2.646-4.646", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_50.wav", "onoffCaption": "tapping clicking clanking at 1.23-4.67", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_122.wav", "onoffCaption": "dog barking at 0.436-2.436, 4.526-6.526", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_170.wav", "onoffCaption": "sneeze at 0.528-3.736, 4.834-8.042", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_195.wav", "onoffCaption": "thump thud at 2.18-4.642, 5.753-8.215", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_218.wav", "onoffCaption": "door slamming at 2.198-4.198", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_303.wav", "onoffCaption": "cow mooing at 3.336-8.316 and dog barking at 7.061-9.061", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_351.wav", "onoffCaption": "gunshot at 0.026-2.026 and whistling at 0.049-2.924 and duck quacking at 2.674-4.674, 6.611-8.611", "frequencyCaption": "gunshot one times and whistling one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_526.wav", "onoffCaption": "duck quacking at 2.365-4.365, 5.347-7.347", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_574.wav", "onoffCaption": "door knocking at 2.41-6.026, 6.947-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_591.wav", "onoffCaption": "dog barking at 0.373-2.373 and door knocking at 4.578-8.728", "frequencyCaption": "dog barking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_707.wav", "onoffCaption": "sneeze at 0.383-4.912 and duck quacking at 6.614-8.614", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_755.wav", "onoffCaption": "thump thud at 0.247-4.697, 5.344-7.86", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_822.wav", "onoffCaption": "gunshot at 2.509-4.983, 7.46-9.934 and tapping clicking clanking at 3.058-6.498 and spraying at 3.322-4.103, 5.476-6.208, 7.466-8.116", "frequencyCaption": "gunshot two times and tapping clicking clanking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_895.wav", "onoffCaption": "door slamming at 2.693-4.217, 4.832-6.356 and woman laughing at 3.667-6.752", "frequencyCaption": "door slamming two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_939.wav", "onoffCaption": "tapping clicking clanking at 0.706-4.146 and sheep goat bleating at 2.877-4.877", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1016.wav", "onoffCaption": "whistling at 0.256-8.641 and woman laughing at 0.274-2.372 and gunshot at 3.743-5.743", "frequencyCaption": "whistling one times and woman laughing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1044.wav", "onoffCaption": "door knocking at 1.427-6.427 and thump thud at 2.18-6.63 and duck quacking at 4.935-6.935", "frequencyCaption": "door knocking one times and thump thud one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1237.wav", "onoffCaption": "sheep goat bleating at 1.542-3.542, 4.23-6.23 and thump thud at 3.688-6.188", "frequencyCaption": "sheep goat bleating two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1265.wav", "onoffCaption": "door slamming at 2.016-5.016", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1280.wav", "onoffCaption": "cat meowing at 0.119-1.479, 3.298-4.445, 5.477-6.752", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1412.wav", "onoffCaption": "duck quacking at 3.869-5.869, 7.87-9.87", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1440.wav", "onoffCaption": "whistling at 0.032-2.907, 5.337-8.212", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1509.wav", "onoffCaption": "sheep goat bleating at 2.97-6.89", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1633.wav", "onoffCaption": "door knocking at 0.593-3.13, 4.121-6.748, 7.286-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1661.wav", "onoffCaption": "door slamming at 0.077-1.096", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1683.wav", "onoffCaption": "door slamming at 0.183-1.574, 2.15-3.001, 4.15-6.068 and explosion at 7.733-10.0", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1684.wav", "onoffCaption": "train horn at 1.734-8.203", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1728.wav", "onoffCaption": "cow mooing at 3.139-6.121, 6.742-9.724", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1798.wav", "onoffCaption": "gunshot at 0.031-2.031, 3.728-5.728, 6.569-8.569", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1944.wav", "onoffCaption": "tapping clicking clanking at 2.026-5.466, 6.277-9.717", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3074.wav", "onoffCaption": "explosion at 0.526-2.59 and burping belching at 2.463-4.494, 5.45-7.481 and tapping clicking clanking at 4.892-8.332", "frequencyCaption": "explosion one times and burping belching two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3091.wav", "onoffCaption": "cow mooing at 0.669-5.649, 7.382-10.0 and door knocking at 1.429-3.781", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3207.wav", "onoffCaption": "woman laughing at 3.018-6.594, 7.428-9.674", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3252.wav", "onoffCaption": "woman laughing at 2.935-6.987, 7.556-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3255.wav", "onoffCaption": "cow mooing at 2.968-5.978 and spraying at 5.505-6.255, 7.07-7.82, 9.091-9.841", "frequencyCaption": "cow mooing one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3349.wav", "onoffCaption": "duck quacking at 4.337-6.337", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3422.wav", "onoffCaption": "door knocking at 0.853-4.603, 6.378-8.921", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3495.wav", "onoffCaption": "whistling at 0.504-2.733, 4.408-6.637", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3539.wav", "onoffCaption": "door slamming at 0.962-2.215, 3.921-5.174, 7.427-8.68 and thump thud at 2.092-6.542 and gunshot at 5.373-7.373", "frequencyCaption": "door slamming three times and thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3603.wav", "onoffCaption": "cat meowing at 1.309-2.88", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3604.wav", "onoffCaption": "sneeze at 2.882-5.199, 5.801-8.262", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3651.wav", "onoffCaption": "sheep goat bleating at 2.163-4.163, 5.115-7.115", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3718.wav", "onoffCaption": "door slamming at 0.076-1.591, 2.399-3.25, 3.84-5.131 and duck quacking at 2.94-4.94, 5.948-7.948", "frequencyCaption": "door slamming three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3926.wav", "onoffCaption": "door slamming at 1.11-2.634, 3.498-5.022, 6.331-7.855 and gunshot at 2.786-4.786, 6.387-8.387 and spraying at 7.305-8.305", "frequencyCaption": "door slamming three times and gunshot two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3974.wav", "onoffCaption": "tapping clicking clanking at 1.133-4.573, 5.176-7.376 and burping belching at 3.032-7.032", "frequencyCaption": "tapping clicking clanking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3991.wav", "onoffCaption": "dog barking at 0.342-2.342 and duck quacking at 6.157-8.157", "frequencyCaption": "dog barking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_57.wav", "onoffCaption": "gunshot at 1.054-3.054 and door slamming at 5.275-7.471", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_125.wav", "onoffCaption": "duck quacking at 1.445-3.445, 4.236-6.236, 7.055-9.055 and sneeze at 2.323-4.282", "frequencyCaption": "duck quacking three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_257.wav", "onoffCaption": "car horn honking at 2.061-6.461, 7.737-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_356.wav", "onoffCaption": "gunshot at 0.745-2.745 and cow mooing at 4.159-8.588", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_468.wav", "onoffCaption": "cow mooing at 3.409-6.378", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_472.wav", "onoffCaption": "spraying at 0.583-3.602", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_497.wav", "onoffCaption": "train horn at 0.289-4.409, 6.425-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_569.wav", "onoffCaption": "thump thud at 0.004-2.504, 4.066-6.566", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_573.wav", "onoffCaption": "tapping clicking clanking at 0.237-3.677, 5.378-7.737", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_596.wav", "onoffCaption": "thump thud at 1.237-4.284, 5.393-7.909", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_601.wav", "onoffCaption": "dog barking at 0.134-2.134 and burping belching at 5.233-10.0", "frequencyCaption": "dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_700.wav", "onoffCaption": "dog barking at 1.016-3.016, 5.055-7.055", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_838.wav", "onoffCaption": "car horn honking at 2.969-7.369", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_877.wav", "onoffCaption": "door knocking at 3.48-6.327, 6.917-8.966", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_888.wav", "onoffCaption": "duck quacking at 0.569-2.569, 3.902-5.902, 6.419-8.419", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_892.wav", "onoffCaption": "door slamming at 2.212-5.173, 6.664-7.642", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_976.wav", "onoffCaption": "train horn at 0.105-4.173, 5.738-9.806", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_989.wav", "onoffCaption": "cat meowing at 0.055-1.204, 3.495-4.507, 5.761-7.755", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_993.wav", "onoffCaption": "door knocking at 2.388-6.921 and sheep goat bleating at 2.785-6.081, 7.459-10.0", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1043.wav", "onoffCaption": "cat meowing at 0.138-1.709, 2.9-5.761 and door knocking at 4.303-6.683, 7.227-9.607", "frequencyCaption": "cat meowing two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1059.wav", "onoffCaption": "car horn honking at 0.031-4.431, 4.972-6.972, 7.512-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1142.wav", "onoffCaption": "tapping clicking clanking at 1.363-4.803, 5.875-8.084", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1158.wav", "onoffCaption": "door knocking at 0.327-2.827, 3.933-6.701 and car horn honking at 2.786-7.035", "frequencyCaption": "door knocking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1230.wav", "onoffCaption": "thump thud at 0.0-3.667, 6.045-9.712", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1331.wav", "onoffCaption": "gunshot at 0.021-2.021 and explosion at 0.033-2.762, 5.092-7.821", "frequencyCaption": "gunshot one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1415.wav", "onoffCaption": "door slamming at 0.772-1.577, 2.542-5.516 and sneeze at 1.092-2.592", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1514.wav", "onoffCaption": "door knocking at 0.852-5.291, 6.978-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1666.wav", "onoffCaption": "tapping clicking clanking at 0.654-4.094", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1699.wav", "onoffCaption": "burping belching at 0.365-2.396, 4.748-7.307", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1767.wav", "onoffCaption": "door slamming at 2.159-3.276, 4.526-7.487", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1782.wav", "onoffCaption": "sheep goat bleating at 0.888-2.888, 4.833-6.833 and train horn at 5.074-7.748", "frequencyCaption": "sheep goat bleating two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1810.wav", "onoffCaption": "sneeze at 0.452-1.698 and duck quacking at 0.778-2.778, 3.694-5.694, 6.436-8.436", "frequencyCaption": "sneeze one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1911.wav", "onoffCaption": "woman laughing at 2.363-5.463, 7.505-9.873", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3021.wav", "onoffCaption": "explosion at 0.157-5.157, 6.381-8.972", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3120.wav", "onoffCaption": "dog barking at 0.042-2.042 and woman laughing at 1.67-4.025", "frequencyCaption": "dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3175.wav", "onoffCaption": "burping belching at 3.834-7.093, 7.803-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3248.wav", "onoffCaption": "explosion at 0.563-3.291", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3353.wav", "onoffCaption": "dog barking at 1.301-4.222, 5.025-7.946 and cat meowing at 4.497-7.738", "frequencyCaption": "dog barking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3438.wav", "onoffCaption": "spraying at 1.37-2.151, 3.483-4.052, 4.616-5.467 and door knocking at 7.804-10.0", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3477.wav", "onoffCaption": "dog barking at 1.283-3.283, 5.184-7.184", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3488.wav", "onoffCaption": "burping belching at 0.015-3.559, 4.328-7.328, 7.932-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3492.wav", "onoffCaption": "door knocking at 0.811-5.321 and dog barking at 2.454-4.454 and sneeze at 4.318-6.001", "frequencyCaption": "door knocking one times and dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3576.wav", "onoffCaption": "thump thud at 0.352-4.802 and cat meowing at 7.774-9.722", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3589.wav", "onoffCaption": "duck quacking at 0.293-2.293, 3.357-5.357, 6.355-8.355", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3593.wav", "onoffCaption": "dog barking at 0.036-2.036, 3.013-5.013, 5.719-7.719", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3705.wav", "onoffCaption": "dog barking at 3.735-5.735", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3868.wav", "onoffCaption": "train horn at 0.332-4.772, 6.048-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3872.wav", "onoffCaption": "door slamming at 0.661-1.342, 3.734-4.415, 6.72-7.401", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3897.wav", "onoffCaption": "thump thud at 2.086-4.857", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3969.wav", "onoffCaption": "thump thud at 2.95-5.45, 6.105-8.605", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3973.wav", "onoffCaption": "car horn honking at 2.448-4.961, 6.029-8.542", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3996.wav", "onoffCaption": "burping belching at 2.047-4.082, 4.884-6.919, 7.802-9.837 and explosion at 2.284-7.284 and sneeze at 2.504-3.961, 4.923-6.38, 7.442-8.899", "frequencyCaption": "burping belching three times and explosion one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_18.wav", "onoffCaption": "thump thud at 0.402-2.63, 4.475-6.937", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_69.wav", "onoffCaption": "car horn honking at 1.273-5.114 and cat meowing at 1.976-3.517, 4.196-5.223", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_202.wav", "onoffCaption": "sheep goat bleating at 3.609-5.609, 7.145-9.145", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_273.wav", "onoffCaption": "tapping clicking clanking at 0.069-3.509, 4.304-6.615", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_296.wav", "onoffCaption": "door slamming at 1.008-2.549, 4.839-6.604, 7.937-9.328", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_319.wav", "onoffCaption": "door slamming at 0.803-1.781", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_368.wav", "onoffCaption": "burping belching at 0.832-3.593, 4.807-7.037, 7.772-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_427.wav", "onoffCaption": "cow mooing at 0.73-3.699", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_456.wav", "onoffCaption": "sneeze at 0.319-1.864, 3.011-4.556 and train horn at 7.979-10.0", "frequencyCaption": "sneeze two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_625.wav", "onoffCaption": "door slamming at 1.746-2.863, 3.942-5.466", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_654.wav", "onoffCaption": "tapping clicking clanking at 0.273-3.713, 4.322-7.189, 7.877-10.0", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_849.wav", "onoffCaption": "cow mooing at 2.184-5.482, 7.665-10.0 and woman laughing at 2.751-5.034", "frequencyCaption": "cow mooing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_923.wav", "onoffCaption": "tapping clicking clanking at 3.813-7.253, 7.927-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_952.wav", "onoffCaption": "gunshot at 0.715-2.885 and whistling at 0.92-9.305", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1098.wav", "onoffCaption": "cow mooing at 0.964-3.933, 4.576-7.273", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1117.wav", "onoffCaption": "sheep goat bleating at 0.773-2.773, 3.983-5.983, 6.891-9.811", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1166.wav", "onoffCaption": "cow mooing at 2.671-5.653, 7.472-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1183.wav", "onoffCaption": "door slamming at 2.346-4.346, 5.646-7.646", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1315.wav", "onoffCaption": "sneeze at 1.326-2.938 and door slamming at 6.273-7.251, 8.082-9.06", "frequencyCaption": "sneeze one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1364.wav", "onoffCaption": "car horn honking at 1.483-5.883, 7.161-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1381.wav", "onoffCaption": "whistling at 0.727-8.598 and sneeze at 2.712-4.0", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1530.wav", "onoffCaption": "tapping clicking clanking at 2.798-6.238 and cow mooing at 2.997-6.007, 7.616-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1541.wav", "onoffCaption": "sheep goat bleating at 0.579-2.579 and cat meowing at 3.673-4.684", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1580.wav", "onoffCaption": "woman laughing at 2.859-6.435, 7.53-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1629.wav", "onoffCaption": "dog barking at 0.638-2.638", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1658.wav", "onoffCaption": "spraying at 0.41-1.011, 3.235-3.836, 5.746-6.347 and thump thud at 6.685-9.147", "frequencyCaption": "spraying three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1732.wav", "onoffCaption": "sheep goat bleating at 0.343-2.343, 3.836-5.836", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1743.wav", "onoffCaption": "explosion at 2.23-5.102, 6.265-9.265", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1834.wav", "onoffCaption": "explosion at 3.121-5.993, 6.705-9.577", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1845.wav", "onoffCaption": "cow mooing at 1.29-4.259, 6.477-8.876", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3104.wav", "onoffCaption": "burping belching at 0.345-4.345 and duck quacking at 1.432-3.432", "frequencyCaption": "burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3151.wav", "onoffCaption": "cat meowing at 0.059-1.144", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3190.wav", "onoffCaption": "sneeze at 0.959-5.488, 6.743-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3289.wav", "onoffCaption": "cat meowing at 0.933-4.174 and door knocking at 1.443-3.818 and dog barking at 1.738-3.738", "frequencyCaption": "cat meowing one times and door knocking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3306.wav", "onoffCaption": "tapping clicking clanking at 0.901-4.341 and spraying at 2.023-3.718, 4.646-5.513 and gunshot at 7.421-9.421", "frequencyCaption": "tapping clicking clanking one times and spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3377.wav", "onoffCaption": "whistling at 1.018-3.027, 4.596-7.243 and sheep goat bleating at 1.491-3.491, 4.68-6.68", "frequencyCaption": "whistling two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3392.wav", "onoffCaption": "whistling at 2.377-5.352", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3449.wav", "onoffCaption": "car horn honking at 2.799-7.311 and spraying at 7.997-8.624", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3507.wav", "onoffCaption": "door knocking at 1.768-6.143, 7.28-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3523.wav", "onoffCaption": "cat meowing at 2.864-3.874, 4.793-6.329, 7.107-8.107", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3552.wav", "onoffCaption": "burping belching at 0.815-4.375, 5.148-8.069", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3721.wav", "onoffCaption": "dog barking at 0.194-2.194, 4.11-6.11 and sneeze at 2.888-4.847, 6.691-8.65", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3750.wav", "onoffCaption": "duck quacking at 1.107-3.107, 3.755-5.755, 7.819-9.819", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3827.wav", "onoffCaption": "door slamming at 3.427-4.718, 6.79-8.181 and burping belching at 5.419-8.419", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3856.wav", "onoffCaption": "sneeze at 2.238-4.197 and woman laughing at 6.756-9.361", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_26.wav", "onoffCaption": "whistling at 2.265-9.298", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_154.wav", "onoffCaption": "cow mooing at 0.083-3.065, 4.203-6.641, 7.573-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_226.wav", "onoffCaption": "cow mooing at 3.057-6.355", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_327.wav", "onoffCaption": "duck quacking at 2.283-4.283, 6.514-8.514", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_403.wav", "onoffCaption": "sneeze at 3.343-4.577 and thump thud at 7.198-9.969", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_419.wav", "onoffCaption": "explosion at 0.829-3.429 and spraying at 3.018-3.518, 4.092-4.592, 5.533-6.033", "frequencyCaption": "explosion one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_502.wav", "onoffCaption": "whistling at 0.199-8.584 and sneeze at 3.066-5.312", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_518.wav", "onoffCaption": "duck quacking at 0.924-2.924, 4.185-6.185, 7.586-9.586", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_670.wav", "onoffCaption": "tapping clicking clanking at 0.126-3.566, 4.832-8.272", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_695.wav", "onoffCaption": "thump thud at 0.106-4.024 and door slamming at 0.509-3.29", "frequencyCaption": "thump thud one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_771.wav", "onoffCaption": "whistling at 1.179-9.19", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_794.wav", "onoffCaption": "door slamming at 2.185-5.185", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_806.wav", "onoffCaption": "sheep goat bleating at 0.148-2.148, 3.428-5.428, 6.242-8.242 and dog barking at 1.14-3.14, 5.434-7.434 and spraying at 1.374-2.225", "frequencyCaption": "sheep goat bleating three times and dog barking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_907.wav", "onoffCaption": "door slamming at 0.681-1.486, 2.744-3.549, 4.715-5.52 and car horn honking at 2.057-6.457", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1028.wav", "onoffCaption": "door knocking at 0.558-3.326, 4.653-7.472", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1032.wav", "onoffCaption": "gunshot at 0.97-3.243, 5.737-8.01 and whistling at 1.394-5.878, 6.976-9.839 and dog barking at 6.431-8.431", "frequencyCaption": "gunshot two times and whistling two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1129.wav", "onoffCaption": "duck quacking at 1.938-3.938, 4.687-6.687", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1133.wav", "onoffCaption": "door knocking at 0.162-2.662 and gunshot at 6.396-8.396", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1241.wav", "onoffCaption": "door knocking at 0.891-5.593, 6.815-9.127", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1340.wav", "onoffCaption": "tapping clicking clanking at 0.427-3.867 and train horn at 6.281-9.161", "frequencyCaption": "tapping clicking clanking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1464.wav", "onoffCaption": "sneeze at 1.81-4.127, 5.404-7.323 and dog barking at 2.707-4.707, 6.039-8.039", "frequencyCaption": "sneeze two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1481.wav", "onoffCaption": "gunshot at 2.87-4.87, 6.198-8.198", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1565.wav", "onoffCaption": "explosion at 0.188-2.276, 3.785-5.873 and woman laughing at 1.189-4.577, 5.689-7.781", "frequencyCaption": "explosion two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1617.wav", "onoffCaption": "spraying at 4.123-6.707, 8.0-8.867", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1716.wav", "onoffCaption": "burping belching at 0.962-3.288, 5.002-7.328", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1861.wav", "onoffCaption": "duck quacking at 0.358-2.358, 4.545-6.545 and sneeze at 5.937-7.214", "frequencyCaption": "duck quacking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1884.wav", "onoffCaption": "train horn at 0.134-2.934 and door knocking at 5.52-7.895", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1960.wav", "onoffCaption": "whistling at 2.883-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1985.wav", "onoffCaption": "door knocking at 1.766-3.926, 4.914-7.074", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3050.wav", "onoffCaption": "thump thud at 2.535-6.985", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3223.wav", "onoffCaption": "sneeze at 1.918-3.082", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3239.wav", "onoffCaption": "car horn honking at 1.927-6.834, 7.521-10.0 and door knocking at 3.368-8.368", "frequencyCaption": "car horn honking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3276.wav", "onoffCaption": "thump thud at 0.898-3.398", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3322.wav", "onoffCaption": "door slamming at 3.403-5.403, 7.253-9.253", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3338.wav", "onoffCaption": "door knocking at 0.462-2.65 and car horn honking at 6.561-10.0", "frequencyCaption": "door knocking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3406.wav", "onoffCaption": "sneeze at 0.145-2.064 and thump thud at 4.654-7.116", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3620.wav", "onoffCaption": "train horn at 3.384-6.584", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3675.wav", "onoffCaption": "dog barking at 3.128-5.566, 7.331-9.769", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3690.wav", "onoffCaption": "sneeze at 0.221-1.515, 2.707-4.001, 4.655-5.949 and woman laughing at 4.131-6.736 and door slamming at 4.556-7.273", "frequencyCaption": "sneeze three times and woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3774.wav", "onoffCaption": "woman laughing at 0.469-3.523, 4.471-7.24", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3791.wav", "onoffCaption": "gunshot at 2.794-4.794", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3803.wav", "onoffCaption": "car horn honking at 3.31-5.775", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3819.wav", "onoffCaption": "dog barking at 1.059-3.059, 4.623-7.061", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3902.wav", "onoffCaption": "sneeze at 2.138-3.384, 4.942-6.188, 8.009-9.255", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3918.wav", "onoffCaption": "explosion at 0.429-3.301", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_73.wav", "onoffCaption": "cat meowing at 1.611-6.611", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_96.wav", "onoffCaption": "door knocking at 0.339-4.107, 5.594-8.094", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_101.wav", "onoffCaption": "spraying at 3.192-5.628, 6.912-9.348", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_140.wav", "onoffCaption": "burping belching at 0.155-2.385 and spraying at 1.861-3.594 and cat meowing at 3.753-5.941", "frequencyCaption": "burping belching one times and spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_269.wav", "onoffCaption": "car horn honking at 0.696-5.208, 7.272-9.272", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_372.wav", "onoffCaption": "spraying at 3.375-3.883, 6.226-6.734", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_397.wav", "onoffCaption": "sneeze at 3.122-4.579 and gunshot at 6.347-8.347", "frequencyCaption": "sneeze one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_516.wav", "onoffCaption": "explosion at 2.108-4.636, 5.554-8.082 and cat meowing at 2.607-3.651, 5.572-6.581", "frequencyCaption": "explosion two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_557.wav", "onoffCaption": "cow mooing at 0.457-3.467", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_724.wav", "onoffCaption": "sneeze at 2.444-5.554, 6.077-7.371", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_812.wav", "onoffCaption": "door slamming at 2.125-3.144, 5.228-6.128", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_853.wav", "onoffCaption": "tapping clicking clanking at 0.486-3.926, 5.27-8.038 and spraying at 1.615-2.615, 3.253-4.51, 5.128-7.256", "frequencyCaption": "tapping clicking clanking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_909.wav", "onoffCaption": "tapping clicking clanking at 0.702-4.142, 5.289-7.949", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_948.wav", "onoffCaption": "explosion at 3.11-6.95", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1067.wav", "onoffCaption": "thump thud at 0.112-3.159 and burping belching at 0.191-2.556, 4.378-6.804", "frequencyCaption": "thump thud one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1082.wav", "onoffCaption": "sneeze at 2.318-4.263, 5.955-7.9", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1199.wav", "onoffCaption": "dog barking at 1.241-6.878", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1214.wav", "onoffCaption": "sheep goat bleating at 2.417-4.417, 5.379-7.613 and cow mooing at 5.246-8.228", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1431.wav", "onoffCaption": "thump thud at 0.131-3.178, 5.554-8.167", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1470.wav", "onoffCaption": "spraying at 0.07-1.803, 2.542-3.623, 4.2-4.722", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1495.wav", "onoffCaption": "thump thud at 0.204-3.871 and door knocking at 5.99-10.0", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1642.wav", "onoffCaption": "woman laughing at 1.271-8.716", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1759.wav", "onoffCaption": "cat meowing at 3.734-5.044, 5.722-7.032, 7.879-9.189", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1935.wav", "onoffCaption": "door slamming at 0.387-1.068, 3.499-5.499, 6.392-8.392", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1974.wav", "onoffCaption": "train horn at 0.923-3.363, 4.373-6.429 and sneeze at 4.66-6.16", "frequencyCaption": "train horn two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1991.wav", "onoffCaption": "sneeze at 4.039-5.142, 5.76-7.288 and whistling at 5.589-7.598", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3005.wav", "onoffCaption": "door slamming at 2.507-3.831, 4.569-7.052, 8.333-9.233", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3044.wav", "onoffCaption": "door knocking at 0.268-4.778, 5.999-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3293.wav", "onoffCaption": "sneeze at 0.238-3.886 and train horn at 0.981-4.221, 6.222-8.722 and whistling at 1.282-4.257, 4.777-7.669", "frequencyCaption": "sneeze one times and train horn two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3388.wav", "onoffCaption": "cow mooing at 2.335-5.345, 7.12-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3412.wav", "onoffCaption": "spraying at 0.422-2.55, 3.472-5.6", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3453.wav", "onoffCaption": "dog barking at 0.309-2.309, 2.831-4.831, 5.437-7.437 and tapping clicking clanking at 2.146-5.586, 6.531-9.971", "frequencyCaption": "dog barking three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3509.wav", "onoffCaption": "tapping clicking clanking at 0.602-4.042, 4.582-6.631 and burping belching at 6.007-8.237 and car horn honking at 6.487-8.952", "frequencyCaption": "tapping clicking clanking two times and burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3548.wav", "onoffCaption": "burping belching at 1.491-4.75, 5.696-8.955 and gunshot at 1.925-4.018 and duck quacking at 3.075-5.075", "frequencyCaption": "burping belching two times and gunshot one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3916.wav", "onoffCaption": "burping belching at 0.422-2.453, 4.46-7.46 and explosion at 4.672-7.544", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3957.wav", "onoffCaption": "tapping clicking clanking at 0.344-3.784, 5.183-8.623 and dog barking at 3.175-5.175", "frequencyCaption": "tapping clicking clanking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_32.wav", "onoffCaption": "whistling at 1.978-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_67.wav", "onoffCaption": "sneeze at 1.634-3.947", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_82.wav", "onoffCaption": "sheep goat bleating at 2.09-4.09, 5.408-7.408 and door knocking at 3.653-6.055", "frequencyCaption": "sheep goat bleating two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_115.wav", "onoffCaption": "thump thud at 1.139-3.639, 4.976-7.438, 7.978-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_228.wav", "onoffCaption": "burping belching at 1.339-3.37, 4.917-7.12 and dog barking at 5.396-7.396, 7.946-9.946", "frequencyCaption": "burping belching two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_267.wav", "onoffCaption": "explosion at 0.424-2.517 and door knocking at 6.923-9.111", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_282.wav", "onoffCaption": "train horn at 2.266-4.746", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_298.wav", "onoffCaption": "door slamming at 0.022-0.703, 1.758-3.011", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_333.wav", "onoffCaption": "dog barking at 0.293-2.293 and door slamming at 4.141-4.946, 5.517-7.88, 8.508-9.189", "frequencyCaption": "dog barking one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_366.wav", "onoffCaption": "spraying at 2.509-3.017 and cat meowing at 6.585-8.169", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_383.wav", "onoffCaption": "door knocking at 0.352-3.727, 4.364-7.739", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_399.wav", "onoffCaption": "cow mooing at 2.658-5.627, 7.837-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_458.wav", "onoffCaption": "whistling at 1.057-9.068 and explosion at 4.286-7.158", "frequencyCaption": "whistling one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_543.wav", "onoffCaption": "cow mooing at 1.056-4.038, 5.843-8.812", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_631.wav", "onoffCaption": "gunshot at 2.915-4.915, 6.448-8.448", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_730.wav", "onoffCaption": "cat meowing at 3.413-4.562, 6.613-7.762", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_765.wav", "onoffCaption": "car horn honking at 0.945-4.786, 5.97-9.811", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_780.wav", "onoffCaption": "whistling at 0.72-8.09", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_847.wav", "onoffCaption": "burping belching at 1.931-9.099", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1026.wav", "onoffCaption": "tapping clicking clanking at 2.614-6.054 and gunshot at 2.933-4.933, 6.04-8.04 and door slamming at 3.285-4.402", "frequencyCaption": "tapping clicking clanking one times and gunshot two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1073.wav", "onoffCaption": "train horn at 2.629-5.949, 6.477-9.797", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1096.wav", "onoffCaption": "door slamming at 2.215-4.215 and whistling at 7.117-10.0", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1168.wav", "onoffCaption": "duck quacking at 0.384-2.384", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1200.wav", "onoffCaption": "tapping clicking clanking at 0.097-3.537, 4.07-6.583 and train horn at 3.145-6.679", "frequencyCaption": "tapping clicking clanking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1255.wav", "onoffCaption": "woman laughing at 0.763-4.151 and cow mooing at 5.988-8.97", "frequencyCaption": "woman laughing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1425.wav", "onoffCaption": "gunshot at 0.046-2.046, 2.925-4.925, 5.487-7.487", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1603.wav", "onoffCaption": "sneeze at 0.369-1.533", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1656.wav", "onoffCaption": "dog barking at 0.624-2.624", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1718.wav", "onoffCaption": "woman laughing at 1.553-3.651", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1921.wav", "onoffCaption": "duck quacking at 0.196-2.196 and tapping clicking clanking at 4.003-7.443", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3011.wav", "onoffCaption": "door knocking at 0.914-3.484 and sheep goat bleating at 2.889-4.889, 6.023-8.023", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3237.wav", "onoffCaption": "door knocking at 2.281-4.53 and cow mooing at 7.956-10.0", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3262.wav", "onoffCaption": "cat meowing at 1.237-4.582, 5.9-7.26 and spraying at 2.83-3.562, 4.76-5.492, 7.812-8.544 and burping belching at 4.274-7.476", "frequencyCaption": "cat meowing two times and spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3278.wav", "onoffCaption": "duck quacking at 3.644-5.644, 7.565-9.565", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3287.wav", "onoffCaption": "sheep goat bleating at 0.66-2.66, 3.969-5.969", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3363.wav", "onoffCaption": "dog barking at 2.469-4.469, 4.984-6.984 and sneeze at 5.02-6.52", "frequencyCaption": "dog barking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3379.wav", "onoffCaption": "cow mooing at 2.657-5.667, 6.343-9.006", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3447.wav", "onoffCaption": "car horn honking at 0.089-3.308 and cow mooing at 0.324-4.753, 6.437-9.419", "frequencyCaption": "car horn honking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3634.wav", "onoffCaption": "cow mooing at 0.997-5.426, 6.53-10.0 and duck quacking at 2.647-4.647, 5.454-7.454", "frequencyCaption": "cow mooing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3661.wav", "onoffCaption": "dog barking at 1.926-4.326, 6.456-8.456", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3684.wav", "onoffCaption": "duck quacking at 2.617-4.617, 5.935-7.935", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3735.wav", "onoffCaption": "door knocking at 1.259-5.634 and tapping clicking clanking at 3.553-6.993", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3858.wav", "onoffCaption": "door knocking at 2.954-5.203, 5.759-8.008", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3943.wav", "onoffCaption": "spraying at 0.374-2.81, 4.757-5.507", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_28.wav", "onoffCaption": "thump thud at 1.425-4.472, 5.233-8.28", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_98.wav", "onoffCaption": "sneeze at 3.369-5.482 and sheep goat bleating at 7.725-9.725", "frequencyCaption": "sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_232.wav", "onoffCaption": "train horn at 2.722-9.191", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_329.wav", "onoffCaption": "woman laughing at 0.281-2.386, 3.414-5.66 and gunshot at 0.769-2.769", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_417.wav", "onoffCaption": "tapping clicking clanking at 0.114-3.554 and door slamming at 5.506-7.03, 8.007-9.007", "frequencyCaption": "tapping clicking clanking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_442.wav", "onoffCaption": "car horn honking at 0.262-3.175 and spraying at 2.001-3.763, 4.393-4.915", "frequencyCaption": "car horn honking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_559.wav", "onoffCaption": "door slamming at 0.382-2.84, 4.65-6.129, 7.21-9.436 and spraying at 1.394-2.144, 3.499-4.249, 5.971-6.721", "frequencyCaption": "door slamming three times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_664.wav", "onoffCaption": "car horn honking at 2.537-7.049", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_681.wav", "onoffCaption": "duck quacking at 3.168-5.168, 6.271-8.271", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_808.wav", "onoffCaption": "explosion at 3.702-7.542", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_913.wav", "onoffCaption": "explosion at 1.558-4.43, 5.523-8.276", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_946.wav", "onoffCaption": "duck quacking at 0.315-2.315, 4.783-6.783 and dog barking at 4.076-6.076", "frequencyCaption": "duck quacking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1069.wav", "onoffCaption": "train horn at 1.208-4.448, 5.077-7.082", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1127.wav", "onoffCaption": "car horn honking at 0.006-2.792 and spraying at 6.323-6.845, 7.424-7.946", "frequencyCaption": "car horn honking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1172.wav", "onoffCaption": "gunshot at 0.688-2.688, 4.069-6.069", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1197.wav", "onoffCaption": "train horn at 0.206-3.006, 4.395-7.195", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1301.wav", "onoffCaption": "cow mooing at 1.996-6.425", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1354.wav", "onoffCaption": "cow mooing at 1.681-4.663", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1524.wav", "onoffCaption": "dog barking at 0.558-2.558, 3.395-5.395, 6.074-8.074 and sheep goat bleating at 1.962-3.962", "frequencyCaption": "dog barking three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1571.wav", "onoffCaption": "thump thud at 0.702-2.93, 4.42-7.262 and explosion at 1.683-4.555 and burping belching at 2.168-5.427", "frequencyCaption": "thump thud two times and explosion one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1594.wav", "onoffCaption": "thump thud at 0.537-4.987, 6.972-9.172", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1619.wav", "onoffCaption": "dog barking at 2.648-5.086, 5.864-8.302 and woman laughing at 6.145-8.227", "frequencyCaption": "dog barking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1702.wav", "onoffCaption": "cat meowing at 2.721-6.066, 6.608-7.693, 8.471-9.618", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1757.wav", "onoffCaption": "gunshot at 1.51-3.51, 5.146-7.386 and thump thud at 2.06-4.56", "frequencyCaption": "gunshot two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1796.wav", "onoffCaption": "tapping clicking clanking at 0.181-3.621 and duck quacking at 2.364-4.364 and train horn at 7.722-10.0", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1820.wav", "onoffCaption": "cat meowing at 0.661-4.006, 4.781-6.047, 6.6-7.685 and woman laughing at 6.484-9.569", "frequencyCaption": "cat meowing three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1875.wav", "onoffCaption": "dog barking at 2.881-4.881 and cow mooing at 7.22-10.0", "frequencyCaption": "dog barking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1890.wav", "onoffCaption": "spraying at 0.707-1.648, 3.552-4.493, 6.603-7.544", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3110.wav", "onoffCaption": "cat meowing at 0.442-2.436 and duck quacking at 1.45-3.45, 3.969-5.969, 6.517-8.517", "frequencyCaption": "cat meowing one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3145.wav", "onoffCaption": "train horn at 0.327-6.796 and door knocking at 6.15-8.525", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3336.wav", "onoffCaption": "thump thud at 2.128-4.628, 5.384-8.155", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3347.wav", "onoffCaption": "explosion at 3.829-7.669", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3386.wav", "onoffCaption": "door knocking at 2.413-6.029, 6.842-9.217", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3408.wav", "onoffCaption": "explosion at 3.363-8.363", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3513.wav", "onoffCaption": "woman laughing at 1.221-5.273, 6.213-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3546.wav", "onoffCaption": "gunshot at 2.824-4.824", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3711.wav", "onoffCaption": "spraying at 0.063-0.585, 1.438-1.96, 3.397-3.919 and cow mooing at 1.311-4.293, 6.776-9.397", "frequencyCaption": "spraying three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3760.wav", "onoffCaption": "duck quacking at 0.042-2.042, 3.248-5.248, 6.411-8.411", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3785.wav", "onoffCaption": "tapping clicking clanking at 0.561-4.001, 6.477-9.917", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3817.wav", "onoffCaption": "spraying at 0.006-0.61", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3842.wav", "onoffCaption": "sneeze at 0.021-1.521, 2.134-3.634", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3959.wav", "onoffCaption": "burping belching at 1.098-4.098, 5.432-7.873", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_59.wav", "onoffCaption": "train horn at 1.154-5.484 and cat meowing at 8.009-9.021", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_164.wav", "onoffCaption": "burping belching at 0.151-4.151, 6.609-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_181.wav", "onoffCaption": "sneeze at 1.324-4.972, 6.303-9.951", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_216.wav", "onoffCaption": "thump thud at 0.386-2.886, 4.342-6.391 and door slamming at 0.64-1.779", "frequencyCaption": "thump thud two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_243.wav", "onoffCaption": "gunshot at 0.792-2.792, 4.535-6.535, 7.287-9.287", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_358.wav", "onoffCaption": "door knocking at 2.31-6.843", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_429.wav", "onoffCaption": "dog barking at 0.504-2.504, 3.032-5.032 and sheep goat bleating at 7.803-9.803", "frequencyCaption": "dog barking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_433.wav", "onoffCaption": "duck quacking at 0.816-2.816 and door knocking at 5.455-9.205", "frequencyCaption": "duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_466.wav", "onoffCaption": "duck quacking at 0.463-2.463, 3.794-5.794, 7.209-9.209", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_483.wav", "onoffCaption": "whistling at 3.824-6.799, 7.476-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_528.wav", "onoffCaption": "tapping clicking clanking at 2.682-6.122, 7.378-9.544", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_598.wav", "onoffCaption": "whistling at 0.363-2.592 and woman laughing at 6.028-8.311", "frequencyCaption": "whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_615.wav", "onoffCaption": "whistling at 0.199-2.428", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_640.wav", "onoffCaption": "thump thud at 0.872-4.79, 5.882-9.8", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_836.wav", "onoffCaption": "whistling at 1.112-6.226, 7.367-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_879.wav", "onoffCaption": "door knocking at 2.646-5.702, 7.465-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_937.wav", "onoffCaption": "dog barking at 4.0-6.0, 7.139-9.139", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_962.wav", "onoffCaption": "cat meowing at 0.143-5.143, 6.242-10.0 and car horn honking at 1.548-3.895", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_987.wav", "onoffCaption": "car horn honking at 0.823-4.477, 6.669-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1018.wav", "onoffCaption": "car horn honking at 0.255-4.767, 5.748-8.386", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1103.wav", "onoffCaption": "car horn honking at 3.214-7.536", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1156.wav", "onoffCaption": "whistling at 3.274-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1325.wav", "onoffCaption": "sneeze at 2.663-5.124, 6.447-8.908 and gunshot at 3.303-5.303, 6.37-8.37", "frequencyCaption": "sneeze two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1370.wav", "onoffCaption": "cow mooing at 0.829-5.258, 6.489-9.114", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1395.wav", "onoffCaption": "duck quacking at 2.874-4.874, 7.229-9.229", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1500.wav", "onoffCaption": "thump thud at 1.523-4.294 and burping belching at 3.024-6.283", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1555.wav", "onoffCaption": "whistling at 0.923-3.798, 5.507-7.736 and gunshot at 2.133-4.133", "frequencyCaption": "whistling two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1668.wav", "onoffCaption": "burping belching at 0.939-4.141 and car horn honking at 2.97-5.788, 7.057-9.875", "frequencyCaption": "burping belching one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1726.wav", "onoffCaption": "explosion at 2.333-7.333", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1773.wav", "onoffCaption": "woman laughing at 2.701-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1804.wav", "onoffCaption": "explosion at 0.379-4.379, 6.146-8.899 and cat meowing at 0.504-1.516 and whistling at 3.877-9.052", "frequencyCaption": "explosion two times and cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1851.wav", "onoffCaption": "car horn honking at 3.222-6.008, 7.231-9.239", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1950.wav", "onoffCaption": "whistling at 3.046-8.221 and sheep goat bleating at 7.003-9.003", "frequencyCaption": "whistling one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3060.wav", "onoffCaption": "door knocking at 1.078-5.911, 7.79-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3134.wav", "onoffCaption": "thump thud at 2.316-4.655, 6.441-8.89", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3161.wav", "onoffCaption": "thump thud at 1.467-4.514, 5.451-8.498", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3184.wav", "onoffCaption": "thump thud at 0.155-2.617, 4.858-7.467 and dog barking at 6.665-8.665", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3209.wav", "onoffCaption": "car horn honking at 0.261-4.102, 5.871-9.712", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3312.wav", "onoffCaption": "door knocking at 0.796-3.108 and thump thud at 6.247-10.0", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3436.wav", "onoffCaption": "door knocking at 2.451-6.219, 6.734-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3479.wav", "onoffCaption": "car horn honking at 0.4-2.865", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3537.wav", "onoffCaption": "duck quacking at 0.27-2.27 and car horn honking at 6.138-10.0", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3562.wav", "onoffCaption": "sneeze at 2.315-3.592, 5.084-6.372, 7.543-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3587.wav", "onoffCaption": "car horn honking at 1.174-3.687", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3744.wav", "onoffCaption": "tapping clicking clanking at 1.635-5.075, 5.921-9.361", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3829.wav", "onoffCaption": "duck quacking at 0.57-2.57, 3.69-5.69", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3833.wav", "onoffCaption": "duck quacking at 2.933-4.933, 6.082-8.082", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3866.wav", "onoffCaption": "cow mooing at 0.249-3.259 and sheep goat bleating at 2.129-4.129, 4.999-6.999", "frequencyCaption": "cow mooing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3883.wav", "onoffCaption": "whistling at 0.366-8.377", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3928.wav", "onoffCaption": "gunshot at 2.58-4.58, 6.96-8.96 and door knocking at 3.521-5.898", "frequencyCaption": "gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3998.wav", "onoffCaption": "car horn honking at 1.23-4.817, 7.133-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_16.wav", "onoffCaption": "woman laughing at 1.302-3.894, 4.898-7.49", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_43.wav", "onoffCaption": "sheep goat bleating at 1.381-3.381, 4.024-6.942, 7.88-9.88", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_131.wav", "onoffCaption": "train horn at 2.615-5.975", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_259.wav", "onoffCaption": "dog barking at 1.087-3.087, 5.583-7.583", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_317.wav", "onoffCaption": "door slamming at 0.245-2.01, 3.859-5.624 and explosion at 1.731-6.731", "frequencyCaption": "door slamming two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_342.wav", "onoffCaption": "woman laughing at 0.341-2.907, 4.104-6.67, 7.981-10.0 and train horn at 3.333-5.973", "frequencyCaption": "woman laughing three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_499.wav", "onoffCaption": "woman laughing at 0.549-3.634, 4.963-7.2", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_532.wav", "onoffCaption": "woman laughing at 2.264-4.856, 5.681-7.7", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_567.wav", "onoffCaption": "whistling at 0.302-8.052", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_582.wav", "onoffCaption": "sheep goat bleating at 0.147-2.147, 4.118-6.118", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_714.wav", "onoffCaption": "door slamming at 1.142-3.142, 5.231-7.231", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_741.wav", "onoffCaption": "explosion at 2.483-5.483, 5.99-8.99", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_863.wav", "onoffCaption": "spraying at 2.737-4.432, 5.492-7.187, 7.813-9.508", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_886.wav", "onoffCaption": "dog barking at 0.055-2.055, 3.364-5.364, 6.632-8.632 and sneeze at 2.661-4.978", "frequencyCaption": "dog barking three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_978.wav", "onoffCaption": "duck quacking at 3.758-5.758, 6.667-8.667", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1002.wav", "onoffCaption": "cat meowing at 2.9-5.804, 8.273-9.583", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1057.wav", "onoffCaption": "train horn at 0.616-3.056 and woman laughing at 2.221-5.502", "frequencyCaption": "train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1119.wav", "onoffCaption": "gunshot at 0.328-2.328, 3.264-5.264, 6.146-8.146", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1224.wav", "onoffCaption": "sheep goat bleating at 1.04-3.04", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1271.wav", "onoffCaption": "woman laughing at 3.563-6.658, 7.737-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1293.wav", "onoffCaption": "burping belching at 1.074-4.276, 5.175-7.481", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1294.wav", "onoffCaption": "dog barking at 0.663-2.663, 3.258-5.258 and explosion at 0.759-5.759 and sheep goat bleating at 1.279-3.279, 5.169-7.169", "frequencyCaption": "dog barking two times and explosion one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1388.wav", "onoffCaption": "whistling at 0.114-8.125 and gunshot at 1.695-3.695, 4.819-6.819 and explosion at 4.022-7.022", "frequencyCaption": "whistling one times and gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1401.wav", "onoffCaption": "thump thud at 0.379-2.841, 3.369-5.708, 6.811-9.748", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1454.wav", "onoffCaption": "door knocking at 2.104-4.941 and explosion at 5.967-10.0", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1627.wav", "onoffCaption": "thump thud at 0.558-3.058, 4.154-6.977", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1672.wav", "onoffCaption": "cat meowing at 2.314-7.314 and spraying at 5.588-8.048", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1697.wav", "onoffCaption": "woman laughing at 0.865-3.634, 4.773-7.73", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1769.wav", "onoffCaption": "tapping clicking clanking at 0.115-3.555 and duck quacking at 5.928-7.928 and dog barking at 6.357-8.795", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1905.wav", "onoffCaption": "thump thud at 2.283-4.783, 5.933-8.433 and sneeze at 7.176-9.095", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3035.wav", "onoffCaption": "whistling at 1.36-6.535", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3085.wav", "onoffCaption": "tapping clicking clanking at 2.948-6.388, 7.912-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3213.wav", "onoffCaption": "door slamming at 0.257-1.736 and whistling at 0.27-2.279, 3.595-6.47", "frequencyCaption": "door slamming one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3214.wav", "onoffCaption": "thump thud at 2.382-4.721, 5.845-8.607", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3246.wav", "onoffCaption": "tapping clicking clanking at 0.331-3.771 and cow mooing at 6.725-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3308.wav", "onoffCaption": "woman laughing at 0.073-5.112 and duck quacking at 0.649-2.649", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3463.wav", "onoffCaption": "dog barking at 0.095-2.095, 3.294-5.294, 6.96-8.96", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3486.wav", "onoffCaption": "sheep goat bleating at 0.432-3.432, 4.38-6.38", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3578.wav", "onoffCaption": "dog barking at 0.095-2.495, 3.497-5.935, 7.9-9.9", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3610.wav", "onoffCaption": "explosion at 0.757-3.757, 5.449-8.449", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3642.wav", "onoffCaption": "door knocking at 0.246-2.746, 5.236-7.736", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3645.wav", "onoffCaption": "tapping clicking clanking at 1.919-5.359", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3759.wav", "onoffCaption": "door knocking at 2.919-5.649, 6.984-9.048", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3899.wav", "onoffCaption": "woman laughing at 2.366-5.135, 6.523-9.277", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3932.wav", "onoffCaption": "woman laughing at 0.413-2.696, 3.687-6.456, 7.42-9.645", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3967.wav", "onoffCaption": "woman laughing at 0.22-2.312, 3.688-5.78", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3982.wav", "onoffCaption": "whistling at 0.296-5.796, 6.441-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_11.wav", "onoffCaption": "spraying at 0.048-0.675, 2.079-3.163 and explosion at 6.554-10.0", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_44.wav", "onoffCaption": "sheep goat bleating at 0.24-2.24, 3.277-5.277, 7.394-9.394", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_163.wav", "onoffCaption": "duck quacking at 0.386-2.386, 3.148-5.148 and sheep goat bleating at 3.949-5.949", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_186.wav", "onoffCaption": "door slamming at 0.044-1.161 and burping belching at 1.036-4.316", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_310.wav", "onoffCaption": "sneeze at 1.983-4.386, 5.995-7.495", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_345.wav", "onoffCaption": "spraying at 1.07-3.462 and explosion at 4.971-9.971", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_535.wav", "onoffCaption": "woman laughing at 2.988-5.343, 6.095-8.681", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_608.wav", "onoffCaption": "explosion at 2.457-5.329", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_713.wav", "onoffCaption": "duck quacking at 0.237-2.237, 3.979-5.979, 7.934-9.934", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_746.wav", "onoffCaption": "train horn at 0.123-4.453, 5.099-9.429", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_831.wav", "onoffCaption": "dog barking at 0.341-2.341, 2.889-4.889, 6.503-8.503", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_881.wav", "onoffCaption": "tapping clicking clanking at 0.683-4.123 and thump thud at 7.869-10.0", "frequencyCaption": "tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1005.wav", "onoffCaption": "door knocking at 1.384-7.444 and explosion at 5.109-8.109", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1050.wav", "onoffCaption": "explosion at 2.575-4.639, 5.936-8.753", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1223.wav", "onoffCaption": "sheep goat bleating at 1.401-3.401, 5.028-7.028", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1276.wav", "onoffCaption": "gunshot at 2.945-4.945, 6.205-8.706", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1338.wav", "onoffCaption": "tapping clicking clanking at 0.123-3.563, 4.576-8.016 and sheep goat bleating at 6.188-8.188", "frequencyCaption": "tapping clicking clanking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1453.wav", "onoffCaption": "train horn at 2.239-5.119, 5.847-8.727", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1548.wav", "onoffCaption": "thump thud at 0.102-3.149, 3.878-6.469", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1620.wav", "onoffCaption": "gunshot at 0.078-2.552 and thump thud at 0.946-4.864", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1675.wav", "onoffCaption": "door knocking at 2.094-4.469, 5.66-8.035", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1690.wav", "onoffCaption": "door slamming at 1.034-3.517 and spraying at 7.495-8.099", "frequencyCaption": "door slamming one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1957.wav", "onoffCaption": "sneeze at 1.883-3.59, 4.837-6.544, 8.083-9.79", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3067.wav", "onoffCaption": "thump thud at 0.31-2.81 and burping belching at 6.33-9.091", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3082.wav", "onoffCaption": "cow mooing at 0.949-3.959, 5.766-7.948", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3199.wav", "onoffCaption": "thump thud at 1.786-4.833, 6.026-9.073", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3241.wav", "onoffCaption": "sheep goat bleating at 2.492-4.492, 5.259-7.259, 7.956-9.956", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3431.wav", "onoffCaption": "woman laughing at 0.558-3.258, 3.954-6.039, 6.909-9.007 and whistling at 2.627-7.111", "frequencyCaption": "woman laughing three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3617.wav", "onoffCaption": "train horn at 0.479-6.948, 7.832-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3935.wav", "onoffCaption": "sneeze at 1.497-3.958, 4.871-6.197", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_136.wav", "onoffCaption": "thump thud at 0.761-3.532 and whistling at 5.796-10.0", "frequencyCaption": "thump thud one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_244.wav", "onoffCaption": "spraying at 0.842-1.709 and door slamming at 2.747-3.247", "frequencyCaption": "spraying one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_461.wav", "onoffCaption": "cat meowing at 2.228-3.978 and explosion at 3.017-5.024, 6.181-8.188 and cow mooing at 5.927-8.909", "frequencyCaption": "cat meowing one times and explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_484.wav", "onoffCaption": "car horn honking at 1.94-5.781", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_560.wav", "onoffCaption": "dog barking at 2.223-4.223, 5.485-7.485", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_585.wav", "onoffCaption": "spraying at 3.005-3.609, 4.518-6.646, 7.52-9.215", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_612.wav", "onoffCaption": "dog barking at 0.012-2.012, 4.228-6.228, 6.945-8.945", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_864.wav", "onoffCaption": "explosion at 1.183-4.183 and woman laughing at 7.821-10.0", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_965.wav", "onoffCaption": "dog barking at 2.009-4.009, 5.762-7.762 and explosion at 2.108-4.861, 5.845-8.598", "frequencyCaption": "dog barking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_980.wav", "onoffCaption": "cat meowing at 0.306-1.446, 2.819-3.959", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1151.wav", "onoffCaption": "sneeze at 0.567-3.181 and explosion at 1.075-6.075", "frequencyCaption": "sneeze one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1406.wav", "onoffCaption": "sheep goat bleating at 1.684-3.684, 4.252-6.252, 7.758-9.758", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1507.wav", "onoffCaption": "cat meowing at 0.787-2.328, 4.605-5.915 and burping belching at 1.025-6.149, 7.732-9.857", "frequencyCaption": "cat meowing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1791.wav", "onoffCaption": "duck quacking at 0.18-2.18 and cat meowing at 0.29-3.531, 4.935-7.682", "frequencyCaption": "duck quacking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1803.wav", "onoffCaption": "whistling at 2.821-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1819.wav", "onoffCaption": "sneeze at 0.143-2.389, 2.913-4.832, 5.475-7.588", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1902.wav", "onoffCaption": "dog barking at 2.405-4.405, 6.495-8.495", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1918.wav", "onoffCaption": "burping belching at 3.973-6.973", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3028.wav", "onoffCaption": "sneeze at 0.602-1.879", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3032.wav", "onoffCaption": "woman laughing at 1.827-5.215, 6.908-9.263", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3129.wav", "onoffCaption": "spraying at 1.764-2.828, 4.27-5.334 and door knocking at 6.965-9.502", "frequencyCaption": "spraying two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3133.wav", "onoffCaption": "woman laughing at 1.628-3.836, 5.97-8.178", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3340.wav", "onoffCaption": "door slamming at 0.33-2.526 and cow mooing at 6.04-9.022", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3464.wav", "onoffCaption": "dog barking at 3.184-5.184, 6.269-8.269", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3481.wav", "onoffCaption": "explosion at 1.66-4.528, 6.089-8.957", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3565.wav", "onoffCaption": "cow mooing at 0.862-3.872 and spraying at 7.02-7.624, 9.258-9.862", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3580.wav", "onoffCaption": "whistling at 0.212-7.611", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3716.wav", "onoffCaption": "thump thud at 3.076-7.526", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3861.wav", "onoffCaption": "whistling at 0.139-7.889", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3884.wav", "onoffCaption": "cat meowing at 3.525-8.525", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3960.wav", "onoffCaption": "cow mooing at 3.13-6.099, 6.661-9.63", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3985.wav", "onoffCaption": "thump thud at 1.04-4.958", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_108.wav", "onoffCaption": "cow mooing at 0.556-4.985 and train horn at 0.71-8.91 and woman laughing at 2.098-4.466, 5.962-8.33", "frequencyCaption": "cow mooing one times and train horn one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_179.wav", "onoffCaption": "gunshot at 2.115-4.115, 5.617-7.617", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_211.wav", "onoffCaption": "burping belching at 0.658-6.602, 7.664-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_434.wav", "onoffCaption": "explosion at 2.431-5.16 and train horn at 7.14-10.0", "frequencyCaption": "explosion one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_647.wav", "onoffCaption": "sheep goat bleating at 1.751-3.751", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_709.wav", "onoffCaption": "cow mooing at 0.149-3.447, 4.752-7.721", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_930.wav", "onoffCaption": "duck quacking at 2.399-4.399, 5.505-7.505 and woman laughing at 3.444-5.526, 6.964-9.448 and dog barking at 4.535-6.535", "frequencyCaption": "duck quacking two times and woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_941.wav", "onoffCaption": "cow mooing at 0.379-5.359, 7.153-9.591", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1104.wav", "onoffCaption": "door slamming at 0.251-2.016, 2.613-4.378, 5.27-7.035 and gunshot at 2.093-4.093, 5.224-7.243 and sheep goat bleating at 2.203-4.203, 6.075-8.075", "frequencyCaption": "door slamming three times and gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1190.wav", "onoffCaption": "woman laughing at 0.835-3.035 and spraying at 6.648-7.223", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1239.wav", "onoffCaption": "woman laughing at 0.599-3.185, 4.736-7.341", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1289.wav", "onoffCaption": "car horn honking at 3.261-6.047, 7.282-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1322.wav", "onoffCaption": "duck quacking at 3.022-5.022, 6.303-8.303", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1377.wav", "onoffCaption": "sheep goat bleating at 0.256-2.256 and sneeze at 4.422-6.105, 7.36-9.043", "frequencyCaption": "sheep goat bleating one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1392.wav", "onoffCaption": "car horn honking at 0.604-3.117, 5.468-7.981", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1449.wav", "onoffCaption": "explosion at 2.052-4.313, 6.672-8.933", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1552.wav", "onoffCaption": "burping belching at 0.716-3.716 and duck quacking at 5.025-7.025, 7.592-9.592", "frequencyCaption": "burping belching one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1721.wav", "onoffCaption": "cow mooing at 1.991-4.973, 5.974-8.943", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1774.wav", "onoffCaption": "door slamming at 1.63-2.863, 4.075-6.075, 6.681-9.11", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1827.wav", "onoffCaption": "woman laughing at 3.327-5.919", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1856.wav", "onoffCaption": "sneeze at 2.355-4.969", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3098.wav", "onoffCaption": "car horn honking at 1.715-6.037", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3117.wav", "onoffCaption": "sneeze at 0.541-2.604, 4.626-6.689 and sheep goat bleating at 2.158-4.158, 5.39-7.39", "frequencyCaption": "sneeze two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3166.wav", "onoffCaption": "spraying at 2.954-3.805", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3183.wav", "onoffCaption": "sheep goat bleating at 0.991-2.991, 3.74-5.74, 6.273-8.273", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3315.wav", "onoffCaption": "duck quacking at 0.019-2.019, 3.177-5.177, 6.468-8.468", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3530.wav", "onoffCaption": "woman laughing at 2.851-5.443", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3541.wav", "onoffCaption": "spraying at 0.143-0.793, 1.737-2.588, 3.538-4.27 and train horn at 5.735-10.0", "frequencyCaption": "spraying three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3658.wav", "onoffCaption": "cat meowing at 0.634-2.529, 3.868-5.763 and explosion at 3.879-6.879, 7.905-10.0", "frequencyCaption": "cat meowing two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_3743.wav", "onoffCaption": "thump thud at 2.308-6.758", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3834.wav", "onoffCaption": "sneeze at 1.737-4.397 and woman laughing at 6.52-9.106", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3845.wav", "onoffCaption": "tapping clicking clanking at 3.386-6.826 and door slamming at 6.172-8.63", "frequencyCaption": "tapping clicking clanking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_260.wav", "onoffCaption": "car horn honking at 0.679-3.497, 4.414-6.414, 7.4-9.747 and thump thud at 0.765-3.104", "frequencyCaption": "car horn honking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_285.wav", "onoffCaption": "cat meowing at 2.662-3.674", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_410.wav", "onoffCaption": "sneeze at 2.922-4.918", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_445.wav", "onoffCaption": "burping belching at 2.349-4.443 and sheep goat bleating at 7.468-9.468", "frequencyCaption": "burping belching one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_636.wav", "onoffCaption": "thump thud at 0.085-3.752, 4.485-8.152", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_778.wav", "onoffCaption": "cat meowing at 0.073-1.968, 4.277-5.463", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_914.wav", "onoffCaption": "cat meowing at 2.239-3.266, 3.834-4.861, 5.407-6.434", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1120.wav", "onoffCaption": "sheep goat bleating at 3.176-5.176, 6.418-8.418", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1175.wav", "onoffCaption": "spraying at 0.156-2.284", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1248.wav", "onoffCaption": "woman laughing at 0.652-3.447, 4.195-6.99 and burping belching at 2.568-6.591, 7.652-10.0", "frequencyCaption": "woman laughing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1306.wav", "onoffCaption": "woman laughing at 0.786-3.154, 5.194-7.562 and spraying at 1.837-2.412, 4.389-7.386 and explosion at 5.217-8.211", "frequencyCaption": "woman laughing two times and spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1353.wav", "onoffCaption": "woman laughing at 2.632-4.832 and tapping clicking clanking at 7.798-10.0", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1438.wav", "onoffCaption": "dog barking at 0.127-5.764 and tapping clicking clanking at 7.741-10.0", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1488.wav", "onoffCaption": "woman laughing at 0.084-3.365, 4.111-7.392", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1523.wav", "onoffCaption": "cat meowing at 3.057-4.628, 5.333-6.904, 8.188-9.759", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1576.wav", "onoffCaption": "woman laughing at 0.377-2.602", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1593.wav", "onoffCaption": "duck quacking at 2.419-4.419, 5.123-7.123, 7.708-9.708", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1705.wav", "onoffCaption": "gunshot at 0.188-2.188, 3.006-5.006, 6.0-8.0 and cow mooing at 2.515-5.813, 6.954-9.655", "frequencyCaption": "gunshot three times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1750.wav", "onoffCaption": "thump thud at 0.121-4.039, 5.53-8.03", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1872.wav", "onoffCaption": "cat meowing at 0.257-1.267 and train horn at 5.374-7.814", "frequencyCaption": "cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1897.wav", "onoffCaption": "cow mooing at 3.107-8.087", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1969.wav", "onoffCaption": "duck quacking at 0.218-2.218", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3059.wav", "onoffCaption": "burping belching at 0.083-2.286", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3142.wav", "onoffCaption": "woman laughing at 0.185-2.885, 4.611-7.311 and burping belching at 2.382-5.822 and duck quacking at 5.807-7.807", "frequencyCaption": "woman laughing two times and burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3364.wav", "onoffCaption": "door slamming at 1.137-3.333, 4.292-6.488, 7.604-9.8", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3381.wav", "onoffCaption": "duck quacking at 2.558-4.558 and dog barking at 3.089-5.089, 6.993-8.993", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3514.wav", "onoffCaption": "sneeze at 1.47-6.47, 7.74-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3629.wav", "onoffCaption": "duck quacking at 2.888-4.888, 7.107-9.107", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3699.wav", "onoffCaption": "woman laughing at 0.287-3.568, 4.142-6.944, 7.835-9.884", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3732.wav", "onoffCaption": "cow mooing at 0.364-3.662, 4.801-7.605", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3782.wav", "onoffCaption": "spraying at 0.427-2.887, 4.194-5.451", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3810.wav", "onoffCaption": "thump thud at 2.015-5.682, 7.857-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_35.wav", "onoffCaption": "thump thud at 3.337-5.676, 7.377-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_60.wav", "onoffCaption": "cat meowing at 0.035-1.619 and explosion at 0.384-3.256, 3.9-6.772 and spraying at 3.01-5.47", "frequencyCaption": "cat meowing one times and explosion two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_147.wav", "onoffCaption": "gunshot at 0.792-2.792, 3.335-5.335", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_235.wav", "onoffCaption": "cat meowing at 0.778-1.787, 2.553-3.565 and gunshot at 2.067-4.067, 6.151-8.151", "frequencyCaption": "cat meowing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_334.wav", "onoffCaption": "duck quacking at 3.078-5.078", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_361.wav", "onoffCaption": "car horn honking at 2.942-5.289, 5.9-8.413", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_384.wav", "onoffCaption": "duck quacking at 1.589-3.589, 4.206-6.206", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_511.wav", "onoffCaption": "woman laughing at 0.627-2.91 and thump thud at 6.446-9.217", "frequencyCaption": "woman laughing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_663.wav", "onoffCaption": "whistling at 0.402-4.886 and dog barking at 2.751-4.751, 6.775-8.775 and door slamming at 8.642-9.142", "frequencyCaption": "whistling one times and dog barking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_679.wav", "onoffCaption": "sneeze at 0.263-2.326, 3.042-5.105, 6.243-8.306", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_686.wav", "onoffCaption": "cat meowing at 0.012-3.357, 5.746-7.478", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_737.wav", "onoffCaption": "thump thud at 3.081-7.531", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_762.wav", "onoffCaption": "car horn honking at 3.016-7.528", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_787.wav", "onoffCaption": "sneeze at 0.267-2.506, 3.642-5.881 and cat meowing at 1.741-6.101", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_815.wav", "onoffCaption": "dog barking at 0.838-2.838, 4.348-6.348 and woman laughing at 5.774-7.982", "frequencyCaption": "dog barking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1021.wav", "onoffCaption": "explosion at 2.342-5.342, 5.92-8.92 and burping belching at 2.438-5.438", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1074.wav", "onoffCaption": "burping belching at 2.126-4.233, 5.913-8.02", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1207.wav", "onoffCaption": "sheep goat bleating at 0.427-2.427, 3.863-5.863, 6.741-8.741", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1252.wav", "onoffCaption": "thump thud at 2.779-5.55, 6.397-8.569", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1349.wav", "onoffCaption": "burping belching at 3.22-5.251, 5.977-8.943", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1477.wav", "onoffCaption": "door knocking at 0.762-5.058", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1492.wav", "onoffCaption": "train horn at 0.546-3.706", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1539.wav", "onoffCaption": "whistling at 0.69-5.174, 5.721-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1589.wav", "onoffCaption": "cow mooing at 1.966-4.935, 6.103-9.072", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1604.wav", "onoffCaption": "sneeze at 0.952-5.481, 6.461-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1651.wav", "onoffCaption": "thump thud at 0.031-4.481, 5.06-7.831 and cow mooing at 1.949-6.929 and gunshot at 2.458-4.458, 5.594-7.594", "frequencyCaption": "thump thud two times and cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1868.wav", "onoffCaption": "dog barking at 2.736-4.736, 5.262-7.262", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1973.wav", "onoffCaption": "sheep goat bleating at 0.478-2.478", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1996.wav", "onoffCaption": "sheep goat bleating at 0.337-2.337, 4.071-6.071", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3043.wav", "onoffCaption": "duck quacking at 1.343-3.343 and door slamming at 6.658-8.576", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3158.wav", "onoffCaption": "spraying at 2.079-3.841, 4.379-6.141, 7.114-8.876", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3230.wav", "onoffCaption": "car horn honking at 0.711-5.111 and burping belching at 7.655-10.0", "frequencyCaption": "car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3265.wav", "onoffCaption": "gunshot at 0.842-2.861, 3.712-5.731, 6.886-8.905", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3280.wav", "onoffCaption": "thump thud at 0.098-4.548 and train horn at 1.037-4.807, 5.772-8.572", "frequencyCaption": "thump thud one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3331.wav", "onoffCaption": "explosion at 0.213-3.213, 4.115-6.28 and gunshot at 7.473-9.473", "frequencyCaption": "explosion two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3415.wav", "onoffCaption": "tapping clicking clanking at 1.663-5.103 and thump thud at 1.668-4.168, 5.582-8.082", "frequencyCaption": "tapping clicking clanking one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3633.wav", "onoffCaption": "cow mooing at 0.348-4.777", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3666.wav", "onoffCaption": "whistling at 3.006-7.49", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3683.wav", "onoffCaption": "tapping clicking clanking at 0.355-3.795, 6.028-8.443", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3728.wav", "onoffCaption": "door knocking at 0.613-2.993 and sheep goat bleating at 1.374-3.374, 4.594-6.594", "frequencyCaption": "door knocking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3767.wav", "onoffCaption": "cow mooing at 1.682-4.651, 6.711-9.68", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3798.wav", "onoffCaption": "sneeze at 2.406-6.054, 7.169-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3911.wav", "onoffCaption": "explosion at 3.713-6.713, 7.614-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_5000.wav", "onoffCaption": "woman laughing at 2.643-4.868, 7.346-9.571", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_85.wav", "onoffCaption": "thump thud at 0.092-4.542, 5.186-9.636", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_112.wav", "onoffCaption": "door slamming at 2.357-3.474, 4.082-6.0", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_175.wav", "onoffCaption": "whistling at 0.475-8.713", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_190.wav", "onoffCaption": "burping belching at 2.79-5.969", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_248.wav", "onoffCaption": "whistling at 0.668-3.643, 6.008-8.983", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_306.wav", "onoffCaption": "whistling at 2.191-9.535", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_353.wav", "onoffCaption": "explosion at 1.105-6.105, 7.952-9.971", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_438.wav", "onoffCaption": "sheep goat bleating at 1.945-3.945", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_523.wav", "onoffCaption": "explosion at 2.55-5.141, 5.651-8.242", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_544.wav", "onoffCaption": "gunshot at 1.072-3.242, 3.961-6.002, 7.293-9.293", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_705.wav", "onoffCaption": "train horn at 0.471-5.329", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_750.wav", "onoffCaption": "cow mooing at 2.139-5.149, 5.874-8.836", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_827.wav", "onoffCaption": "tapping clicking clanking at 0.111-3.551, 5.819-8.222 and whistling at 3.405-8.58 and explosion at 5.426-8.154", "frequencyCaption": "tapping clicking clanking two times and whistling one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_840.wav", "onoffCaption": "gunshot at 0.056-2.056 and whistling at 4.934-10.0", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1013.wav", "onoffCaption": "door slamming at 0.266-0.766, 1.741-3.659, 4.964-6.964 and explosion at 1.158-6.158", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1091.wav", "onoffCaption": "explosion at 0.702-3.302", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1108.wav", "onoffCaption": "door knocking at 0.744-3.047, 4.096-6.399", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1235.wav", "onoffCaption": "explosion at 0.214-2.967", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1260.wav", "onoffCaption": "train horn at 1.145-3.945 and explosion at 6.009-8.609", "frequencyCaption": "train horn one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1285.wav", "onoffCaption": "thump thud at 0.028-2.799 and gunshot at 0.633-2.633, 4.968-6.968", "frequencyCaption": "thump thud one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1422.wav", "onoffCaption": "whistling at 1.541-6.716", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1445.wav", "onoffCaption": "gunshot at 2.192-4.192, 6.123-8.123", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1636.wav", "onoffCaption": "cow mooing at 0.692-3.674, 4.245-7.07 and explosion at 1.708-4.576, 6.62-9.614 and door slamming at 7.173-8.312", "frequencyCaption": "cow mooing two times and explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1663.wav", "onoffCaption": "thump thud at 1.73-3.958, 5.73-8.425", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1686.wav", "onoffCaption": "spraying at 1.212-3.648, 4.274-6.71, 7.857-10.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1778.wav", "onoffCaption": "dog barking at 2.24-4.24 and explosion at 6.124-10.0", "frequencyCaption": "dog barking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1926.wav", "onoffCaption": "whistling at 3.395-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1941.wav", "onoffCaption": "whistling at 0.213-7.868 and gunshot at 3.4-5.4", "frequencyCaption": "whistling one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3016.wav", "onoffCaption": "cow mooing at 2.11-5.408", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3071.wav", "onoffCaption": "car horn honking at 0.372-2.837, 5.051-7.551", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3094.wav", "onoffCaption": "gunshot at 2.875-4.875", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3202.wav", "onoffCaption": "duck quacking at 0.628-2.628, 3.29-5.29, 6.192-8.192", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3257.wav", "onoffCaption": "car horn honking at 0.974-3.321, 5.028-7.375 and train horn at 3.119-5.759, 6.861-9.328", "frequencyCaption": "car horn honking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3319.wav", "onoffCaption": "sneeze at 3.802-5.036, 6.027-8.488", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3427.wav", "onoffCaption": "spraying at 0.506-2.966, 4.117-5.85, 7.285-9.895 and woman laughing at 1.012-4.588", "frequencyCaption": "spraying three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3440.wav", "onoffCaption": "car horn honking at 1.313-5.713, 7.865-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3497.wav", "onoffCaption": "duck quacking at 1.589-3.589, 4.21-6.21, 7.047-9.047", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3601.wav", "onoffCaption": "spraying at 0.049-1.133 and burping belching at 5.237-8.237", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3654.wav", "onoffCaption": "sneeze at 2.713-4.325", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3838.wav", "onoffCaption": "train horn at 2.023-7.738", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3923.wav", "onoffCaption": "gunshot at 0.039-2.209, 3.821-5.821", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3944.wav", "onoffCaption": "sheep goat bleating at 0.704-2.704, 4.835-6.835 and cat meowing at 1.039-2.252, 3.829-5.245", "frequencyCaption": "sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_48.wav", "onoffCaption": "door knocking at 1.798-3.925, 6.24-8.864", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_52.wav", "onoffCaption": "cat meowing at 0.243-5.243 and whistling at 1.921-4.15, 5.629-8.103", "frequencyCaption": "cat meowing one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_120.wav", "onoffCaption": "duck quacking at 2.114-4.114, 4.766-6.766, 7.77-9.77", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_252.wav", "onoffCaption": "gunshot at 0.236-2.236, 4.669-6.71", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_349.wav", "onoffCaption": "gunshot at 2.895-4.895, 7.074-9.074 and duck quacking at 3.758-5.758", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_422.wav", "onoffCaption": "spraying at 2.662-5.054, 5.777-8.169", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_477.wav", "onoffCaption": "dog barking at 2.616-4.616, 5.85-7.85", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_488.wav", "onoffCaption": "thump thud at 0.294-4.744 and explosion at 2.961-5.833", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_492.wav", "onoffCaption": "burping belching at 3.015-6.515, 7.914-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_539.wav", "onoffCaption": "duck quacking at 0.626-2.626 and gunshot at 0.647-2.647 and sheep goat bleating at 6.015-8.015", "frequencyCaption": "duck quacking one times and gunshot one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_576.wav", "onoffCaption": "cow mooing at 0.017-2.986", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_589.wav", "onoffCaption": "sneeze at 0.227-2.34", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_593.wav", "onoffCaption": "spraying at 0.736-1.386, 2.43-4.89 and burping belching at 2.475-5.734 and car horn honking at 3.11-7.51", "frequencyCaption": "spraying two times and burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_604.wav", "onoffCaption": "sneeze at 0.064-2.381, 4.864-7.181 and woman laughing at 1.889-4.658, 5.873-8.11", "frequencyCaption": "sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_868.wav", "onoffCaption": "duck quacking at 1.107-3.107, 5.382-7.382", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_872.wav", "onoffCaption": "sheep goat bleating at 1.92-4.92", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_897.wav", "onoffCaption": "woman laughing at 2.093-7.132, 7.959-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_969.wav", "onoffCaption": "thump thud at 0.204-3.871, 4.373-6.774 and duck quacking at 0.295-2.295, 4.429-6.429", "frequencyCaption": "thump thud two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_973.wav", "onoffCaption": "woman laughing at 2.152-4.636, 7.017-9.501", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_996.wav", "onoffCaption": "door knocking at 1.037-3.127 and train horn at 4.574-7.054", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1009.wav", "onoffCaption": "explosion at 0.047-5.047, 6.829-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1046.wav", "onoffCaption": "thump thud at 2.605-7.055", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1112.wav", "onoffCaption": "dog barking at 0.813-3.251", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1147.wav", "onoffCaption": "door slamming at 0.008-0.986, 2.768-3.746 and burping belching at 6.192-9.192", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1334.wav", "onoffCaption": "burping belching at 2.724-9.892", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1410.wav", "onoffCaption": "spraying at 2.671-3.298, 4.313-4.94, 5.78-6.407", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1511.wav", "onoffCaption": "cow mooing at 2.906-5.875, 6.784-9.753", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1544.wav", "onoffCaption": "tapping clicking clanking at 0.031-3.471, 4.249-7.055", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1679.wav", "onoffCaption": "whistling at 0.412-5.912", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1762.wav", "onoffCaption": "sneeze at 0.149-5.149", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1787.wav", "onoffCaption": "tapping clicking clanking at 3.695-7.135", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1815.wav", "onoffCaption": "cat meowing at 2.792-3.978, 4.638-6.388", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1840.wav", "onoffCaption": "whistling at 0.127-7.877 and duck quacking at 0.626-2.626, 4.514-6.514", "frequencyCaption": "whistling one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1914.wav", "onoffCaption": "duck quacking at 0.36-2.36 and spraying at 5.948-6.68, 7.206-7.938, 9.061-9.793", "frequencyCaption": "duck quacking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3024.wav", "onoffCaption": "whistling at 1.51-4.385, 4.941-7.136 and sheep goat bleating at 4.617-6.617", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3125.wav", "onoffCaption": "cow mooing at 0.525-3.535 and sneeze at 7.273-10.0", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3170.wav", "onoffCaption": "thump thud at 2.4-6.775 and cow mooing at 3.142-7.571", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3195.wav", "onoffCaption": "duck quacking at 3.632-5.632, 6.157-8.157", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3356.wav", "onoffCaption": "dog barking at 0.004-2.004 and sneeze at 0.554-2.867", "frequencyCaption": "dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3468.wav", "onoffCaption": "tapping clicking clanking at 1.654-5.094, 6.11-9.55", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3472.wav", "onoffCaption": "thump thud at 0.086-3.133, 4.191-7.238 and cow mooing at 2.851-6.149, 7.782-10.0", "frequencyCaption": "thump thud two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3526.wav", "onoffCaption": "burping belching at 2.329-5.329, 5.967-8.967 and spraying at 3.37-5.498", "frequencyCaption": "burping belching two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3569.wav", "onoffCaption": "woman laughing at 3.645-6.717", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3573.wav", "onoffCaption": "tapping clicking clanking at 0.078-3.518 and woman laughing at 2.07-4.438", "frequencyCaption": "tapping clicking clanking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3596.wav", "onoffCaption": "gunshot at 0.42-2.42, 3.855-5.855 and whistling at 0.821-3.796, 5.538-7.962", "frequencyCaption": "gunshot two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3700.wav", "onoffCaption": "thump thud at 2.566-6.484", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3822.wav", "onoffCaption": "tapping clicking clanking at 0.026-3.466, 4.038-7.478 and sneeze at 0.281-2.594 and whistling at 0.577-8.588", "frequencyCaption": "tapping clicking clanking two times and sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3877.wav", "onoffCaption": "sheep goat bleating at 2.689-5.985, 7.249-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3888.wav", "onoffCaption": "gunshot at 0.011-2.011, 4.076-6.076 and woman laughing at 2.145-4.353, 5.515-7.798 and burping belching at 3.798-6.321, 7.51-9.581", "frequencyCaption": "gunshot two times and woman laughing two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3892.wav", "onoffCaption": "whistling at 0.003-5.178, 6.434-10.0 and gunshot at 0.281-2.281, 4.664-6.664", "frequencyCaption": "whistling two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3939.wav", "onoffCaption": "door knocking at 2.256-5.103, 6.83-8.99", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3976.wav", "onoffCaption": "sheep goat bleating at 4.316-6.316, 7.624-9.624", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3989.wav", "onoffCaption": "duck quacking at 0.753-2.753, 4.201-6.201", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3993.wav", "onoffCaption": "sheep goat bleating at 0.539-3.619, 5.059-8.139", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_39.wav", "onoffCaption": "explosion at 0.263-2.327, 4.015-6.079", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_89.wav", "onoffCaption": "sheep goat bleating at 1.399-3.399, 4.145-6.145, 6.832-9.296", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_207.wav", "onoffCaption": "train horn at 0.007-4.188 and cat meowing at 0.737-2.714", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_276.wav", "onoffCaption": "dog barking at 0.676-2.676 and sheep goat bleating at 1.893-3.893, 4.654-7.441", "frequencyCaption": "dog barking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_293.wav", "onoffCaption": "spraying at 3.538-4.795", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_338.wav", "onoffCaption": "tapping clicking clanking at 3.456-6.896", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_388.wav", "onoffCaption": "car horn honking at 0.262-4.662, 6.687-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_406.wav", "onoffCaption": "sheep goat bleating at 1.038-3.038, 4.725-6.725, 7.491-9.491", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_453.wav", "onoffCaption": "tapping clicking clanking at 0.529-3.969, 5.003-7.612", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_548.wav", "onoffCaption": "burping belching at 3.305-8.305", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_620.wav", "onoffCaption": "spraying at 1.449-1.971, 3.051-3.573, 4.44-4.962", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_651.wav", "onoffCaption": "explosion at 0.475-5.396, 6.197-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_819.wav", "onoffCaption": "cow mooing at 0.503-3.485, 4.402-7.384 and dog barking at 3.328-5.328, 6.489-8.489", "frequencyCaption": "cow mooing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_902.wav", "onoffCaption": "explosion at 0.945-5.945, 7.669-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_926.wav", "onoffCaption": "woman laughing at 1.915-5.196, 6.014-9.295", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_957.wav", "onoffCaption": "spraying at 3.156-3.783, 4.327-4.896, 5.659-6.743", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1078.wav", "onoffCaption": "dog barking at 2.572-4.572", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1136.wav", "onoffCaption": "sneeze at 1.816-3.761 and woman laughing at 6.146-9.722", "frequencyCaption": "sneeze one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1163.wav", "onoffCaption": "thump thud at 0.616-3.663, 4.265-6.765 and cat meowing at 1.658-6.658, 7.892-8.904", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1186.wav", "onoffCaption": "tapping clicking clanking at 2.584-6.024, 7.048-9.723", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1310.wav", "onoffCaption": "cat meowing at 0.11-1.122, 1.672-3.404, 4.651-7.141", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1345.wav", "onoffCaption": "tapping clicking clanking at 0.045-3.485, 4.395-7.835", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1361.wav", "onoffCaption": "whistling at 2.651-8.484", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1384.wav", "onoffCaption": "burping belching at 1.138-3.959 and explosion at 7.473-10.0", "frequencyCaption": "burping belching one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1535.wav", "onoffCaption": "cat meowing at 0.403-1.616, 2.396-3.951", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1560.wav", "onoffCaption": "dog barking at 0.055-2.055, 2.803-4.803, 6.123-8.123", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1585.wav", "onoffCaption": "car horn honking at 1.4-5.649", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1608.wav", "onoffCaption": "tapping clicking clanking at 0.833-4.273 and sneeze at 1.635-2.738", "frequencyCaption": "tapping clicking clanking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1713.wav", "onoffCaption": "burping belching at 0.208-3.708 and train horn at 5.339-8.219", "frequencyCaption": "burping belching one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1737.wav", "onoffCaption": "woman laughing at 2.723-4.948, 6.995-9.6", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1746.wav", "onoffCaption": "cat meowing at 2.19-4.167, 5.515-7.651", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1831.wav", "onoffCaption": "gunshot at 2.959-4.959, 6.448-8.448", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1864.wav", "onoffCaption": "duck quacking at 2.461-4.461, 5.492-7.492 and woman laughing at 3.466-6.261", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1881.wav", "onoffCaption": "sheep goat bleating at 2.941-4.941, 5.649-7.649", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3101.wav", "onoffCaption": "door slamming at 0.408-2.628, 4.108-6.328 and woman laughing at 2.274-4.474 and door knocking at 3.199-6.752, 7.928-10.0", "frequencyCaption": "door slamming two times and woman laughing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3154.wav", "onoffCaption": "gunshot at 2.635-4.635", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3218.wav", "onoffCaption": "gunshot at 0.374-2.374 and sneeze at 5.184-7.798", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3269.wav", "onoffCaption": "woman laughing at 2.804-5.573", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3303.wav", "onoffCaption": "train horn at 0.168-3.408, 4.333-7.573", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3372.wav", "onoffCaption": "whistling at 2.433-5.408, 7.427-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3397.wav", "onoffCaption": "train horn at 3.047-8.791 and spraying at 3.625-4.229, 6.445-7.445", "frequencyCaption": "train horn one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3419.wav", "onoffCaption": "duck quacking at 0.225-2.225 and dog barking at 1.184-3.184, 4.339-6.339", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3502.wav", "onoffCaption": "sneeze at 0.887-3.501, 4.361-6.975 and door slamming at 2.409-3.214, 5.124-7.905", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3557.wav", "onoffCaption": "thump thud at 1.144-5.519, 6.289-8.744 and sheep goat bleating at 2.984-6.904, 7.408-9.604", "frequencyCaption": "thump thud two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3724.wav", "onoffCaption": "duck quacking at 0.417-2.417, 3.645-5.645", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3755.wav", "onoffCaption": "dog barking at 1.109-3.109, 4.896-6.896, 7.76-9.76 and duck quacking at 3.606-5.606", "frequencyCaption": "dog barking three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3794.wav", "onoffCaption": "whistling at 0.449-3.324, 4.92-7.795", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3806.wav", "onoffCaption": "car horn honking at 2.923-5.27, 6.695-9.042", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3853.wav", "onoffCaption": "car horn honking at 2.237-6.749, 7.707-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3948.wav", "onoffCaption": "cow mooing at 2.363-7.343 and gunshot at 6.261-8.501", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_23.wav", "onoffCaption": "sheep goat bleating at 2.024-4.024, 4.681-6.681, 7.699-9.699", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_76.wav", "onoffCaption": "cow mooing at 0.517-5.497", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_151.wav", "onoffCaption": "burping belching at 0.578-4.447 and door slamming at 2.144-3.468", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_223.wav", "onoffCaption": "thump thud at 0.521-3.292 and sneeze at 2.031-3.285", "frequencyCaption": "thump thud one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_239.wav", "onoffCaption": "sheep goat bleating at 0.693-2.693, 4.197-6.197", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_289.wav", "onoffCaption": "cat meowing at 0.229-1.25, 2.111-3.132, 4.459-5.48 and woman laughing at 2.584-5.684, 7.136-9.504", "frequencyCaption": "cat meowing three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_322.wav", "onoffCaption": "door slamming at 1.672-4.551, 7.01-8.401 and thump thud at 2.44-4.779", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_377.wav", "onoffCaption": "whistling at 2.592-5.567, 6.838-9.813", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_392.wav", "onoffCaption": "door slamming at 0.562-2.69, 3.661-5.789, 7.08-9.208", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_507.wav", "onoffCaption": "duck quacking at 2.858-4.858", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_675.wav", "onoffCaption": "woman laughing at 2.855-5.063, 5.692-8.111", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_690.wav", "onoffCaption": "thump thud at 2.745-4.973, 7.024-9.486", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_721.wav", "onoffCaption": "car horn honking at 0.361-4.761, 7.058-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_774.wav", "onoffCaption": "burping belching at 2.738-4.832 and cow mooing at 7.019-10.0", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_791.wav", "onoffCaption": "train horn at 0.204-2.671, 3.684-6.181 and whistling at 0.286-8.671 and burping belching at 5.965-8.195", "frequencyCaption": "train horn two times and whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_803.wav", "onoffCaption": "dog barking at 1.213-3.213, 5.389-7.389 and explosion at 3.633-6.505, 7.382-10.0", "frequencyCaption": "dog barking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_918.wav", "onoffCaption": "door knocking at 1.957-4.084, 6.555-8.682", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1037.wav", "onoffCaption": "cow mooing at 0.323-3.292, 4.556-7.542", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1211.wav", "onoffCaption": "train horn at 0.755-4.515, 5.812-9.572", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1244.wav", "onoffCaption": "gunshot at 0.224-2.698 and woman laughing at 6.812-9.912", "frequencyCaption": "gunshot one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1461.wav", "onoffCaption": "duck quacking at 0.523-2.523, 4.907-6.907 and woman laughing at 1.227-3.421", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1484.wav", "onoffCaption": "burping belching at 2.002-8.979 and tapping clicking clanking at 3.197-6.637", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1612.wav", "onoffCaption": "whistling at 2.42-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1647.wav", "onoffCaption": "gunshot at 0.059-2.059, 2.889-4.889, 5.928-7.928 and sneeze at 0.303-1.986, 2.839-4.133, 5.618-8.191", "frequencyCaption": "gunshot three times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1709.wav", "onoffCaption": "dog barking at 3.101-5.101", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1965.wav", "onoffCaption": "spraying at 3.478-4.329 and thump thud at 6.947-10.0", "frequencyCaption": "spraying one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1980.wav", "onoffCaption": "car horn honking at 3.67-6.135, 7.16-9.625", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3055.wav", "onoffCaption": "thump thud at 0.803-3.85", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3226.wav", "onoffCaption": "whistling at 2.624-7.108, 7.973-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3273.wav", "onoffCaption": "cow mooing at 3.464-6.446", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3296.wav", "onoffCaption": "sheep goat bleating at 2.22-4.22, 5.571-7.571", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3327.wav", "onoffCaption": "sheep goat bleating at 0.979-2.979, 5.085-7.085", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3368.wav", "onoffCaption": "car horn honking at 2.877-6.531, 7.446-9.989", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3403.wav", "onoffCaption": "door knocking at 2.848-5.472, 7.49-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3518.wav", "onoffCaption": "cat meowing at 0.029-1.924, 3.136-5.031", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3625.wav", "onoffCaption": "whistling at 1.721-4.696, 5.425-8.4", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3670.wav", "onoffCaption": "whistling at 1.871-9.882", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3695.wav", "onoffCaption": "explosion at 2.44-5.993 and sheep goat bleating at 4.285-6.285", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3771.wav", "onoffCaption": "explosion at 2.574-5.574, 6.762-9.762", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3907.wav", "onoffCaption": "train horn at 0.688-2.825 and door slamming at 1.492-2.47 and sneeze at 5.398-7.01, 7.662-9.274", "frequencyCaption": "train horn one times and door slamming one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_24.wav", "onoffCaption": "door slamming at 0.185-2.098, 3.22-5.133, 7.597-9.51", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_93.wav", "onoffCaption": "woman laughing at 3.523-6.804", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_104.wav", "onoffCaption": "car horn honking at 1.439-3.786, 5.888-8.792", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_156.wav", "onoffCaption": "cow mooing at 0.403-3.413 and cat meowing at 1.774-6.774", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_449.wav", "onoffCaption": "cat meowing at 0.961-2.377 and car horn honking at 1.892-5.067", "frequencyCaption": "cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_500.wav", "onoffCaption": "train horn at 0.129-3.489, 5.32-7.76", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_552.wav", "onoffCaption": "duck quacking at 1.62-3.62, 4.619-6.619 and cow mooing at 1.955-4.965, 5.474-8.484", "frequencyCaption": "duck quacking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_668.wav", "onoffCaption": "door knocking at 0.117-3.617", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_773.wav", "onoffCaption": "woman laughing at 2.284-6.336", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_796.wav", "onoffCaption": "door slamming at 0.302-1.475, 2.628-3.801, 5.758-6.931 and car horn honking at 2.659-7.566", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_804.wav", "onoffCaption": "sneeze at 1.835-4.91, 5.489-8.564", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_856.wav", "onoffCaption": "cow mooing at 0.081-5.061 and explosion at 1.242-3.833, 4.45-6.514, 7.335-10.0", "frequencyCaption": "cow mooing one times and explosion three times"}
+{"filepath": "data/multi_event_train/syn_1030.wav", "onoffCaption": "cat meowing at 0.829-1.84, 3.044-4.055, 4.829-5.84 and gunshot at 2.193-4.193, 4.87-6.87, 7.81-9.81", "frequencyCaption": "cat meowing three times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1062.wav", "onoffCaption": "whistling at 0.568-8.579 and train horn at 1.664-6.566, 7.957-10.0", "frequencyCaption": "whistling one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_1087.wav", "onoffCaption": "train horn at 2.12-5.52, 6.427-9.827 and thump thud at 2.993-6.911", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1179.wav", "onoffCaption": "thump thud at 0.225-2.725, 4.517-7.017 and train horn at 2.293-5.653, 6.302-9.662", "frequencyCaption": "thump thud two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_1434.wav", "onoffCaption": "cat meowing at 2.858-7.858", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1466.wav", "onoffCaption": "train horn at 0.155-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1483.wav", "onoffCaption": "sneeze at 1.223-2.906 and sheep goat bleating at 1.495-3.495 and dog barking at 2.465-4.465", "frequencyCaption": "sneeze one times and sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1598.wav", "onoffCaption": "door slamming at 0.59-1.707, 3.721-5.045", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1879.wav", "onoffCaption": "woman laughing at 0.565-8.01 and gunshot at 2.025-4.025", "frequencyCaption": "woman laughing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1930.wav", "onoffCaption": "train horn at 0.044-4.484 and thump thud at 0.511-2.85, 4.648-6.987 and sneeze at 3.344-7.4", "frequencyCaption": "train horn one times and thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1962.wav", "onoffCaption": "explosion at 2.685-5.685, 6.915-9.787", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1987.wav", "onoffCaption": "burping belching at 0.826-6.427, 6.931-9.038 and dog barking at 1.463-3.463, 4.507-6.507 and thump thud at 4.019-6.519", "frequencyCaption": "burping belching two times and dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3000.wav", "onoffCaption": "sneeze at 2.175-4.421, 6.906-9.152", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3052.wav", "onoffCaption": "duck quacking at 3.237-5.237", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3149.wav", "onoffCaption": "whistling at 0.327-8.712", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3221.wav", "onoffCaption": "whistling at 0.692-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3404.wav", "onoffCaption": "door knocking at 1.875-6.314", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3456.wav", "onoffCaption": "explosion at 0.557-3.557, 4.707-7.707 and gunshot at 0.902-2.902, 3.819-6.059, 7.052-9.052", "frequencyCaption": "explosion two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3677.wav", "onoffCaption": "woman laughing at 2.965-5.063", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3849.wav", "onoffCaption": "car horn honking at 0.921-5.828, 6.938-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3900.wav", "onoffCaption": "gunshot at 0.379-2.379, 3.498-5.498, 6.33-8.33", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3952.wav", "onoffCaption": "tapping clicking clanking at 0.053-3.493, 5.295-8.735 and door slamming at 3.514-4.687, 5.589-6.762, 7.398-8.571", "frequencyCaption": "tapping clicking clanking two times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_71.wav", "onoffCaption": "car horn honking at 0.054-2.054, 2.567-5.48, 6.163-8.547 and dog barking at 1.396-3.396", "frequencyCaption": "car horn honking three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_94.wav", "onoffCaption": "whistling at 0.664-2.673 and dog barking at 5.547-7.547", "frequencyCaption": "whistling one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_103.wav", "onoffCaption": "train horn at 3.383-6.917, 7.644-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_119.wav", "onoffCaption": "explosion at 1.17-6.17", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_224.wav", "onoffCaption": "explosion at 0.69-3.443, 4.949-7.678", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_271.wav", "onoffCaption": "sheep goat bleating at 0.252-2.252, 3.168-5.863, 7.483-9.483", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_294.wav", "onoffCaption": "dog barking at 4.009-6.009, 6.713-8.713", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_325.wav", "onoffCaption": "train horn at 2.454-4.934", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_370.wav", "onoffCaption": "woman laughing at 0.539-3.611", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_395.wav", "onoffCaption": "explosion at 0.154-5.154, 6.077-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_454.wav", "onoffCaption": "cat meowing at 0.659-1.744, 2.58-3.665 and car horn honking at 6.169-10.0", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_555.wav", "onoffCaption": "train horn at 2.939-6.139, 7.128-9.199", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_627.wav", "onoffCaption": "sneeze at 0.592-2.299, 3.698-5.783", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_726.wav", "onoffCaption": "sneeze at 1.533-3.061, 4.94-6.468", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_769.wav", "onoffCaption": "burping belching at 3.011-6.19, 7.094-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_851.wav", "onoffCaption": "duck quacking at 0.562-2.562, 4.942-6.942", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_950.wav", "onoffCaption": "sheep goat bleating at 0.278-2.278, 3.82-6.577", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1065.wav", "onoffCaption": "burping belching at 0.199-2.293, 3.025-5.119, 6.426-8.52", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_1080.wav", "onoffCaption": "woman laughing at 2.234-8.968", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1164.wav", "onoffCaption": "door knocking at 2.967-5.319, 6.283-8.635", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1181.wav", "onoffCaption": "cat meowing at 0.605-2.793, 3.637-5.597, 7.0-8.011", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1216.wav", "onoffCaption": "whistling at 0.333-7.988", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1243.wav", "onoffCaption": "cat meowing at 0.456-1.596 and sheep goat bleating at 0.866-2.866 and sneeze at 1.641-3.186, 5.174-7.413, 8.442-9.534", "frequencyCaption": "cat meowing one times and sheep goat bleating one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1259.wav", "onoffCaption": "explosion at 0.006-2.094, 3.253-5.341", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1317.wav", "onoffCaption": "door knocking at 0.43-3.539 and duck quacking at 2.429-4.429", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1342.wav", "onoffCaption": "train horn at 2.787-6.147, 7.196-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1358.wav", "onoffCaption": "burping belching at 1.915-3.95, 4.703-6.738", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1429.wav", "onoffCaption": "sneeze at 2.999-4.091, 5.749-7.988 and door slamming at 5.615-8.332", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1433.wav", "onoffCaption": "thump thud at 2.715-6.382, 7.011-9.511", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1528.wav", "onoffCaption": "car horn honking at 3.34-5.84, 6.745-9.245", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1532.wav", "onoffCaption": "door slamming at 0.251-1.089, 2.484-3.322, 5.322-6.16", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1615.wav", "onoffCaption": "car horn honking at 3.507-7.907", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1640.wav", "onoffCaption": "thump thud at 2.133-4.633", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1714.wav", "onoffCaption": "cow mooing at 2.052-7.032 and cat meowing at 3.252-6.493 and burping belching at 3.449-5.48, 6.955-9.955", "frequencyCaption": "cow mooing one times and cat meowing one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1741.wav", "onoffCaption": "cat meowing at 3.699-4.708, 7.085-8.094, 8.677-9.686", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1836.wav", "onoffCaption": "whistling at 1.123-3.998, 6.213-9.088", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1937.wav", "onoffCaption": "explosion at 0.473-5.473, 7.065-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3007.wav", "onoffCaption": "spraying at 3.438-4.379, 5.961-8.545", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3106.wav", "onoffCaption": "thump thud at 0.143-4.518 and burping belching at 7.45-9.485", "frequencyCaption": "thump thud one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3274.wav", "onoffCaption": "sheep goat bleating at 2.531-4.531, 5.953-8.314", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3291.wav", "onoffCaption": "cat meowing at 2.873-4.768, 5.368-7.308, 8.311-9.866", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3375.wav", "onoffCaption": "gunshot at 1.097-3.097, 3.963-5.963, 7.832-9.832 and sneeze at 3.728-5.016", "frequencyCaption": "gunshot three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3390.wav", "onoffCaption": "train horn at 1.323-4.643", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3451.wav", "onoffCaption": "door knocking at 1.209-3.299 and spraying at 3.065-3.916, 5.992-6.896", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3550.wav", "onoffCaption": "car horn honking at 1.828-4.341", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3622.wav", "onoffCaption": "door knocking at 3.698-8.137", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3638.wav", "onoffCaption": "dog barking at 3.178-5.178, 7.122-9.122", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3688.wav", "onoffCaption": "explosion at 0.51-2.512, 3.826-6.694, 7.756-10.0 and dog barking at 1.669-3.669, 6.087-8.087 and tapping clicking clanking at 4.081-7.521", "frequencyCaption": "explosion three times and dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3692.wav", "onoffCaption": "cat meowing at 0.796-2.691, 3.33-5.225", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3723.wav", "onoffCaption": "door knocking at 0.865-4.633, 5.792-9.56", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3739.wav", "onoffCaption": "explosion at 1.129-4.969, 5.71-8.318 and sneeze at 5.171-7.488, 8.553-9.656", "frequencyCaption": "explosion two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3789.wav", "onoffCaption": "train horn at 1.777-5.977, 7.19-10.0 and whistling at 3.008-8.183", "frequencyCaption": "train horn two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3793.wav", "onoffCaption": "woman laughing at 0.114-2.397, 4.314-6.412", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3854.wav", "onoffCaption": "sheep goat bleating at 0.091-2.091", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3955.wav", "onoffCaption": "burping belching at 0.1-2.303 and sneeze at 0.938-2.226, 2.785-4.073, 5.144-6.432 and spraying at 2.713-3.235", "frequencyCaption": "burping belching one times and sneeze three times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_255.wav", "onoffCaption": "duck quacking at 2.246-4.246 and car horn honking at 6.104-8.104", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_401.wav", "onoffCaption": "explosion at 1.188-4.188, 6.166-8.234 and door slamming at 6.603-7.284", "frequencyCaption": "explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_495.wav", "onoffCaption": "tapping clicking clanking at 1.277-4.717 and spraying at 6.204-7.268, 9.239-9.814", "frequencyCaption": "tapping clicking clanking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_603.wav", "onoffCaption": "door slamming at 0.31-1.21", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_672.wav", "onoffCaption": "sheep goat bleating at 2.385-5.705, 7.194-9.194", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_697.wav", "onoffCaption": "car horn honking at 2.257-5.17, 6.998-9.911", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_718.wav", "onoffCaption": "explosion at 0.272-3.272, 3.864-6.864", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_905.wav", "onoffCaption": "dog barking at 0.109-2.109, 2.928-4.928, 6.585-8.585", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_974.wav", "onoffCaption": "cow mooing at 0.23-3.24, 3.796-6.806", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1131.wav", "onoffCaption": "thump thud at 3.383-7.301", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1228.wav", "onoffCaption": "spraying at 2.431-3.181, 4.888-5.829, 7.392-9.784", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1333.wav", "onoffCaption": "tapping clicking clanking at 0.156-3.596", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1499.wav", "onoffCaption": "door knocking at 0.679-2.9, 4.216-6.437, 7.091-9.312", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1567.wav", "onoffCaption": "spraying at 1.396-2.653", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1582.wav", "onoffCaption": "tapping clicking clanking at 0.881-4.321, 5.587-9.027 and cow mooing at 0.881-5.861", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1765.wav", "onoffCaption": "gunshot at 3.529-5.529, 7.826-9.826", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1780.wav", "onoffCaption": "whistling at 0.548-3.523", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1863.wav", "onoffCaption": "dog barking at 0.423-2.423 and whistling at 6.542-10.0", "frequencyCaption": "dog barking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1886.wav", "onoffCaption": "duck quacking at 3.169-5.169 and train horn at 7.932-10.0", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1909.wav", "onoffCaption": "gunshot at 3.317-5.317, 7.169-9.169", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1978.wav", "onoffCaption": "door knocking at 1.1-5.61 and gunshot at 3.2-5.2, 6.765-9.271", "frequencyCaption": "door knocking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3039.wav", "onoffCaption": "thump thud at 0.21-2.71, 3.351-5.851, 6.52-9.02", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_3048.wav", "onoffCaption": "tapping clicking clanking at 2.359-5.799, 7.591-9.639", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3122.wav", "onoffCaption": "door knocking at 0.12-3.24 and duck quacking at 0.249-2.249, 2.872-4.872 and cow mooing at 0.975-3.944", "frequencyCaption": "door knocking one times and duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3153.wav", "onoffCaption": "spraying at 2.568-3.3, 4.971-5.703", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3320.wav", "onoffCaption": "door slamming at 0.339-3.12, 5.228-6.603 and car horn honking at 1.221-6.128", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3351.wav", "onoffCaption": "whistling at 0.082-2.091, 2.634-4.643, 5.173-7.182", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_3505.wav", "onoffCaption": "gunshot at 0.828-2.828, 3.51-5.551, 6.26-8.26", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3574.wav", "onoffCaption": "dog barking at 0.732-2.732, 3.583-5.583, 6.4-8.4", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3707.wav", "onoffCaption": "thump thud at 2.729-5.776, 7.199-9.97", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3776.wav", "onoffCaption": "sneeze at 0.514-3.174, 3.942-6.345, 7.027-8.527 and duck quacking at 3.844-5.844, 6.531-8.531", "frequencyCaption": "sneeze three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3801.wav", "onoffCaption": "woman laughing at 0.433-3.72, 4.347-7.634", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3895.wav", "onoffCaption": "tapping clicking clanking at 0.957-4.397, 6.045-8.62", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_168.wav", "onoffCaption": "gunshot at 0.302-2.302, 3.007-5.007, 5.65-7.65 and sneeze at 1.729-4.804, 5.308-7.304", "frequencyCaption": "gunshot three times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_200.wav", "onoffCaption": "thump thud at 0.027-4.402, 5.833-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_425.wav", "onoffCaption": "whistling at 0.103-7.853", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_470.wav", "onoffCaption": "dog barking at 0.337-2.337 and cat meowing at 1.04-2.051, 2.899-4.443, 4.964-6.15 and gunshot at 2.741-4.741", "frequencyCaption": "dog barking one times and cat meowing three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_656.wav", "onoffCaption": "explosion at 2.865-7.865", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_820.wav", "onoffCaption": "burping belching at 2.113-7.714", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_921.wav", "onoffCaption": "tapping clicking clanking at 3.542-6.982", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_991.wav", "onoffCaption": "gunshot at 3.601-5.642, 7.694-9.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1115.wav", "onoffCaption": "tapping clicking clanking at 1.078-4.518 and cat meowing at 2.764-4.381, 6.623-8.24", "frequencyCaption": "tapping clicking clanking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1140.wav", "onoffCaption": "car horn honking at 0.279-3.192, 5.549-8.462", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1298.wav", "onoffCaption": "cat meowing at 1.184-2.459, 3.186-4.461", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1366.wav", "onoffCaption": "door slamming at 0.202-1.717, 2.995-4.51, 5.792-7.307 and spraying at 1.267-2.524 and woman laughing at 1.332-3.526, 5.309-7.503", "frequencyCaption": "door slamming three times and spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1383.wav", "onoffCaption": "cow mooing at 2.663-5.961", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1458.wav", "onoffCaption": "door knocking at 2.411-6.85", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1516.wav", "onoffCaption": "sneeze at 2.129-4.442, 6.085-8.893", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1543.wav", "onoffCaption": "car horn honking at 1.994-4.341, 6.759-9.106", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1730.wav", "onoffCaption": "door slamming at 0.66-1.913, 3.512-5.512 and spraying at 7.065-7.666", "frequencyCaption": "door slamming two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1812.wav", "onoffCaption": "sneeze at 0.01-2.134, 4.32-6.444", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1847.wav", "onoffCaption": "train horn at 2.659-6.419", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3076.wav", "onoffCaption": "cow mooing at 0.452-5.432, 7.721-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3089.wav", "onoffCaption": "explosion at 2.8-5.4, 7.474-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3177.wav", "onoffCaption": "thump thud at 1.086-4.133", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3192.wav", "onoffCaption": "woman laughing at 3.884-6.239, 6.838-9.193", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3304.wav", "onoffCaption": "train horn at 2.616-6.816, 7.46-9.94", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3420.wav", "onoffCaption": "door knocking at 0.306-3.153, 5.11-7.174 and car horn honking at 4.113-7.608", "frequencyCaption": "door knocking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3521.wav", "onoffCaption": "tapping clicking clanking at 1.082-4.522, 5.042-7.335", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3591.wav", "onoffCaption": "door knocking at 2.49-5.546 and car horn honking at 7.149-10.0", "frequencyCaption": "door knocking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3649.wav", "onoffCaption": "dog barking at 0.503-3.823, 4.345-7.665", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3752.wav", "onoffCaption": "sheep goat bleating at 1.849-3.849, 6.339-8.339 and duck quacking at 2.367-4.367, 5.96-7.96", "frequencyCaption": "sheep goat bleating two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3825.wav", "onoffCaption": "tapping clicking clanking at 0.088-3.528, 4.234-7.674 and sneeze at 0.459-2.455, 3.231-5.593, 6.22-7.394", "frequencyCaption": "tapping clicking clanking two times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3870.wav", "onoffCaption": "dog barking at 0.159-2.159 and burping belching at 3.641-10.0", "frequencyCaption": "dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_55.wav", "onoffCaption": "car horn honking at 0.516-3.302, 4.306-7.092", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_127.wav", "onoffCaption": "explosion at 0.637-2.701, 3.39-5.454, 6.295-8.359", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_172.wav", "onoffCaption": "woman laughing at 0.259-2.678, 3.762-6.181, 7.293-9.712", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_197.wav", "onoffCaption": "sheep goat bleating at 1.284-3.284, 3.836-5.836, 7.193-9.193", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_301.wav", "onoffCaption": "door knocking at 3.503-6.57", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_354.wav", "onoffCaption": "explosion at 2.583-6.136, 6.737-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_524.wav", "onoffCaption": "duck quacking at 0.427-2.427, 3.177-5.177, 6.113-8.113 and burping belching at 2.667-5.667, 6.53-9.53", "frequencyCaption": "duck quacking three times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_571.wav", "onoffCaption": "cow mooing at 0.128-3.138, 4.344-7.354 and dog barking at 4.593-6.593", "frequencyCaption": "cow mooing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_594.wav", "onoffCaption": "cat meowing at 2.648-3.66, 4.641-5.653", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_619.wav", "onoffCaption": "whistling at 0.822-7.241", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_702.wav", "onoffCaption": "duck quacking at 0.54-2.54", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_757.wav", "onoffCaption": "cow mooing at 0.723-3.705, 4.306-6.958 and train horn at 1.18-7.649", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_875.wav", "onoffCaption": "duck quacking at 1.336-3.336, 5.54-7.54 and car horn honking at 1.58-6.487, 7.594-9.594 and gunshot at 2.079-4.079, 5.251-7.251", "frequencyCaption": "duck quacking two times and car horn honking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_890.wav", "onoffCaption": "sneeze at 3.586-5.505, 6.712-8.631", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1014.wav", "onoffCaption": "train horn at 0.202-2.357 and spraying at 6.382-6.882", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1041.wav", "onoffCaption": "cat meowing at 0.344-1.353 and sheep goat bleating at 4.842-6.842, 7.989-9.989", "frequencyCaption": "cat meowing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1232.wav", "onoffCaption": "explosion at 0.237-5.237", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1267.wav", "onoffCaption": "car horn honking at 0.041-3.882, 5.425-7.938 and spraying at 2.028-2.895", "frequencyCaption": "car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1282.wav", "onoffCaption": "gunshot at 1.822-3.822 and cow mooing at 2.568-7.548", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1329.wav", "onoffCaption": "door knocking at 0.595-3.651, 4.593-6.72", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1399.wav", "onoffCaption": "sneeze at 0.117-3.765, 4.445-5.537 and explosion at 0.46-3.46 and car horn honking at 1.044-3.044", "frequencyCaption": "sneeze two times and explosion one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1417.wav", "onoffCaption": "sheep goat bleating at 1.462-3.462, 4.727-6.727 and tapping clicking clanking at 2.217-5.657, 7.32-10.0", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1442.wav", "onoffCaption": "tapping clicking clanking at 0.277-3.717, 5.707-9.147", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1559.wav", "onoffCaption": "thump thud at 0.397-4.064, 5.612-9.279", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1631.wav", "onoffCaption": "car horn honking at 0.189-2.702", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1664.wav", "onoffCaption": "train horn at 0.002-4.07", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1808.wav", "onoffCaption": "woman laughing at 2.822-5.876, 6.644-9.698", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1913.wav", "onoffCaption": "sneeze at 1.993-3.239, 4.458-6.571, 7.242-9.645", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1946.wav", "onoffCaption": "burping belching at 2.326-6.662 and door slamming at 4.842-5.82", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3023.wav", "onoffCaption": "door slamming at 0.184-2.312, 3.655-5.783, 6.912-9.04", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3093.wav", "onoffCaption": "woman laughing at 0.977-3.569", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3138.wav", "onoffCaption": "train horn at 2.088-7.165 and explosion at 2.61-5.61 and door slamming at 3.402-4.655", "frequencyCaption": "train horn one times and explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3188.wav", "onoffCaption": "sneeze at 0.033-1.321, 2.162-4.776, 5.79-7.402", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3205.wav", "onoffCaption": "thump thud at 1.166-4.213, 4.916-7.242 and gunshot at 6.167-8.167", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3250.wav", "onoffCaption": "duck quacking at 0.323-2.323, 4.398-6.398", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3475.wav", "onoffCaption": "woman laughing at 0.214-2.806, 3.565-5.773", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3490.wav", "onoffCaption": "spraying at 0.003-0.784, 2.076-4.204, 5.949-6.457", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3606.wav", "onoffCaption": "cow mooing at 0.694-3.676, 5.839-8.821", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3653.wav", "onoffCaption": "thump thud at 0.056-3.974, 5.313-9.231", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3748.wav", "onoffCaption": "door slamming at 0.261-1.74", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3924.wav", "onoffCaption": "woman laughing at 0.478-3.532, 5.351-8.05", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3971.wav", "onoffCaption": "sheep goat bleating at 1.131-3.131 and thump thud at 4.915-9.365 and dog barking at 6.582-8.582", "frequencyCaption": "sheep goat bleating one times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3994.wav", "onoffCaption": "burping belching at 1.116-3.739, 5.597-8.418", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_14.wav", "onoffCaption": "gunshot at 1.876-3.876, 5.278-7.278 and explosion at 2.328-5.057", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_133.wav", "onoffCaption": "cow mooing at 1.851-6.831", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_166.wav", "onoffCaption": "car horn honking at 1.533-3.88, 6.373-8.72 and tapping clicking clanking at 3.047-6.487", "frequencyCaption": "car horn honking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_183.wav", "onoffCaption": "spraying at 0.373-1.314, 2.792-3.733 and cat meowing at 2.599-3.611", "frequencyCaption": "spraying two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_315.wav", "onoffCaption": "door slamming at 0.256-1.489, 2.931-4.164", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_340.wav", "onoffCaption": "door slamming at 2.894-5.323 and cow mooing at 7.457-10.0", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_530.wav", "onoffCaption": "whistling at 0.753-9.138", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_565.wav", "onoffCaption": "cow mooing at 0.66-3.629 and woman laughing at 7.388-10.0", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_580.wav", "onoffCaption": "door slamming at 3.578-6.578, 7.636-9.832", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_658.wav", "onoffCaption": "explosion at 2.357-7.357", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_716.wav", "onoffCaption": "dog barking at 0.108-2.108, 2.668-4.668, 5.421-7.421 and gunshot at 1.186-3.186, 5.192-7.192 and door knocking at 4.129-6.193", "frequencyCaption": "dog barking three times and gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_743.wav", "onoffCaption": "cow mooing at 0.853-3.835, 5.646-8.205", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_834.wav", "onoffCaption": "duck quacking at 2.186-4.186, 5.978-7.978", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_861.wav", "onoffCaption": "thump thud at 2.521-6.439, 7.515-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1000.wav", "onoffCaption": "gunshot at 2.613-4.613, 6.416-8.416 and door knocking at 3.178-5.399, 6.317-8.686", "frequencyCaption": "gunshot two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1055.wav", "onoffCaption": "door slamming at 1.103-2.22, 4.078-5.195, 7.409-8.526", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1226.wav", "onoffCaption": "dog barking at 1.16-3.16", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1273.wav", "onoffCaption": "whistling at 0.636-2.865", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1296.wav", "onoffCaption": "whistling at 0.45-2.679", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1368.wav", "onoffCaption": "door knocking at 2.254-4.656, 6.714-8.841", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1456.wav", "onoffCaption": "tapping clicking clanking at 1.369-4.809", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1518.wav", "onoffCaption": "explosion at 0.098-5.098 and tapping clicking clanking at 0.832-4.272", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1625.wav", "onoffCaption": "cat meowing at 0.696-2.446, 3.021-4.771, 5.412-7.162", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1670.wav", "onoffCaption": "burping belching at 2.796-4.919, 5.517-7.64", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1681.wav", "onoffCaption": "cow mooing at 1.387-4.356, 6.831-9.813", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1695.wav", "onoffCaption": "gunshot at 0.724-2.724, 4.717-6.717 and duck quacking at 1.185-3.185, 4.117-6.117, 7.028-9.028", "frequencyCaption": "gunshot two times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1849.wav", "onoffCaption": "thump thud at 0.642-5.092, 5.855-7.909", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1907.wav", "onoffCaption": "explosion at 1.151-6.151", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1952.wav", "onoffCaption": "door knocking at 0.422-3.975 and duck quacking at 1.2-3.2 and car horn honking at 6.424-9.242", "frequencyCaption": "door knocking one times and duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3037.wav", "onoffCaption": "duck quacking at 0.241-2.241, 3.858-5.858 and thump thud at 4.667-6.895, 7.566-10.0", "frequencyCaption": "duck quacking two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3062.wav", "onoffCaption": "woman laughing at 2.849-5.049", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3087.wav", "onoffCaption": "duck quacking at 0.031-2.031, 3.164-5.164 and burping belching at 7.688-10.0", "frequencyCaption": "duck quacking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3179.wav", "onoffCaption": "gunshot at 1.377-3.377, 4.054-6.054", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3211.wav", "onoffCaption": "sheep goat bleating at 0.633-4.553, 5.305-7.305 and tapping clicking clanking at 1.74-5.18, 7.248-10.0", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3244.wav", "onoffCaption": "cow mooing at 2.773-6.071, 7.377-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3434.wav", "onoffCaption": "duck quacking at 3.054-5.054, 7.036-9.036", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3461.wav", "onoffCaption": "door slamming at 2.061-4.842, 6.857-9.638", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3484.wav", "onoffCaption": "duck quacking at 0.197-2.197, 2.792-4.792, 5.774-7.774 and thump thud at 3.437-6.484", "frequencyCaption": "duck quacking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3612.wav", "onoffCaption": "whistling at 0.85-5.334, 7.349-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3647.wav", "onoffCaption": "dog barking at 0.531-3.851, 5.617-7.617 and train horn at 0.823-3.303, 4.388-6.934 and tapping clicking clanking at 2.164-5.604", "frequencyCaption": "dog barking two times and train horn two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3709.wav", "onoffCaption": "train horn at 0.105-2.242, 2.805-4.942, 5.66-7.797", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_3930.wav", "onoffCaption": "door knocking at 0.868-4.421", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3965.wav", "onoffCaption": "woman laughing at 0.226-3.614, 4.453-7.841 and cow mooing at 3.081-6.05 and door slamming at 3.829-5.829", "frequencyCaption": "woman laughing two times and cow mooing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3980.wav", "onoffCaption": "door slamming at 0.164-3.138 and sheep goat bleating at 0.911-2.911, 5.018-7.018 and car horn honking at 2.298-7.205", "frequencyCaption": "door slamming one times and sheep goat bleating two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_41.wav", "onoffCaption": "tapping clicking clanking at 2.541-5.981", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_129.wav", "onoffCaption": "dog barking at 1.053-3.053, 4.927-6.927 and cat meowing at 4.681-9.041 and sneeze at 6.661-8.273", "frequencyCaption": "dog barking two times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_199.wav", "onoffCaption": "woman laughing at 0.853-7.587 and door knocking at 2.12-4.21 and door slamming at 6.068-7.185, 7.926-9.043", "frequencyCaption": "woman laughing one times and door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_214.wav", "onoffCaption": "sneeze at 2.226-3.838, 4.85-6.395", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_241.wav", "onoffCaption": "explosion at 0.978-5.978", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_431.wav", "onoffCaption": "spraying at 2.906-3.847, 4.661-5.602, 6.824-7.765", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_464.wav", "onoffCaption": "door knocking at 0.881-3.728, 4.455-7.302 and door slamming at 1.699-4.578, 5.851-6.656 and explosion at 2.481-6.481", "frequencyCaption": "door knocking two times and door slamming two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_481.wav", "onoffCaption": "gunshot at 2.466-4.466, 5.127-7.127 and door knocking at 3.505-8.207 and train horn at 3.535-6.935", "frequencyCaption": "gunshot two times and door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_617.wav", "onoffCaption": "spraying at 2.92-3.57, 5.893-6.543, 8.392-9.042", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_642.wav", "onoffCaption": "gunshot at 2.277-4.277, 5.537-7.537 and cat meowing at 6.26-7.26", "frequencyCaption": "gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_759.wav", "onoffCaption": "door slamming at 3.234-5.46 and gunshot at 5.625-7.625", "frequencyCaption": "door slamming one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_884.wav", "onoffCaption": "spraying at 0.199-0.699, 1.763-2.263, 2.935-3.435 and cat meowing at 0.762-2.333, 2.93-4.501", "frequencyCaption": "spraying three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_935.wav", "onoffCaption": "sheep goat bleating at 2.538-4.538, 6.466-8.466", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_960.wav", "onoffCaption": "sneeze at 0.1-3.21, 4.615-6.7", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_985.wav", "onoffCaption": "burping belching at 3.179-6.358, 7.834-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1101.wav", "onoffCaption": "sneeze at 0.201-2.44, 3.115-5.028, 5.81-7.895 and gunshot at 1.528-3.528", "frequencyCaption": "sneeze three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1154.wav", "onoffCaption": "spraying at 0.02-1.104 and whistling at 0.446-5.946", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1170.wav", "onoffCaption": "spraying at 2.656-4.418, 6.913-8.17", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1269.wav", "onoffCaption": "duck quacking at 0.627-2.627, 3.358-5.358", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1327.wav", "onoffCaption": "burping belching at 0.01-2.135, 3.537-5.662, 6.649-8.774 and door knocking at 1.426-3.586", "frequencyCaption": "burping belching three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1372.wav", "onoffCaption": "dog barking at 2.736-4.736, 6.873-8.873", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1397.wav", "onoffCaption": "cat meowing at 0.314-1.898, 2.63-4.214", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1403.wav", "onoffCaption": "explosion at 1.981-4.734, 6.313-9.313", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1419.wav", "onoffCaption": "train horn at 0.012-4.012 and spraying at 6.3-7.381", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1502.wav", "onoffCaption": "duck quacking at 0.082-2.082 and car horn honking at 0.559-4.054 and spraying at 5.939-8.523", "frequencyCaption": "duck quacking one times and car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1526.wav", "onoffCaption": "thump thud at 0.761-2.989, 4.754-6.982", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1557.wav", "onoffCaption": "train horn at 0.528-8.728", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1724.wav", "onoffCaption": "door slamming at 0.416-1.267, 2.1-2.951, 4.414-5.265", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1771.wav", "onoffCaption": "thump thud at 0.394-2.622, 3.385-5.613, 6.502-8.73", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1794.wav", "onoffCaption": "door knocking at 0.846-3.576, 5.727-8.351", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1806.wav", "onoffCaption": "explosion at 1.47-6.47 and door slamming at 1.716-4.497, 5.429-6.908 and burping belching at 3.429-7.429", "frequencyCaption": "explosion one times and door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1853.wav", "onoffCaption": "cow mooing at 0.434-3.416, 3.954-6.789, 7.523-9.86", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_1948.wav", "onoffCaption": "train horn at 1.774-6.851 and tapping clicking clanking at 3.442-6.882", "frequencyCaption": "train horn one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3078.wav", "onoffCaption": "sneeze at 0.026-1.483 and thump thud at 3.614-6.114, 7.183-10.0", "frequencyCaption": "sneeze one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3136.wav", "onoffCaption": "train horn at 0.866-5.768, 7.973-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3163.wav", "onoffCaption": "car horn honking at 0.131-2.478", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3186.wav", "onoffCaption": "spraying at 1.654-3.416, 4.687-7.123 and cat meowing at 2.384-5.625, 7.118-10.0", "frequencyCaption": "spraying two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3310.wav", "onoffCaption": "train horn at 1.864-5.398 and door knocking at 4.175-6.239, 6.937-9.001", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3345.wav", "onoffCaption": "door knocking at 3.904-8.176", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3535.wav", "onoffCaption": "gunshot at 1.061-3.061, 4.558-6.558 and duck quacking at 3.044-5.044", "frequencyCaption": "gunshot two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3560.wav", "onoffCaption": "car horn honking at 0.198-4.71, 6.293-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3585.wav", "onoffCaption": "cat meowing at 2.877-4.017", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3608.wav", "onoffCaption": "cow mooing at 2.655-5.953 and sneeze at 7.997-10.0", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3713.wav", "onoffCaption": "whistling at 0.531-5.015, 7.0-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3746.wav", "onoffCaption": "car horn honking at 0.427-3.213", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3831.wav", "onoffCaption": "sheep goat bleating at 2.297-4.297, 6.101-8.101", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3864.wav", "onoffCaption": "sneeze at 0.203-1.703", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3881.wav", "onoffCaption": "sneeze at 0.434-2.117, 2.96-4.643, 5.251-6.934", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_158.wav", "onoffCaption": "sheep goat bleating at 0.22-2.22, 4.379-6.379, 7.198-9.198", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_230.wav", "onoffCaption": "thump thud at 0.17-2.509, 4.2-6.539", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_265.wav", "onoffCaption": "burping belching at 0.549-2.58, 3.304-5.335 and cow mooing at 1.658-4.64, 5.717-8.699", "frequencyCaption": "burping belching two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_280.wav", "onoffCaption": "sheep goat bleating at 3.352-5.352, 6.27-9.27", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_331.wav", "onoffCaption": "woman laughing at 0.703-2.986, 4.385-6.668 and burping belching at 3.892-6.892", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_415.wav", "onoffCaption": "gunshot at 0.476-2.476 and cat meowing at 0.509-2.503, 3.064-4.648, 5.727-7.962", "frequencyCaption": "gunshot one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_440.wav", "onoffCaption": "duck quacking at 0.184-2.184, 3.112-5.112", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_633.wav", "onoffCaption": "cat meowing at 0.287-1.434, 2.463-3.61", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_666.wav", "onoffCaption": "cow mooing at 2.642-5.94", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_683.wav", "onoffCaption": "door slamming at 2.21-3.601", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_699.wav", "onoffCaption": "door slamming at 2.043-4.269, 4.82-6.211, 8.048-9.572", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_728.wav", "onoffCaption": "train horn at 2.321-6.389, 7.291-9.903", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_767.wav", "onoffCaption": "duck quacking at 0.926-2.926, 4.515-6.515 and cat meowing at 1.965-3.925", "frequencyCaption": "duck quacking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_798.wav", "onoffCaption": "whistling at 0.489-6.899", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_911.wav", "onoffCaption": "cat meowing at 0.48-1.666 and whistling at 0.61-10.0", "frequencyCaption": "cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_944.wav", "onoffCaption": "burping belching at 1.37-5.393, 6.183-9.183", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1125.wav", "onoffCaption": "spraying at 2.52-3.371, 4.122-6.582, 7.329-7.956", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1195.wav", "onoffCaption": "door knocking at 0.96-3.2", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1218.wav", "onoffCaption": "tapping clicking clanking at 1.668-5.108", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1257.wav", "onoffCaption": "sneeze at 1.556-2.844, 3.917-6.002 and duck quacking at 2.435-4.435, 5.674-7.674 and spraying at 3.075-3.659, 5.829-6.413", "frequencyCaption": "sneeze two times and duck quacking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1303.wav", "onoffCaption": "car horn honking at 1.45-3.797 and spraying at 2.235-3.086 and dog barking at 6.931-9.852", "frequencyCaption": "car horn honking one times and spraying one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1356.wav", "onoffCaption": "door knocking at 3.378-6.931", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1468.wav", "onoffCaption": "cat meowing at 0.276-1.378, 2.979-4.245 and gunshot at 7.143-9.143", "frequencyCaption": "cat meowing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1573.wav", "onoffCaption": "dog barking at 0.216-2.216, 3.355-5.355, 6.148-8.148 and spraying at 0.276-0.845 and tapping clicking clanking at 5.304-8.744", "frequencyCaption": "dog barking three times and spraying one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1596.wav", "onoffCaption": "cat meowing at 2.368-3.554, 4.623-7.014", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1601.wav", "onoffCaption": "door knocking at 0.127-2.149 and cow mooing at 3.366-6.376", "frequencyCaption": "door knocking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1700.wav", "onoffCaption": "whistling at 1.146-8.764", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1755.wav", "onoffCaption": "sneeze at 0.305-1.892, 4.262-6.375", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1822.wav", "onoffCaption": "train horn at 1.537-4.417, 6.155-9.035", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1877.wav", "onoffCaption": "gunshot at 2.109-4.109 and whistling at 7.924-10.0", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1892.wav", "onoffCaption": "explosion at 0.321-3.377", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1939.wav", "onoffCaption": "door knocking at 0.279-3.126, 3.663-6.51", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1989.wav", "onoffCaption": "sheep goat bleating at 0.042-2.042 and woman laughing at 4.477-7.069", "frequencyCaption": "sheep goat bleating one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3009.wav", "onoffCaption": "sheep goat bleating at 0.509-2.509", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3112.wav", "onoffCaption": "woman laughing at 3.607-6.212 and duck quacking at 5.845-7.845", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3147.wav", "onoffCaption": "whistling at 2.305-6.789, 7.817-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3334.wav", "onoffCaption": "spraying at 0.947-1.697, 3.096-3.846, 4.38-5.13 and whistling at 7.494-10.0", "frequencyCaption": "spraying three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3361.wav", "onoffCaption": "dog barking at 0.659-2.659, 4.075-6.075 and cow mooing at 2.011-6.991, 7.547-10.0", "frequencyCaption": "dog barking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3384.wav", "onoffCaption": "gunshot at 0.991-2.991 and spraying at 3.058-4.753", "frequencyCaption": "gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3511.wav", "onoffCaption": "cat meowing at 0.223-1.232, 1.827-2.836 and whistling at 0.758-8.413", "frequencyCaption": "cat meowing two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3544.wav", "onoffCaption": "tapping clicking clanking at 1.273-4.713, 5.387-8.827 and car horn honking at 1.592-3.939 and sheep goat bleating at 1.733-3.733", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3679.wav", "onoffCaption": "spraying at 0.11-0.737, 1.279-1.787, 3.007-5.467 and door knocking at 7.869-10.0", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3686.wav", "onoffCaption": "explosion at 1.044-6.044, 6.567-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3737.wav", "onoffCaption": "sheep goat bleating at 0.425-2.425, 4.108-6.108 and door slamming at 7.848-8.748", "frequencyCaption": "sheep goat bleating two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3762.wav", "onoffCaption": "duck quacking at 2.729-4.729", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3787.wav", "onoffCaption": "burping belching at 2.836-6.095, 7.143-9.467", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3815.wav", "onoffCaption": "cow mooing at 2.141-5.151 and explosion at 2.372-7.372", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3840.wav", "onoffCaption": "duck quacking at 1.918-3.918, 5.313-7.313", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_30.wav", "onoffCaption": "duck quacking at 1.581-3.581, 5.379-7.379", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_65.wav", "onoffCaption": "gunshot at 0.859-2.859", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_80.wav", "onoffCaption": "sneeze at 0.482-2.478, 3.427-5.423, 7.453-9.449 and explosion at 0.882-3.754, 4.584-6.904", "frequencyCaption": "sneeze three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_117.wav", "onoffCaption": "explosion at 1.659-4.715, 6.7-9.756", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_142.wav", "onoffCaption": "spraying at 0.225-0.733, 1.499-2.007 and dog barking at 0.699-4.019, 4.823-8.143 and door knocking at 0.984-4.734, 5.73-7.951", "frequencyCaption": "spraying two times and dog barking two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_364.wav", "onoffCaption": "car horn honking at 0.766-3.692, 5.467-8.228", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_381.wav", "onoffCaption": "thump thud at 0.034-3.952, 6.217-8.717 and gunshot at 1.044-3.085", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_514.wav", "onoffCaption": "cow mooing at 0.534-3.516 and door slamming at 3.555-4.574, 6.074-7.093", "frequencyCaption": "cow mooing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_541.wav", "onoffCaption": "car horn honking at 2.908-6.749", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_629.wav", "onoffCaption": "burping belching at 1.02-6.144, 6.845-9.4", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_732.wav", "onoffCaption": "explosion at 0.038-2.126 and thump thud at 0.211-2.982, 4.408-6.482", "frequencyCaption": "explosion one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_782.wav", "onoffCaption": "thump thud at 1.036-3.375, 4.184-6.955 and cat meowing at 2.618-5.859, 7.36-10.0", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_810.wav", "onoffCaption": "dog barking at 0.258-2.258, 4.132-6.132, 7.99-9.99", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_845.wav", "onoffCaption": "thump thud at 2.152-6.07", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1024.wav", "onoffCaption": "woman laughing at 0.196-2.277, 4.543-6.624", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1071.wav", "onoffCaption": "door slamming at 0.205-2.205, 3.528-5.528, 6.22-7.473 and duck quacking at 2.613-4.613", "frequencyCaption": "door slamming three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1094.wav", "onoffCaption": "sheep goat bleating at 2.335-4.335, 5.743-7.743", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1202.wav", "onoffCaption": "cat meowing at 0.067-1.076", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1319.wav", "onoffCaption": "train horn at 0.993-3.873, 4.733-7.613", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1472.wav", "onoffCaption": "thump thud at 0.435-4.353 and sheep goat bleating at 1.679-3.679, 5.085-7.085", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1497.wav", "onoffCaption": "door slamming at 0.326-2.326, 4.646-6.646", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1569.wav", "onoffCaption": "cow mooing at 0.08-3.09, 4.676-7.686 and door slamming at 7.197-9.197", "frequencyCaption": "cow mooing two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1654.wav", "onoffCaption": "tapping clicking clanking at 3.172-6.612", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1838.wav", "onoffCaption": "sheep goat bleating at 1.288-4.608, 6.25-9.57 and duck quacking at 2.365-4.365", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1888.wav", "onoffCaption": "tapping clicking clanking at 0.995-4.435 and cow mooing at 2.244-5.226, 6.414-9.383", "frequencyCaption": "tapping clicking clanking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1923.wav", "onoffCaption": "gunshot at 1.011-3.011, 4.239-6.239, 7.057-9.057", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1976.wav", "onoffCaption": "cat meowing at 2.801-4.795, 5.356-7.544, 8.821-9.961", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1993.wav", "onoffCaption": "woman laughing at 2.007-4.089, 6.48-8.562", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3013.wav", "onoffCaption": "sheep goat bleating at 2.166-4.166 and cow mooing at 4.473-7.442 and gunshot at 6.86-8.86", "frequencyCaption": "sheep goat bleating one times and cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3046.wav", "onoffCaption": "sheep goat bleating at 0.979-2.979 and door slamming at 5.08-6.604, 7.268-9.697", "frequencyCaption": "sheep goat bleating one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3108.wav", "onoffCaption": "spraying at 0.384-2.512, 4.662-7.003 and sneeze at 0.575-2.821", "frequencyCaption": "spraying two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3235.wav", "onoffCaption": "cat meowing at 0.814-1.825", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3260.wav", "onoffCaption": "cat meowing at 0.62-2.56, 3.135-5.075, 6.248-8.188", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3285.wav", "onoffCaption": "gunshot at 0.533-2.533", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3410.wav", "onoffCaption": "duck quacking at 0.698-2.698, 4.798-6.798", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3445.wav", "onoffCaption": "cow mooing at 1.175-4.185, 4.703-7.672 and door knocking at 6.169-8.549", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3636.wav", "onoffCaption": "train horn at 1.218-3.873, 5.846-8.552", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3663.wav", "onoffCaption": "explosion at 2.042-4.049", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3778.wav", "onoffCaption": "train horn at 2.917-6.237, 7.741-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3914.wav", "onoffCaption": "thump thud at 0.368-2.868, 3.643-6.143, 7.238-9.738", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_3941.wav", "onoffCaption": "car horn honking at 0.289-5.196, 7.372-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_37.wav", "onoffCaption": "burping belching at 0.202-5.202, 6.756-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_62.wav", "onoffCaption": "door slamming at 0.077-0.758, 1.309-3.309 and car horn honking at 7.208-10.0", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_87.wav", "onoffCaption": "cat meowing at 3.989-5.929, 7.639-9.389", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_110.wav", "onoffCaption": "woman laughing at 1.572-3.654, 4.928-7.01", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_145.wav", "onoffCaption": "cat meowing at 2.595-3.808", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_336.wav", "onoffCaption": "spraying at 2.725-3.725", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_408.wav", "onoffCaption": "spraying at 0.198-0.979, 2.793-3.42, 4.839-6.572", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_513.wav", "onoffCaption": "thump thud at 0.607-4.525, 5.536-9.454 and woman laughing at 3.351-7.403", "frequencyCaption": "thump thud two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_546.wav", "onoffCaption": "burping belching at 0.499-2.597", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_760.wav", "onoffCaption": "car horn honking at 2.233-6.555, 7.446-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_785.wav", "onoffCaption": "sheep goat bleating at 0.214-2.214 and tapping clicking clanking at 4.133-7.573", "frequencyCaption": "sheep goat bleating one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_817.wav", "onoffCaption": "burping belching at 0.67-2.768, 5.205-7.33", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_842.wav", "onoffCaption": "explosion at 3.225-6.225", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_959.wav", "onoffCaption": "door knocking at 0.278-2.581, 4.559-6.961, 7.965-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1023.wav", "onoffCaption": "spraying at 0.154-1.005, 3.277-4.972", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1076.wav", "onoffCaption": "woman laughing at 0.237-7.249, 7.811-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1093.wav", "onoffCaption": "spraying at 0.32-1.052, 3.287-5.415 and burping belching at 6.898-10.0", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1138.wav", "onoffCaption": "car horn honking at 0.532-2.532, 3.585-6.576, 7.794-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1188.wav", "onoffCaption": "cow mooing at 3.604-8.033", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1250.wav", "onoffCaption": "sheep goat bleating at 3.127-5.127, 5.942-7.942", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1420.wav", "onoffCaption": "door slamming at 3.092-4.07, 5.862-6.713 and thump thud at 5.014-7.514", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1427.wav", "onoffCaption": "tapping clicking clanking at 3.687-7.127", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1475.wav", "onoffCaption": "dog barking at 2.524-4.524 and duck quacking at 2.891-4.891, 5.41-7.41", "frequencyCaption": "dog barking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1490.wav", "onoffCaption": "woman laughing at 0.474-3.269", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1606.wav", "onoffCaption": "train horn at 3.052-6.372", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1924.wav", "onoffCaption": "woman laughing at 0.787-3.353, 4.339-6.905 and door knocking at 2.189-5.957", "frequencyCaption": "woman laughing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1971.wav", "onoffCaption": "duck quacking at 0.795-2.795, 4.693-6.693", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1994.wav", "onoffCaption": "burping belching at 0.079-2.11", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3014.wav", "onoffCaption": "duck quacking at 2.193-4.193, 4.922-6.922, 7.799-9.799", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3041.wav", "onoffCaption": "door slamming at 1.161-2.926 and whistling at 7.103-10.0", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3232.wav", "onoffCaption": "cat meowing at 2.11-4.05, 5.92-7.86", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3267.wav", "onoffCaption": "thump thud at 0.009-4.459, 6.04-8.228", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3329.wav", "onoffCaption": "thump thud at 0.967-4.014, 6.031-9.078", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3417.wav", "onoffCaption": "spraying at 0.093-1.034", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3442.wav", "onoffCaption": "gunshot at 0.47-2.47, 4.029-6.029 and cow mooing at 4.86-7.842", "frequencyCaption": "gunshot two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3559.wav", "onoffCaption": "burping belching at 0.082-2.176, 4.473-7.294 and woman laughing at 0.393-2.676", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3631.wav", "onoffCaption": "sneeze at 3.473-4.761, 6.665-9.279", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3664.wav", "onoffCaption": "tapping clicking clanking at 2.176-5.616, 6.881-9.797", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3681.wav", "onoffCaption": "tapping clicking clanking at 0.021-3.461", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3808.wav", "onoffCaption": "car horn honking at 0.792-3.139 and woman laughing at 1.428-3.783, 4.399-7.212", "frequencyCaption": "car horn honking one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3913.wav", "onoffCaption": "train horn at 0.685-4.045", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3946.wav", "onoffCaption": "car horn honking at 0.998-3.911 and thump thud at 2.478-6.396", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_78.wav", "onoffCaption": "door slamming at 2.514-3.494, 4.288-5.268, 6.573-7.553 and thump thud at 3.927-6.698", "frequencyCaption": "door slamming three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_237.wav", "onoffCaption": "dog barking at 1.736-3.736, 5.817-7.817 and door slamming at 3.506-4.006, 4.74-5.24", "frequencyCaption": "dog barking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_262.wav", "onoffCaption": "explosion at 0.161-2.879", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_278.wav", "onoffCaption": "whistling at 0.888-3.863, 5.403-8.378 and spraying at 1.588-3.98", "frequencyCaption": "whistling two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_287.wav", "onoffCaption": "train horn at 2.767-6.887", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_363.wav", "onoffCaption": "car horn honking at 0.079-2.897, 4.356-7.174 and thump thud at 5.249-7.588", "frequencyCaption": "car horn honking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_379.wav", "onoffCaption": "cat meowing at 0.541-3.445 and sheep goat bleating at 3.471-5.471, 7.835-9.835 and burping belching at 5.151-8.651", "frequencyCaption": "cat meowing one times and sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_386.wav", "onoffCaption": "tapping clicking clanking at 3.089-6.529", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_447.wav", "onoffCaption": "door knocking at 1.369-7.429 and cat meowing at 1.777-3.725, 5.36-7.308", "frequencyCaption": "door knocking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_634.wav", "onoffCaption": "sheep goat bleating at 3.048-5.048, 6.235-8.355 and woman laughing at 4.232-6.348, 7.658-9.852", "frequencyCaption": "sheep goat bleating two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_661.wav", "onoffCaption": "car horn honking at 0.639-5.039, 6.08-8.118", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_684.wav", "onoffCaption": "whistling at 1.082-8.737", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_735.wav", "onoffCaption": "thump thud at 2.444-4.672, 5.658-7.997 and gunshot at 2.509-4.528", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_858.wav", "onoffCaption": "woman laughing at 1.337-3.62 and spraying at 6.255-7.339", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_916.wav", "onoffCaption": "duck quacking at 0.154-2.154, 3.43-5.43", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_943.wav", "onoffCaption": "thump thud at 0.617-4.992", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1089.wav", "onoffCaption": "duck quacking at 0.838-2.838, 4.658-6.658 and sheep goat bleating at 3.276-5.276", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1177.wav", "onoffCaption": "door slamming at 0.485-2.398, 4.742-7.703", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1192.wav", "onoffCaption": "burping belching at 3.097-7.097", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1205.wav", "onoffCaption": "door slamming at 1.534-2.434, 3.424-4.324 and burping belching at 6.945-9.945", "frequencyCaption": "door slamming two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1304.wav", "onoffCaption": "sneeze at 0.136-2.382, 3.132-5.378, 6.294-8.54 and cat meowing at 3.431-5.391, 6.289-8.237", "frequencyCaption": "sneeze three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1351.wav", "onoffCaption": "door slamming at 0.002-1.002, 2.305-4.305", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1521.wav", "onoffCaption": "sheep goat bleating at 0.099-2.099 and whistling at 0.517-8.267", "frequencyCaption": "sheep goat bleating one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1591.wav", "onoffCaption": "cow mooing at 0.046-3.015, 4.496-7.465, 7.986-10.0 and door slamming at 0.486-1.719", "frequencyCaption": "cow mooing three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1649.wav", "onoffCaption": "sheep goat bleating at 1.382-3.382 and duck quacking at 6.325-8.325", "frequencyCaption": "sheep goat bleating one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1653.wav", "onoffCaption": "duck quacking at 2.236-4.236", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1707.wav", "onoffCaption": "train horn at 0.201-3.001 and car horn honking at 0.44-2.44", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1748.wav", "onoffCaption": "car horn honking at 1.705-5.292 and cow mooing at 2.18-7.16", "frequencyCaption": "car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1752.wav", "onoffCaption": "burping belching at 1.373-5.242", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1825.wav", "onoffCaption": "door knocking at 2.788-5.037, 6.324-8.552", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1870.wav", "onoffCaption": "duck quacking at 2.886-4.886, 7.079-9.079", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3115.wav", "onoffCaption": "train horn at 1.259-4.499, 5.745-8.4", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3140.wav", "onoffCaption": "whistling at 0.617-2.626, 3.818-6.71 and door knocking at 2.508-5.132, 6.443-9.067", "frequencyCaption": "whistling two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3228.wav", "onoffCaption": "duck quacking at 1.877-3.877", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3282.wav", "onoffCaption": "cat meowing at 0.19-1.745, 3.887-5.442 and woman laughing at 7.712-9.81", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3298.wav", "onoffCaption": "door slamming at 3.537-6.511", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3333.wav", "onoffCaption": "door slamming at 0.342-1.515, 2.303-3.452, 4.241-6.437", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3366.wav", "onoffCaption": "car horn honking at 3.033-7.282", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3383.wav", "onoffCaption": "dog barking at 0.353-5.99 and door slamming at 5.231-7.594", "frequencyCaption": "dog barking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3399.wav", "onoffCaption": "woman laughing at 1.997-5.284 and car horn honking at 4.179-8.02", "frequencyCaption": "woman laughing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3458.wav", "onoffCaption": "gunshot at 2.555-4.555, 6.02-8.02", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3516.wav", "onoffCaption": "thump thud at 1.147-3.918 and sneeze at 5.679-9.327", "frequencyCaption": "thump thud one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3543.wav", "onoffCaption": "car horn honking at 1.374-5.028, 5.948-8.323", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3730.wav", "onoffCaption": "duck quacking at 2.802-4.802, 5.763-7.763", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3765.wav", "onoffCaption": "cow mooing at 2.874-5.843 and cat meowing at 6.157-7.184", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3780.wav", "onoffCaption": "sheep goat bleating at 1.923-5.563, 7.028-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3847.wav", "onoffCaption": "train horn at 2.831-7.733 and car horn honking at 3.775-8.024", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_213.wav", "onoffCaption": "sheep goat bleating at 4.077-6.077 and whistling at 4.832-7.807 and woman laughing at 5.713-8.132", "frequencyCaption": "sheep goat bleating one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_246.wav", "onoffCaption": "gunshot at 1.206-3.206", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_308.wav", "onoffCaption": "thump thud at 2.127-4.898, 7.007-9.507", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_412.wav", "onoffCaption": "cat meowing at 0.117-1.129 and woman laughing at 2.319-4.435", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_463.wav", "onoffCaption": "sneeze at 1.924-3.027, 3.851-4.954 and explosion at 7.605-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_486.wav", "onoffCaption": "tapping clicking clanking at 2.505-5.945, 7.953-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_509.wav", "onoffCaption": "thump thud at 1.212-4.879, 5.485-8.191", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_578.wav", "onoffCaption": "explosion at 1.569-4.569 and door slamming at 6.423-8.852", "frequencyCaption": "explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_610.wav", "onoffCaption": "tapping clicking clanking at 1.008-4.448, 6.821-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_645.wav", "onoffCaption": "gunshot at 2.581-4.581, 5.606-7.606", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_899.wav", "onoffCaption": "door slamming at 1.574-2.965, 4.028-5.419, 6.017-7.408", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_967.wav", "onoffCaption": "explosion at 2.942-7.863", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_982.wav", "onoffCaption": "tapping clicking clanking at 3.045-6.485", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1039.wav", "onoffCaption": "gunshot at 1.37-3.37, 4.286-6.286, 7.048-9.048", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1048.wav", "onoffCaption": "woman laughing at 0.025-3.11", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1122.wav", "onoffCaption": "thump thud at 3.834-8.284", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1153.wav", "onoffCaption": "sneeze at 0.825-2.437, 4.345-5.957 and dog barking at 2.113-4.513", "frequencyCaption": "sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1320.wav", "onoffCaption": "cow mooing at 0.004-3.302", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1375.wav", "onoffCaption": "explosion at 3.847-6.375, 7.628-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1390.wav", "onoffCaption": "sheep goat bleating at 0.762-2.762, 4.683-6.683", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1505.wav", "onoffCaption": "tapping clicking clanking at 0.221-3.661, 5.651-8.267", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1574.wav", "onoffCaption": "cow mooing at 0.572-3.582, 5.412-8.311", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1638.wav", "onoffCaption": "cow mooing at 1.479-4.777, 5.348-8.33 and duck quacking at 4.231-6.231, 7.487-9.487", "frequencyCaption": "cow mooing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1688.wav", "onoffCaption": "cat meowing at 0.14-1.184 and burping belching at 4.809-6.916", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1723.wav", "onoffCaption": "cow mooing at 0.814-3.824, 5.162-8.172", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1776.wav", "onoffCaption": "train horn at 2.256-6.324 and tapping clicking clanking at 2.356-5.796, 6.811-9.098", "frequencyCaption": "train horn one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1793.wav", "onoffCaption": "sheep goat bleating at 0.474-3.77, 4.683-7.979", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1801.wav", "onoffCaption": "thump thud at 1.867-4.329, 5.915-8.377", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1895.wav", "onoffCaption": "thump thud at 3.141-6.188", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3131.wav", "onoffCaption": "sheep goat bleating at 0.574-2.574, 5.046-7.046 and burping belching at 0.963-4.523", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3259.wav", "onoffCaption": "cat meowing at 1.523-4.553", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3317.wav", "onoffCaption": "tapping clicking clanking at 2.77-6.21", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3342.wav", "onoffCaption": "thump thud at 0.556-3.603, 4.335-7.382", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3499.wav", "onoffCaption": "thump thud at 0.149-3.816", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3567.wav", "onoffCaption": "woman laughing at 1.141-3.246", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3582.wav", "onoffCaption": "train horn at 2.172-4.639, 6.039-8.506 and thump thud at 7.093-9.555", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3714.wav", "onoffCaption": "spraying at 1.752-2.356, 3.004-3.945, 4.761-7.197", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3741.wav", "onoffCaption": "thump thud at 2.691-6.609", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3812.wav", "onoffCaption": "woman laughing at 1.805-4.005, 5.087-7.287", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3863.wav", "onoffCaption": "sneeze at 0.002-1.328, 2.29-5.202 and gunshot at 7.546-9.546", "frequencyCaption": "sneeze two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3886.wav", "onoffCaption": "whistling at 0.26-6.542, 7.991-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3909.wav", "onoffCaption": "woman laughing at 1.825-3.923, 4.569-6.667, 7.773-9.871", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3978.wav", "onoffCaption": "cat meowing at 0.014-1.598, 3.728-5.312", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_13.wav", "onoffCaption": "train horn at 1.267-6.125", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_161.wav", "onoffCaption": "duck quacking at 2.251-4.251, 5.882-7.882", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_184.wav", "onoffCaption": "door slamming at 1.129-4.09, 5.351-8.312", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_209.wav", "onoffCaption": "cow mooing at 0.235-3.217, 4.196-7.178 and duck quacking at 0.314-2.314, 3.528-5.528, 6.184-8.184", "frequencyCaption": "cow mooing two times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_312.wav", "onoffCaption": "door slamming at 2.09-4.548, 5.452-7.91", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_436.wav", "onoffCaption": "sneeze at 0.157-1.84, 2.888-4.801, 5.695-6.859 and explosion at 1.858-6.452", "frequencyCaption": "sneeze three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_537.wav", "onoffCaption": "thump thud at 0.307-4.682, 5.581-9.956", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_744.wav", "onoffCaption": "car horn honking at 0.921-3.834, 6.306-9.219", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_829.wav", "onoffCaption": "thump thud at 1.369-5.036, 5.978-8.234", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_833.wav", "onoffCaption": "thump thud at 0.767-3.106, 4.114-6.453, 7.09-9.429", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_866.wav", "onoffCaption": "sheep goat bleating at 0.53-4.45", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_928.wav", "onoffCaption": "tapping clicking clanking at 1.238-4.678, 6.197-8.498", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_932.wav", "onoffCaption": "woman laughing at 1.002-4.074, 6.289-9.361", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1007.wav", "onoffCaption": "door slamming at 1.093-3.093, 4.313-6.313", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1106.wav", "onoffCaption": "car horn honking at 3.019-6.194, 7.499-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1274.wav", "onoffCaption": "duck quacking at 1.809-3.809 and whistling at 5.361-10.0", "frequencyCaption": "duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1291.wav", "onoffCaption": "train horn at 0.326-3.726, 4.95-7.162", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1451.wav", "onoffCaption": "thump thud at 0.034-3.952, 4.674-7.174 and spraying at 3.911-5.644, 6.476-8.912", "frequencyCaption": "thump thud two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1550.wav", "onoffCaption": "cow mooing at 2.579-5.589, 7.075-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1622.wav", "onoffCaption": "door slamming at 0.324-2.782, 3.662-6.12, 7.194-9.652", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1739.wav", "onoffCaption": "train horn at 0.383-10.0 and sneeze at 3.175-4.421, 5.109-6.355, 7.415-8.661", "frequencyCaption": "train horn one times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1854.wav", "onoffCaption": "spraying at 0.11-0.685, 1.718-2.622, 3.534-4.266", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1955.wav", "onoffCaption": "spraying at 2.243-2.818 and explosion at 2.78-5.78", "frequencyCaption": "spraying one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3030.wav", "onoffCaption": "door knocking at 0.001-2.848, 4.27-7.117", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3065.wav", "onoffCaption": "car horn honking at 2.783-6.624, 7.492-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3080.wav", "onoffCaption": "sheep goat bleating at 0.185-3.825", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3164.wav", "onoffCaption": "door slamming at 3.034-5.492, 6.137-7.678", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3181.wav", "onoffCaption": "thump thud at 0.103-4.553 and cat meowing at 1.125-3.102", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3216.wav", "onoffCaption": "whistling at 3.804-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3429.wav", "onoffCaption": "door knocking at 0.197-3.317, 5.04-7.55", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3433.wav", "onoffCaption": "sheep goat bleating at 1.432-3.432, 5.775-7.775", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3466.wav", "onoffCaption": "tapping clicking clanking at 0.407-3.847, 5.443-8.883", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3528.wav", "onoffCaption": "tapping clicking clanking at 3.215-6.655, 7.623-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3532.wav", "onoffCaption": "gunshot at 1.231-3.231, 4.12-6.12 and explosion at 7.291-10.0", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3640.wav", "onoffCaption": "sneeze at 0.411-2.524 and dog barking at 0.973-2.973 and spraying at 5.929-6.556, 7.556-8.183, 9.228-9.855", "frequencyCaption": "sneeze one times and dog barking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3836.wav", "onoffCaption": "cat meowing at 0.111-1.666, 2.677-4.232, 6.241-7.796", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3937.wav", "onoffCaption": "car horn honking at 0.116-3.029, 3.973-6.427, 7.604-9.604", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_46.wav", "onoffCaption": "duck quacking at 1.528-3.528", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_134.wav", "onoffCaption": "sheep goat bleating at 3.009-7.729", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_222.wav", "onoffCaption": "spraying at 0.061-0.569, 1.14-2.081", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_339.wav", "onoffCaption": "car horn honking at 1.248-3.248, 3.785-5.785 and spraying at 1.875-2.939 and gunshot at 3.353-5.353", "frequencyCaption": "car horn honking two times and spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_347.wav", "onoffCaption": "door knocking at 2.775-5.895, 6.592-9.712", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_407.wav", "onoffCaption": "duck quacking at 1.025-3.025, 4.357-6.357, 7.309-9.309", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_479.wav", "onoffCaption": "car horn honking at 2.446-6.033", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_562.wav", "onoffCaption": "door knocking at 2.335-4.462", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_587.wav", "onoffCaption": "duck quacking at 2.07-4.07, 5.856-7.856", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_674.wav", "onoffCaption": "burping belching at 0.475-2.573, 3.598-6.116, 6.89-9.015", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_691.wav", "onoffCaption": "explosion at 0.294-4.888, 6.62-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_711.wav", "onoffCaption": "woman laughing at 2.479-4.716, 6.449-8.729 and car horn honking at 3.55-6.063", "frequencyCaption": "woman laughing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_883.wav", "onoffCaption": "woman laughing at 1.491-3.975, 4.989-7.473", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_998.wav", "onoffCaption": "tapping clicking clanking at 0.324-3.764 and car horn honking at 6.065-8.851", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1052.wav", "onoffCaption": "woman laughing at 1.457-8.902", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1149.wav", "onoffCaption": "car horn honking at 0.002-4.909, 5.614-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1221.wav", "onoffCaption": "sneeze at 0.416-4.916 and cat meowing at 6.14-8.328", "frequencyCaption": "sneeze one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1344.wav", "onoffCaption": "cow mooing at 0.37-3.352 and train horn at 6.332-9.866", "frequencyCaption": "cow mooing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1404.wav", "onoffCaption": "cow mooing at 0.174-3.156, 3.958-6.94, 7.572-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_1609.wav", "onoffCaption": "tapping clicking clanking at 3.452-6.892", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1677.wav", "onoffCaption": "tapping clicking clanking at 0.138-3.578 and gunshot at 1.678-3.678, 4.883-7.389", "frequencyCaption": "tapping clicking clanking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1692.wav", "onoffCaption": "sneeze at 2.938-4.169 and thump thud at 6.382-9.153", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1712.wav", "onoffCaption": "explosion at 4.023-7.023", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1789.wav", "onoffCaption": "cat meowing at 2.803-4.387, 5.33-6.516 and cow mooing at 4.765-7.775", "frequencyCaption": "cat meowing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1880.wav", "onoffCaption": "spraying at 1.107-3.543", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1900.wav", "onoffCaption": "cat meowing at 0.685-2.256 and duck quacking at 5.166-7.166", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3243.wav", "onoffCaption": "dog barking at 2.647-4.647", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3326.wav", "onoffCaption": "door knocking at 1.874-6.17 and dog barking at 2.187-4.187", "frequencyCaption": "door knocking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3358.wav", "onoffCaption": "thump thud at 0.013-2.241", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3483.wav", "onoffCaption": "sneeze at 1.519-3.98, 5.122-7.583 and door slamming at 5.232-7.595, 8.514-9.747", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3598.wav", "onoffCaption": "train horn at 3.652-7.652", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3615.wav", "onoffCaption": "cat meowing at 2.269-3.84, 5.6-7.171, 7.761-9.332", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3770.wav", "onoffCaption": "spraying at 2.216-3.463, 4.491-7.075", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3795.wav", "onoffCaption": "whistling at 0.521-9.132", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3807.wav", "onoffCaption": "door knocking at 0.858-2.948 and gunshot at 1.961-4.131", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3879.wav", "onoffCaption": "duck quacking at 0.504-2.504, 3.187-5.187", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3962.wav", "onoffCaption": "door slamming at 0.085-0.936 and dog barking at 0.218-2.218", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3987.wav", "onoffCaption": "woman laughing at 0.024-2.616, 4.505-6.534 and cow mooing at 1.114-4.124, 5.734-8.744 and explosion at 1.707-4.575", "frequencyCaption": "woman laughing two times and cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_38.wav", "onoffCaption": "whistling at 2.329-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_77.wav", "onoffCaption": "whistling at 0.632-8.382", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_88.wav", "onoffCaption": "explosion at 0.507-5.507, 6.588-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_105.wav", "onoffCaption": "sheep goat bleating at 2.711-4.711, 5.831-7.831", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_277.wav", "onoffCaption": "sheep goat bleating at 3.395-5.395, 6.002-8.002 and door knocking at 4.127-8.277", "frequencyCaption": "sheep goat bleating two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_292.wav", "onoffCaption": "woman laughing at 2.321-4.437, 6.796-8.912", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_389.wav", "onoffCaption": "woman laughing at 2.815-5.234", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_448.wav", "onoffCaption": "burping belching at 2.701-5.701, 6.788-9.788", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_452.wav", "onoffCaption": "cow mooing at 1.714-5.012, 7.457-9.998 and train horn at 1.894-4.374", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_549.wav", "onoffCaption": "train horn at 2.855-7.932 and dog barking at 5.735-7.735", "frequencyCaption": "train horn one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_553.wav", "onoffCaption": "door knocking at 0.301-2.489, 3.653-5.841", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_621.wav", "onoffCaption": "thump thud at 3.315-6.086, 7.481-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_818.wav", "onoffCaption": "door knocking at 0.557-2.684 and spraying at 5.889-7.064, 8.823-9.998", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_903.wav", "onoffCaption": "car horn honking at 0.806-4.393", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_956.wav", "onoffCaption": "door slamming at 0.088-1.066, 1.694-2.532, 3.141-5.141", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1063.wav", "onoffCaption": "woman laughing at 2.638-5.733, 6.69-8.916", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1079.wav", "onoffCaption": "cat meowing at 0.552-2.688, 5.012-6.159, 7.267-10.0", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1137.wav", "onoffCaption": "whistling at 2.697-7.181, 7.71-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1162.wav", "onoffCaption": "door slamming at 0.678-3.678, 4.346-5.495 and train horn at 7.049-10.0", "frequencyCaption": "door slamming two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1178.wav", "onoffCaption": "thump thud at 3.23-7.148", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1187.wav", "onoffCaption": "dog barking at 1.943-3.943, 4.593-6.593, 7.488-9.497", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1311.wav", "onoffCaption": "spraying at 0.006-1.768, 3.268-5.03 and door knocking at 0.879-3.342, 4.739-7.202 and door slamming at 0.879-2.403, 2.917-4.441", "frequencyCaption": "spraying two times and door knocking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1435.wav", "onoffCaption": "burping belching at 1.147-3.473 and cow mooing at 1.605-4.574", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1534.wav", "onoffCaption": "tapping clicking clanking at 2.592-6.032, 7.042-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1561.wav", "onoffCaption": "sheep goat bleating at 0.655-2.655, 4.69-6.69", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1584.wav", "onoffCaption": "woman laughing at 0.521-3.575 and tapping clicking clanking at 2.057-5.497, 7.773-10.0", "frequencyCaption": "woman laughing one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1747.wav", "onoffCaption": "door knocking at 0.364-3.201", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1830.wav", "onoffCaption": "whistling at 1.244-4.119", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1865.wav", "onoffCaption": "door slamming at 3.362-4.841", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3001.wav", "onoffCaption": "door slamming at 0.226-1.399, 2.054-4.054", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3100.wav", "onoffCaption": "woman laughing at 0.453-2.937 and dog barking at 1.239-3.239 and cat meowing at 5.498-7.438", "frequencyCaption": "woman laughing one times and dog barking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3155.wav", "onoffCaption": "tapping clicking clanking at 1.428-4.868, 5.411-8.851 and dog barking at 2.667-4.667, 5.798-7.798", "frequencyCaption": "tapping clicking clanking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3268.wav", "onoffCaption": "door slamming at 0.163-2.944, 4.553-5.531", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3373.wav", "onoffCaption": "door knocking at 0.756-5.458, 6.45-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3396.wav", "onoffCaption": "spraying at 0.33-2.79, 3.695-6.155 and whistling at 0.506-8.345", "frequencyCaption": "spraying two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3418.wav", "onoffCaption": "tapping clicking clanking at 2.902-6.342, 7.418-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3503.wav", "onoffCaption": "door knocking at 0.979-3.603, 5.118-7.245", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3556.wav", "onoffCaption": "whistling at 1.284-6.459, 7.504-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3725.wav", "onoffCaption": "sheep goat bleating at 0.533-2.533, 4.782-7.45 and spraying at 3.404-4.308", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3848.wav", "onoffCaption": "door slamming at 0.005-2.133 and duck quacking at 0.54-2.54, 3.892-5.892", "frequencyCaption": "door slamming one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3852.wav", "onoffCaption": "spraying at 0.81-1.31 and duck quacking at 3.213-5.213", "frequencyCaption": "spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3949.wav", "onoffCaption": "door slamming at 2.945-5.726, 6.358-9.139", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3953.wav", "onoffCaption": "door slamming at 1.582-3.778 and gunshot at 1.671-3.671, 4.86-6.86, 7.927-9.927 and explosion at 1.95-5.077, 5.781-8.781", "frequencyCaption": "door slamming one times and gunshot three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_22.wav", "onoffCaption": "gunshot at 0.208-2.208, 3.513-5.513", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_92.wav", "onoffCaption": "train horn at 1.487-4.287 and car horn honking at 6.554-10.0", "frequencyCaption": "train horn one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_121.wav", "onoffCaption": "dog barking at 0.074-2.074, 4.268-6.268", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_150.wav", "onoffCaption": "cat meowing at 3.961-5.236, 7.028-8.612 and sheep goat bleating at 5.34-7.34", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_238.wav", "onoffCaption": "cat meowing at 1.32-3.26, 5.223-6.489", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_288.wav", "onoffCaption": "whistling at 2.216-6.7", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_323.wav", "onoffCaption": "woman laughing at 2.454-5.842", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_376.wav", "onoffCaption": "tapping clicking clanking at 0.29-3.73, 6.02-9.46", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_393.wav", "onoffCaption": "burping belching at 2.451-4.777, 5.569-7.604 and cow mooing at 2.659-5.957, 6.533-9.831", "frequencyCaption": "burping belching two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_506.wav", "onoffCaption": "train horn at 0.082-2.756", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_577.wav", "onoffCaption": "spraying at 3.475-4.05, 6.365-7.429", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_720.wav", "onoffCaption": "train horn at 0.504-3.904, 6.135-8.272 and thump thud at 3.277-6.048, 6.725-9.496", "frequencyCaption": "train horn two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_775.wav", "onoffCaption": "cat meowing at 2.974-3.985, 4.638-5.649, 6.693-7.704", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_790.wav", "onoffCaption": "explosion at 0.948-2.95, 3.893-6.893", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_802.wav", "onoffCaption": "woman laughing at 0.011-4.063", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_857.wav", "onoffCaption": "woman laughing at 1.419-4.214", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_896.wav", "onoffCaption": "door slamming at 2.591-5.565 and cat meowing at 8.569-9.879", "frequencyCaption": "door slamming one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_919.wav", "onoffCaption": "explosion at 0.047-3.047 and train horn at 0.453-3.853, 4.62-6.757 and sheep goat bleating at 5.955-7.955", "frequencyCaption": "explosion one times and train horn two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1036.wav", "onoffCaption": "thump thud at 0.983-5.433, 7.61-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1086.wav", "onoffCaption": "door slamming at 0.124-1.273 and explosion at 0.735-2.823, 4.138-6.226, 6.896-8.984", "frequencyCaption": "door slamming one times and explosion three times"}
+{"filepath": "data/multi_event_train/syn_1210.wav", "onoffCaption": "spraying at 0.162-1.246, 1.853-2.937, 3.553-4.637 and burping belching at 0.281-7.449", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1245.wav", "onoffCaption": "whistling at 2.827-5.056, 6.187-8.416", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1411.wav", "onoffCaption": "sneeze at 2.848-5.087", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1460.wav", "onoffCaption": "thump thud at 0.766-3.537, 4.514-6.887", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1485.wav", "onoffCaption": "train horn at 0.635-3.115, 4.967-7.447 and whistling at 2.075-4.95 and dog barking at 2.827-4.827, 5.893-7.893", "frequencyCaption": "train horn two times and whistling one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1613.wav", "onoffCaption": "thump thud at 1.417-3.917, 6.277-9.048", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1646.wav", "onoffCaption": "burping belching at 0.959-4.239, 6.326-8.357", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1708.wav", "onoffCaption": "gunshot at 3.501-5.501, 6.563-8.563", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1931.wav", "onoffCaption": "tapping clicking clanking at 2.988-6.428 and whistling at 3.277-5.286, 5.868-8.31", "frequencyCaption": "tapping clicking clanking one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1964.wav", "onoffCaption": "cat meowing at 2.184-3.916 and spraying at 5.222-6.126, 7.621-8.525", "frequencyCaption": "cat meowing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1981.wav", "onoffCaption": "sneeze at 2.341-4.955 and spraying at 3.88-4.784, 6.693-9.277", "frequencyCaption": "sneeze one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3054.wav", "onoffCaption": "whistling at 3.252-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3227.wav", "onoffCaption": "burping belching at 1.558-5.894", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3272.wav", "onoffCaption": "burping belching at 0.208-2.534, 3.348-5.674", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3297.wav", "onoffCaption": "train horn at 0.472-3.792 and duck quacking at 5.278-7.278, 7.887-9.887", "frequencyCaption": "train horn one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3369.wav", "onoffCaption": "spraying at 0.138-0.989, 2.815-5.251 and cat meowing at 0.541-5.541, 6.668-8.204 and car horn honking at 5.496-7.961", "frequencyCaption": "spraying two times and cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3402.wav", "onoffCaption": "dog barking at 0.304-2.304, 3.917-5.917 and spraying at 3.578-4.228, 4.758-5.408 and duck quacking at 4.149-6.149", "frequencyCaption": "dog barking two times and spraying two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3457.wav", "onoffCaption": "car horn honking at 0.049-4.449 and door slamming at 2.436-4.436, 6.293-7.466", "frequencyCaption": "car horn honking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3519.wav", "onoffCaption": "woman laughing at 0.968-4.356, 6.084-8.682", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3624.wav", "onoffCaption": "door knocking at 1.229-4.845 and cat meowing at 1.724-2.999 and spraying at 7.504-8.679", "frequencyCaption": "door knocking one times and cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3671.wav", "onoffCaption": "door knocking at 3.586-5.898, 6.835-9.147", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3694.wav", "onoffCaption": "dog barking at 1.084-3.084", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3906.wav", "onoffCaption": "gunshot at 1.193-3.193, 5.106-7.106", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3977.wav", "onoffCaption": "dog barking at 0.384-2.822 and train horn at 5.374-7.854", "frequencyCaption": "dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_53.wav", "onoffCaption": "gunshot at 1.059-3.059", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_174.wav", "onoffCaption": "car horn honking at 2.435-5.93, 6.754-9.219", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_191.wav", "onoffCaption": "dog barking at 1.037-3.037, 4.049-6.049, 7.259-9.259", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_206.wav", "onoffCaption": "door slamming at 0.219-1.368, 2.751-3.9 and sheep goat bleating at 2.449-7.329", "frequencyCaption": "door slamming two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_249.wav", "onoffCaption": "spraying at 1.489-3.949, 5.232-5.74, 6.275-7.025 and explosion at 3.381-8.302", "frequencyCaption": "spraying three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_307.wav", "onoffCaption": "gunshot at 0.873-2.873, 4.967-6.967 and duck quacking at 4.095-6.095, 7.153-9.153", "frequencyCaption": "gunshot two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_352.wav", "onoffCaption": "door knocking at 0.553-2.774, 3.711-5.932", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_439.wav", "onoffCaption": "car horn honking at 3.541-7.195", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_489.wav", "onoffCaption": "burping belching at 0.236-3.236, 3.906-6.906, 7.783-10.0 and sheep goat bleating at 0.829-4.149", "frequencyCaption": "burping belching three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_522.wav", "onoffCaption": "sneeze at 1.585-3.824, 4.766-7.005, 7.533-9.772", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_592.wav", "onoffCaption": "car horn honking at 3.597-6.523", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_650.wav", "onoffCaption": "burping belching at 0.679-3.679, 4.496-7.496", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_704.wav", "onoffCaption": "spraying at 0.353-0.853, 2.125-2.625, 4.729-5.229 and train horn at 2.949-7.149", "frequencyCaption": "spraying three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_751.wav", "onoffCaption": "dog barking at 0.236-2.236, 2.934-4.934, 6.462-8.462 and cat meowing at 0.323-1.598", "frequencyCaption": "dog barking three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_826.wav", "onoffCaption": "cat meowing at 1.916-3.46 and tapping clicking clanking at 5.548-8.988", "frequencyCaption": "cat meowing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_873.wav", "onoffCaption": "burping belching at 0.036-3.905, 5.109-7.574 and whistling at 0.109-2.118, 2.672-4.681", "frequencyCaption": "burping belching two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_968.wav", "onoffCaption": "burping belching at 2.876-5.876, 6.506-8.876 and door slamming at 8.343-9.343", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1012.wav", "onoffCaption": "cow mooing at 2.858-7.838", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1047.wav", "onoffCaption": "door knocking at 0.765-3.495 and cat meowing at 6.322-8.217", "frequencyCaption": "door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1109.wav", "onoffCaption": "door knocking at 0.19-2.317, 3.942-6.775", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1234.wav", "onoffCaption": "woman laughing at 0.092-2.792, 4.844-7.544", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1261.wav", "onoffCaption": "dog barking at 0.85-2.85, 3.551-5.551 and door knocking at 1.585-4.694, 5.258-8.367 and spraying at 2.752-4.88", "frequencyCaption": "dog barking two times and door knocking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1284.wav", "onoffCaption": "dog barking at 0.114-2.114, 3.412-5.412 and cat meowing at 2.154-4.342 and gunshot at 2.395-4.395, 6.662-8.662", "frequencyCaption": "dog barking two times and cat meowing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1360.wav", "onoffCaption": "door knocking at 2.493-6.255, 7.002-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1444.wav", "onoffCaption": "gunshot at 0.506-2.506 and burping belching at 0.555-5.555", "frequencyCaption": "gunshot one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1637.wav", "onoffCaption": "tapping clicking clanking at 4.251-7.691", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1662.wav", "onoffCaption": "tapping clicking clanking at 0.128-3.568, 4.812-7.009, 7.776-9.841 and door slamming at 1.132-3.59, 5.164-7.622", "frequencyCaption": "tapping clicking clanking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1687.wav", "onoffCaption": "woman laughing at 0.062-7.507", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1736.wav", "onoffCaption": "train horn at 0.102-3.502, 4.222-6.522", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1779.wav", "onoffCaption": "cat meowing at 0.502-3.532, 5.329-8.359", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1915.wav", "onoffCaption": "train horn at 0.937-3.577, 5.909-8.278", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1940.wav", "onoffCaption": "cat meowing at 3.102-4.251 and sheep goat bleating at 7.5-9.5", "frequencyCaption": "cat meowing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3025.wav", "onoffCaption": "train horn at 0.202-3.002, 5.381-8.181", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3070.wav", "onoffCaption": "spraying at 2.319-2.946, 5.07-5.697", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3095.wav", "onoffCaption": "door knocking at 1.645-4.182, 6.269-9.116", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3203.wav", "onoffCaption": "train horn at 2.208-5.978 and door slamming at 7.84-9.231", "frequencyCaption": "train horn one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3219.wav", "onoffCaption": "gunshot at 2.327-4.327, 6.446-8.446", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3256.wav", "onoffCaption": "cat meowing at 2.573-4.144, 6.409-7.98 and sneeze at 3.459-4.713", "frequencyCaption": "cat meowing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3318.wav", "onoffCaption": "car horn honking at 0.37-4.77 and sneeze at 5.499-6.663, 7.188-8.352", "frequencyCaption": "car horn honking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3426.wav", "onoffCaption": "spraying at 2.304-4.888", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3473.wav", "onoffCaption": "train horn at 3.563-9.623", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3496.wav", "onoffCaption": "door knocking at 1.616-6.318", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3568.wav", "onoffCaption": "spraying at 2.545-3.545, 4.18-5.18, 6.037-7.037", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3600.wav", "onoffCaption": "sneeze at 1.585-3.498, 4.742-7.001", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3655.wav", "onoffCaption": "gunshot at 1.525-3.655, 4.353-6.353", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3839.wav", "onoffCaption": "sheep goat bleating at 1.587-4.883, 6.121-8.121 and door knocking at 2.008-4.383, 5.019-7.248 and duck quacking at 5.02-7.02", "frequencyCaption": "sheep goat bleating two times and door knocking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3889.wav", "onoffCaption": "sheep goat bleating at 0.727-4.647, 5.179-7.179", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3922.wav", "onoffCaption": "tapping clicking clanking at 2.51-5.95, 6.743-9.167", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3992.wav", "onoffCaption": "car horn honking at 0.694-4.943, 6.084-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_49.wav", "onoffCaption": "cat meowing at 2.928-4.545 and tapping clicking clanking at 6.788-10.0", "frequencyCaption": "cat meowing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_169.wav", "onoffCaption": "car horn honking at 3.743-8.65", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_253.wav", "onoffCaption": "car horn honking at 0.476-4.063, 6.176-9.023 and door slamming at 3.689-5.885", "frequencyCaption": "car horn honking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_348.wav", "onoffCaption": "duck quacking at 1.696-3.696 and car horn honking at 6.827-10.0", "frequencyCaption": "duck quacking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_423.wav", "onoffCaption": "sneeze at 0.039-1.984, 3.045-5.358 and duck quacking at 0.331-2.331, 4.696-6.696 and cat meowing at 4.868-6.758, 7.943-9.833", "frequencyCaption": "sneeze two times and duck quacking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_424.wav", "onoffCaption": "dog barking at 2.088-4.088, 5.769-7.769", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_476.wav", "onoffCaption": "cow mooing at 2.637-5.619, 6.636-9.618", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_493.wav", "onoffCaption": "train horn at 2.842-6.082 and thump thud at 4.328-6.79", "frequencyCaption": "train horn one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_538.wav", "onoffCaption": "door slamming at 3.054-4.357 and sneeze at 7.84-9.785", "frequencyCaption": "door slamming one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_588.wav", "onoffCaption": "sneeze at 0.202-3.85, 5.004-6.096", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_605.wav", "onoffCaption": "woman laughing at 3.674-6.26", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_869.wav", "onoffCaption": "gunshot at 3.09-5.09, 7.337-9.337", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_920.wav", "onoffCaption": "tapping clicking clanking at 0.304-3.744, 5.854-9.294", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_927.wav", "onoffCaption": "whistling at 2.837-7.321", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_972.wav", "onoffCaption": "whistling at 0.085-9.75 and cow mooing at 0.316-4.745, 6.432-8.861 and spraying at 0.327-1.059", "frequencyCaption": "whistling one times and cow mooing two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_997.wav", "onoffCaption": "cat meowing at 0.354-1.456, 2.312-3.414 and cow mooing at 1.639-6.068, 7.489-10.0 and duck quacking at 2.356-4.356", "frequencyCaption": "cat meowing two times and cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1008.wav", "onoffCaption": "gunshot at 2.084-4.324 and sneeze at 2.735-4.974", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1113.wav", "onoffCaption": "explosion at 3.046-7.046 and thump thud at 4.891-8.558", "frequencyCaption": "explosion one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1114.wav", "onoffCaption": "car horn honking at 2.766-6.261, 7.379-9.379", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1146.wav", "onoffCaption": "sneeze at 3.579-5.896, 6.434-9.281", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1335.wav", "onoffCaption": "explosion at 1.32-4.873 and sheep goat bleating at 1.995-3.995", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1385.wav", "onoffCaption": "cat meowing at 0.38-1.924, 3.984-5.528, 6.641-8.185", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1459.wav", "onoffCaption": "whistling at 0.017-8.361", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1510.wav", "onoffCaption": "tapping clicking clanking at 1.596-5.036", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1542.wav", "onoffCaption": "car horn honking at 2.759-7.159, 7.739-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1545.wav", "onoffCaption": "thump thud at 1.587-4.358 and car horn honking at 7.312-9.777", "frequencyCaption": "thump thud one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1678.wav", "onoffCaption": "explosion at 0.267-3.82, 5.934-8.806", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1763.wav", "onoffCaption": "whistling at 2.661-5.536, 6.975-9.85", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1786.wav", "onoffCaption": "cow mooing at 0.537-3.835 and thump thud at 5.986-8.757", "frequencyCaption": "cow mooing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1814.wav", "onoffCaption": "woman laughing at 0.504-2.741, 3.614-5.818, 7.061-9.416", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_1841.wav", "onoffCaption": "burping belching at 0.979-5.002", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1846.wav", "onoffCaption": "door knocking at 3.081-5.171, 6.785-8.875", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3088.wav", "onoffCaption": "woman laughing at 1.437-8.882", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3124.wav", "onoffCaption": "sheep goat bleating at 1.512-3.512, 4.025-6.025", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3171.wav", "onoffCaption": "whistling at 0.029-5.427, 6.409-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3176.wav", "onoffCaption": "woman laughing at 2.821-5.104 and spraying at 5.22-6.301, 7.497-8.578", "frequencyCaption": "woman laughing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3193.wav", "onoffCaption": "thump thud at 0.666-3.166 and sneeze at 1.27-5.326, 6.655-8.041", "frequencyCaption": "thump thud one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3194.wav", "onoffCaption": "woman laughing at 2.365-5.437, 6.472-8.84 and spraying at 3.387-4.468, 6.773-7.854", "frequencyCaption": "woman laughing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3302.wav", "onoffCaption": "burping belching at 2.505-4.612, 5.841-7.872", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3357.wav", "onoffCaption": "cow mooing at 0.093-3.075 and explosion at 7.184-10.0", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3469.wav", "onoffCaption": "sheep goat bleating at 3.297-5.297, 7.545-9.545", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3520.wav", "onoffCaption": "cow mooing at 0.447-3.745", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3527.wav", "onoffCaption": "cow mooing at 0.083-3.381", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3572.wav", "onoffCaption": "door knocking at 2.958-7.397 and door slamming at 4.054-4.994", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3597.wav", "onoffCaption": "door slamming at 0.711-2.839, 4.986-6.377", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3701.wav", "onoffCaption": "explosion at 1.073-3.393, 4.548-6.868, 7.945-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_3754.wav", "onoffCaption": "car horn honking at 0.048-4.448, 5.584-7.585 and spraying at 2.765-3.849", "frequencyCaption": "car horn honking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3823.wav", "onoffCaption": "door slamming at 2.956-5.917 and dog barking at 4.289-6.289", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3824.wav", "onoffCaption": "explosion at 0.787-3.787, 4.651-7.651 and sheep goat bleating at 3.192-5.192, 5.983-8.134", "frequencyCaption": "explosion two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3876.wav", "onoffCaption": "burping belching at 0.142-2.507, 3.011-5.376", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3893.wav", "onoffCaption": "duck quacking at 0.014-2.014, 3.379-5.379, 7.391-9.391", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3938.wav", "onoffCaption": "spraying at 1.836-4.228, 5.119-6.183, 6.916-9.044", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3988.wav", "onoffCaption": "door slamming at 2.053-4.416 and burping belching at 7.773-10.0", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_201.wav", "onoffCaption": "duck quacking at 0.022-2.022, 3.195-5.195, 5.934-7.934 and door slamming at 0.033-1.15", "frequencyCaption": "duck quacking three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_471.wav", "onoffCaption": "sneeze at 0.419-1.583 and explosion at 3.872-5.965, 7.719-10.0", "frequencyCaption": "sneeze one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_494.wav", "onoffCaption": "cat meowing at 1.089-2.644, 5.032-6.587 and thump thud at 4.146-6.485 and sheep goat bleating at 5.667-7.667", "frequencyCaption": "cat meowing two times and thump thud one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_657.wav", "onoffCaption": "cow mooing at 0.315-3.613, 5.99-8.959", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_719.wav", "onoffCaption": "cat meowing at 0.265-1.531, 3.646-5.378", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_975.wav", "onoffCaption": "car horn honking at 0.708-5.615", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_990.wav", "onoffCaption": "spraying at 0.108-0.84, 1.923-3.685, 4.653-5.9", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1141.wav", "onoffCaption": "sneeze at 0.18-3.388, 4.291-5.585", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1229.wav", "onoffCaption": "dog barking at 2.048-5.368, 7.558-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1299.wav", "onoffCaption": "cat meowing at 0.318-4.678, 5.599-7.154 and burping belching at 0.472-4.472", "frequencyCaption": "cat meowing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1332.wav", "onoffCaption": "dog barking at 0.323-2.323, 3.124-5.124, 7.606-9.606", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1367.wav", "onoffCaption": "whistling at 1.395-3.624 and duck quacking at 2.012-4.012 and woman laughing at 6.398-8.753", "frequencyCaption": "whistling one times and duck quacking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1382.wav", "onoffCaption": "door knocking at 1.252-3.627, 4.328-6.703, 7.558-9.933", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1517.wav", "onoffCaption": "explosion at 0.423-3.423 and spraying at 1.514-2.141, 4.439-5.008", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1731.wav", "onoffCaption": "thump thud at 0.204-4.654 and door slamming at 0.552-1.691", "frequencyCaption": "thump thud one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1764.wav", "onoffCaption": "door knocking at 2.123-5.873", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1781.wav", "onoffCaption": "dog barking at 1.246-3.246, 5.194-7.194", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1813.wav", "onoffCaption": "spraying at 0.171-2.631, 3.957-4.861, 6.029-8.421", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1908.wav", "onoffCaption": "tapping clicking clanking at 1.365-4.805", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3038.wav", "onoffCaption": "cow mooing at 1.901-4.87", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3123.wav", "onoffCaption": "spraying at 1.022-3.414 and burping belching at 5.862-9.862", "frequencyCaption": "spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3305.wav", "onoffCaption": "cow mooing at 0.655-5.084, 5.786-8.564", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3350.wav", "onoffCaption": "sheep goat bleating at 0.712-2.712, 3.787-6.394 and duck quacking at 6.274-8.274", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3575.wav", "onoffCaption": "tapping clicking clanking at 0.812-4.252, 5.887-8.473", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3590.wav", "onoffCaption": "whistling at 0.635-6.135, 7.948-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3648.wav", "onoffCaption": "spraying at 3.426-4.076 and sheep goat bleating at 7.566-9.566", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3706.wav", "onoffCaption": "whistling at 0.945-8.6", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3753.wav", "onoffCaption": "tapping clicking clanking at 3.307-6.747", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3871.wav", "onoffCaption": "explosion at 3.077-5.83, 6.473-9.226", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3894.wav", "onoffCaption": "sheep goat bleating at 0.201-2.201, 3.431-5.431", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_54.wav", "onoffCaption": "door slamming at 3.055-4.446, 5.885-8.248", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_126.wav", "onoffCaption": "dog barking at 0.327-2.327", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_254.wav", "onoffCaption": "gunshot at 0.466-2.466, 3.589-5.589, 7.552-9.552", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_355.wav", "onoffCaption": "duck quacking at 2.57-4.57, 5.358-7.358 and sheep goat bleating at 7.642-9.642", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_570.wav", "onoffCaption": "train horn at 2.982-5.782, 6.951-9.751", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_595.wav", "onoffCaption": "sneeze at 3.065-7.594", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_602.wav", "onoffCaption": "duck quacking at 0.964-2.964, 5.143-7.143", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_618.wav", "onoffCaption": "door slamming at 2.714-3.853, 4.445-5.584, 6.529-7.668", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_703.wav", "onoffCaption": "tapping clicking clanking at 1.009-4.449, 6.831-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_874.wav", "onoffCaption": "thump thud at 0.65-5.1, 6.105-8.333", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_891.wav", "onoffCaption": "thump thud at 2.63-5.13 and cat meowing at 3.305-5.037, 5.676-7.408", "frequencyCaption": "thump thud one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1040.wav", "onoffCaption": "sheep goat bleating at 1.747-3.747", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1233.wav", "onoffCaption": "gunshot at 3.194-5.194", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1328.wav", "onoffCaption": "gunshot at 0.77-2.77, 4.869-7.375, 7.885-9.885", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1416.wav", "onoffCaption": "door slamming at 0.159-0.84 and whistling at 0.385-4.869", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1665.wav", "onoffCaption": "duck quacking at 1.743-3.743, 4.623-6.623, 7.499-9.499", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1680.wav", "onoffCaption": "train horn at 0.177-2.577, 3.945-6.558 and explosion at 0.391-3.263, 3.995-6.867, 7.672-10.0 and cat meowing at 1.708-6.068", "frequencyCaption": "train horn two times and explosion three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1809.wav", "onoffCaption": "tapping clicking clanking at 0.425-3.865, 4.963-7.504", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1912.wav", "onoffCaption": "tapping clicking clanking at 2.261-5.701", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3022.wav", "onoffCaption": "sheep goat bleating at 0.296-2.296 and train horn at 4.943-8.423", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3139.wav", "onoffCaption": "sheep goat bleating at 2.639-4.639", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3204.wav", "onoffCaption": "duck quacking at 0.444-2.444 and thump thud at 5.921-8.968", "frequencyCaption": "duck quacking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3251.wav", "onoffCaption": "tapping clicking clanking at 0.42-3.86 and spraying at 2.566-3.141", "frequencyCaption": "tapping clicking clanking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3474.wav", "onoffCaption": "tapping clicking clanking at 0.664-4.104, 6.255-8.849", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3491.wav", "onoffCaption": "cat meowing at 0.031-1.391 and door knocking at 4.522-9.224", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3607.wav", "onoffCaption": "thump thud at 0.385-4.303, 5.233-7.577 and duck quacking at 5.528-7.528", "frequencyCaption": "thump thud two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3652.wav", "onoffCaption": "gunshot at 0.653-2.653", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3749.wav", "onoffCaption": "explosion at 0.571-5.165", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3970.wav", "onoffCaption": "dog barking at 3.47-5.47, 6.969-8.969", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3995.wav", "onoffCaption": "dog barking at 0.185-2.185, 2.955-4.955", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_70.wav", "onoffCaption": "dog barking at 1.993-3.993, 4.548-6.548", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_95.wav", "onoffCaption": "thump thud at 0.47-3.517, 4.762-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_102.wav", "onoffCaption": "door knocking at 2.705-5.825, 7.509-9.76", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_173.wav", "onoffCaption": "tapping clicking clanking at 0.086-3.526 and whistling at 0.707-9.805 and explosion at 2.13-4.137, 5.322-7.329", "frequencyCaption": "tapping clicking clanking one times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_196.wav", "onoffCaption": "explosion at 0.777-2.779", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_300.wav", "onoffCaption": "thump thud at 1.121-5.571", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_371.wav", "onoffCaption": "duck quacking at 1.22-3.22, 5.626-7.626", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_394.wav", "onoffCaption": "woman laughing at 2.892-5.138", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_525.wav", "onoffCaption": "explosion at 0.454-2.715, 3.764-6.025 and whistling at 1.919-7.094", "frequencyCaption": "explosion two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_554.wav", "onoffCaption": "cat meowing at 1.653-2.928, 4.67-5.945", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_669.wav", "onoffCaption": "train horn at 2.326-6.326, 7.15-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_727.wav", "onoffCaption": "whistling at 0.48-10.0 and door knocking at 1.258-6.091", "frequencyCaption": "whistling one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_756.wav", "onoffCaption": "duck quacking at 2.343-4.343, 4.875-6.875, 7.865-9.865", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_772.wav", "onoffCaption": "explosion at 2.289-5.161, 5.982-8.854", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_797.wav", "onoffCaption": "sheep goat bleating at 1.447-3.447, 5.844-7.844 and dog barking at 2.75-4.75", "frequencyCaption": "sheep goat bleating two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_821.wav", "onoffCaption": "gunshot at 0.209-2.209, 3.627-5.627 and tapping clicking clanking at 0.683-4.123", "frequencyCaption": "gunshot two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_850.wav", "onoffCaption": "train horn at 0.525-4.295 and door knocking at 0.696-3.233 and cat meowing at 2.715-5.619, 6.598-8.169", "frequencyCaption": "train horn one times and door knocking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1015.wav", "onoffCaption": "car horn honking at 0.191-3.117 and spraying at 2.06-2.568, 3.885-4.393, 5.536-6.044 and door slamming at 4.087-5.027, 6.379-7.319", "frequencyCaption": "car horn honking one times and spraying three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1064.wav", "onoffCaption": "door slamming at 1.414-4.293", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1081.wav", "onoffCaption": "door slamming at 1.436-4.153, 4.833-5.852", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1217.wav", "onoffCaption": "explosion at 0.391-3.391 and dog barking at 2.243-4.243", "frequencyCaption": "explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1242.wav", "onoffCaption": "spraying at 0.567-1.067, 2.157-2.938, 4.063-5.32", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1266.wav", "onoffCaption": "woman laughing at 0.647-2.93, 5.093-7.293 and dog barking at 4.855-6.855, 7.951-9.951", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1283.wav", "onoffCaption": "whistling at 1.596-9.981", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1359.wav", "onoffCaption": "train horn at 0.245-3.645, 4.367-7.767 and woman laughing at 3.448-6.148", "frequencyCaption": "train horn two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1398.wav", "onoffCaption": "woman laughing at 0.118-4.17, 5.424-7.707", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1432.wav", "onoffCaption": "door slamming at 3.069-4.61", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1443.wav", "onoffCaption": "cow mooing at 2.99-7.97", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1529.wav", "onoffCaption": "spraying at 0.045-3.064, 4.628-5.136, 5.647-6.498 and whistling at 0.17-7.92", "frequencyCaption": "spraying three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1558.wav", "onoffCaption": "spraying at 0.1-0.675", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1614.wav", "onoffCaption": "train horn at 2.717-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1630.wav", "onoffCaption": "spraying at 3.453-3.975, 6.279-6.787, 7.637-8.264", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1641.wav", "onoffCaption": "tapping clicking clanking at 0.201-3.641, 4.988-8.428", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1936.wav", "onoffCaption": "woman laughing at 0.152-2.852, 4.099-6.799", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1947.wav", "onoffCaption": "door knocking at 3.182-5.912, 7.132-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3006.wav", "onoffCaption": "door knocking at 3.963-7.516", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3077.wav", "onoffCaption": "whistling at 1.496-5.98", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3092.wav", "onoffCaption": "thump thud at 1.644-6.019, 7.447-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3189.wav", "onoffCaption": "explosion at 0.54-5.54, 6.412-10.0 and door knocking at 2.556-5.286, 6.463-8.926", "frequencyCaption": "explosion two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3220.wav", "onoffCaption": "sneeze at 3.349-4.513, 5.349-7.752", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3275.wav", "onoffCaption": "duck quacking at 0.417-2.417, 3.876-5.876", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3290.wav", "onoffCaption": "tapping clicking clanking at 0.227-3.667, 4.679-6.849 and duck quacking at 3.852-5.852, 7.038-9.038 and explosion at 5.023-9.617", "frequencyCaption": "tapping clicking clanking two times and duck quacking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3421.wav", "onoffCaption": "cat meowing at 0.013-1.584, 2.383-4.331, 5.483-8.387 and door knocking at 0.75-5.283, 7.441-10.0", "frequencyCaption": "cat meowing three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3450.wav", "onoffCaption": "duck quacking at 2.7-4.7, 5.509-7.509", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3623.wav", "onoffCaption": "car horn honking at 3.793-6.579", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3676.wav", "onoffCaption": "burping belching at 1.812-3.843, 6.336-8.367", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3693.wav", "onoffCaption": "sheep goat bleating at 2.944-4.944", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3738.wav", "onoffCaption": "thump thud at 0.424-4.799", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3788.wav", "onoffCaption": "sheep goat bleating at 0.787-5.507, 6.694-8.694", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3925.wav", "onoffCaption": "woman laughing at 0.292-7.737", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3954.wav", "onoffCaption": "sneeze at 0.104-2.023, 2.539-4.484, 6.855-8.101", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_25.wav", "onoffCaption": "gunshot at 1.625-3.625, 5.899-8.172 and whistling at 3.782-8.957", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_157.wav", "onoffCaption": "train horn at 0.233-2.37, 3.224-5.601", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_225.wav", "onoffCaption": "tapping clicking clanking at 2.105-5.545, 7.367-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_324.wav", "onoffCaption": "spraying at 0.009-0.509, 2.341-3.425", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_400.wav", "onoffCaption": "train horn at 2.694-5.574", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_501.wav", "onoffCaption": "cow mooing at 0.234-3.532 and whistling at 1.104-6.737 and explosion at 2.284-7.284, 7.816-10.0", "frequencyCaption": "cow mooing one times and whistling one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_673.wav", "onoffCaption": "woman laughing at 0.788-3.583 and door knocking at 2.854-5.701", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_805.wav", "onoffCaption": "spraying at 1.964-4.548", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_904.wav", "onoffCaption": "car horn honking at 0.948-4.535, 6.325-8.809", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1031.wav", "onoffCaption": "explosion at 0.294-2.468", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1130.wav", "onoffCaption": "woman laughing at 1.1-3.455, 4.433-6.788 and cat meowing at 1.581-3.575", "frequencyCaption": "woman laughing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1467.wav", "onoffCaption": "door knocking at 1.655-5.405, 7.308-9.808", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1482.wav", "onoffCaption": "tapping clicking clanking at 0.246-3.686, 4.457-7.897 and door slamming at 2.901-3.739", "frequencyCaption": "tapping clicking clanking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1498.wav", "onoffCaption": "whistling at 0.046-8.057 and sneeze at 2.099-3.644, 5.951-8.197", "frequencyCaption": "whistling one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1566.wav", "onoffCaption": "train horn at 2.16-7.904", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1583.wav", "onoffCaption": "door slamming at 0.07-2.29, 3.986-6.206", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1599.wav", "onoffCaption": "whistling at 0.128-9.793", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1862.wav", "onoffCaption": "gunshot at 0.694-2.694, 3.323-5.323 and burping belching at 2.026-6.026, 6.855-10.0", "frequencyCaption": "gunshot two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1878.wav", "onoffCaption": "whistling at 0.389-8.4", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1887.wav", "onoffCaption": "cow mooing at 0.512-4.941 and thump thud at 7.964-10.0", "frequencyCaption": "cow mooing one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1963.wav", "onoffCaption": "woman laughing at 1.24-3.595 and tapping clicking clanking at 1.503-4.943", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1979.wav", "onoffCaption": "door slamming at 0.075-1.214, 3.128-5.909", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1986.wav", "onoffCaption": "door slamming at 1.025-3.806, 5.303-7.303", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3049.wav", "onoffCaption": "woman laughing at 2.442-7.481", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3053.wav", "onoffCaption": "woman laughing at 2.766-5.861, 7.24-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3148.wav", "onoffCaption": "tapping clicking clanking at 0.276-3.716", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3152.wav", "onoffCaption": "sheep goat bleating at 0.454-2.454, 3.049-5.049, 6.197-8.197", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3405.wav", "onoffCaption": "whistling at 0.099-8.11", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3504.wav", "onoffCaption": "door knocking at 0.793-5.495 and thump thud at 3.052-5.391", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3800.wav", "onoffCaption": "tapping clicking clanking at 0.979-4.419, 5.28-7.969 and explosion at 2.588-5.588 and sneeze at 7.902-9.196", "frequencyCaption": "tapping clicking clanking two times and explosion one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3901.wav", "onoffCaption": "sheep goat bleating at 0.683-2.683, 4.233-6.233", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_118.wav", "onoffCaption": "train horn at 0.018-2.818, 3.936-6.736, 7.383-10.0", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_270.wav", "onoffCaption": "duck quacking at 0.329-2.329, 3.818-5.818 and sheep goat bleating at 2.212-7.092", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_295.wav", "onoffCaption": "burping belching at 0.305-2.336, 4.218-6.249, 6.891-8.922", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_455.wav", "onoffCaption": "thump thud at 2.92-7.37 and sneeze at 4.19-5.424, 6.465-7.922", "frequencyCaption": "thump thud one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_626.wav", "onoffCaption": "dog barking at 1.493-3.493", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_696.wav", "onoffCaption": "gunshot at 1.907-3.907, 4.559-6.559, 7.45-9.45", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_768.wav", "onoffCaption": "duck quacking at 3.151-5.151, 6.78-8.78", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_951.wav", "onoffCaption": "cow mooing at 2.278-5.26, 6.421-9.403", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1165.wav", "onoffCaption": "spraying at 2.992-3.992, 4.604-5.208", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1180.wav", "onoffCaption": "duck quacking at 0.539-2.539, 3.257-5.257, 6.102-8.102", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1258.wav", "onoffCaption": "whistling at 0.723-6.223 and woman laughing at 1.377-8.111 and sheep goat bleating at 6.748-8.748", "frequencyCaption": "whistling one times and woman laughing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1316.wav", "onoffCaption": "car horn honking at 2.275-5.862", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1343.wav", "onoffCaption": "duck quacking at 1.608-3.608, 4.465-6.465, 7.576-9.576", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_1428.wav", "onoffCaption": "train horn at 0.016-2.153, 3.202-5.339, 5.864-8.001", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_1533.wav", "onoffCaption": "cow mooing at 2.782-5.764, 6.836-9.177", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1715.wav", "onoffCaption": "thump thud at 1.218-3.718, 4.628-7.128", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1740.wav", "onoffCaption": "train horn at 1.767-5.127, 6.095-9.455", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1837.wav", "onoffCaption": "sheep goat bleating at 2.059-6.779 and whistling at 2.884-8.059 and woman laughing at 4.959-7.378", "frequencyCaption": "sheep goat bleating one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3107.wav", "onoffCaption": "door slamming at 0.022-0.873 and cat meowing at 1.093-2.403 and train horn at 5.959-10.0", "frequencyCaption": "door slamming one times and cat meowing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3321.wav", "onoffCaption": "cat meowing at 3.21-4.21, 5.7-6.712", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3374.wav", "onoffCaption": "dog barking at 0.078-2.078, 3.208-5.208, 5.9-7.9", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3391.wav", "onoffCaption": "explosion at 0.269-3.141 and duck quacking at 0.455-2.455 and door knocking at 5.976-8.328", "frequencyCaption": "explosion one times and duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3551.wav", "onoffCaption": "door knocking at 2.84-7.673", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3639.wav", "onoffCaption": "woman laughing at 0.177-2.377, 3.151-5.351, 6.045-8.245", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3689.wav", "onoffCaption": "sheep goat bleating at 2.212-4.212, 5.084-7.569", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3722.wav", "onoffCaption": "cat meowing at 0.333-2.327, 3.269-4.805", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3777.wav", "onoffCaption": "duck quacking at 2.801-4.801 and spraying at 3.013-5.141, 6.062-6.663", "frequencyCaption": "duck quacking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3792.wav", "onoffCaption": "explosion at 2.566-4.659, 7.128-9.212", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3855.wav", "onoffCaption": "duck quacking at 1.625-3.625 and door knocking at 7.025-9.525", "frequencyCaption": "duck quacking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_159.wav", "onoffCaption": "sneeze at 0.134-3.209, 4.258-7.138, 7.933-9.852", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_231.wav", "onoffCaption": "duck quacking at 0.318-2.318, 2.949-4.949, 6.008-8.008", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_264.wav", "onoffCaption": "dog barking at 0.128-2.128 and gunshot at 1.2-3.2, 5.28-7.28", "frequencyCaption": "dog barking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_414.wav", "onoffCaption": "door knocking at 2.845-5.469, 6.563-9.4", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_632.wav", "onoffCaption": "explosion at 2.325-4.389, 4.942-7.006 and tapping clicking clanking at 3.383-6.823", "frequencyCaption": "explosion two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_667.wav", "onoffCaption": "thump thud at 2.149-6.599, 7.743-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_682.wav", "onoffCaption": "cow mooing at 0.843-5.272, 7.321-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_729.wav", "onoffCaption": "door knocking at 0.601-4.369, 5.79-8.142", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_799.wav", "onoffCaption": "sheep goat bleating at 0.047-2.047 and whistling at 1.195-5.679, 7.057-10.0", "frequencyCaption": "sheep goat bleating one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_910.wav", "onoffCaption": "thump thud at 0.422-3.193, 3.817-6.156, 7.264-9.455", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_1124.wav", "onoffCaption": "woman laughing at 2.461-4.829 and explosion at 2.722-4.81, 5.629-7.693", "frequencyCaption": "woman laughing one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1302.wav", "onoffCaption": "explosion at 3.167-5.428, 6.839-9.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1357.wav", "onoffCaption": "explosion at 0.493-3.246, 4.246-6.607 and door slamming at 1.075-2.554", "frequencyCaption": "explosion two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1469.wav", "onoffCaption": "duck quacking at 0.592-2.592, 3.215-5.215, 5.836-7.836 and train horn at 1.241-4.601, 5.638-8.998 and door slamming at 5.79-6.768", "frequencyCaption": "duck quacking three times and train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1572.wav", "onoffCaption": "sneeze at 0.074-1.987, 3.317-4.42, 5.2-7.661 and door knocking at 1.161-5.671, 7.944-10.0", "frequencyCaption": "sneeze three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1597.wav", "onoffCaption": "cat meowing at 0.941-2.496 and dog barking at 1.668-3.668 and explosion at 6.21-10.0", "frequencyCaption": "cat meowing one times and dog barking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1701.wav", "onoffCaption": "spraying at 2.038-2.642", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1754.wav", "onoffCaption": "car horn honking at 0.031-2.817, 4.405-6.905", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1876.wav", "onoffCaption": "woman laughing at 0.697-3.181, 3.691-5.891", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1893.wav", "onoffCaption": "door slamming at 0.078-1.217 and dog barking at 3.861-5.861, 6.925-8.925 and thump thud at 4.033-8.483", "frequencyCaption": "door slamming one times and dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1988.wav", "onoffCaption": "door knocking at 0.078-2.168, 2.716-4.806, 5.45-7.54 and cat meowing at 1.282-2.866, 3.698-5.282, 5.908-7.492", "frequencyCaption": "door knocking three times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_3146.wav", "onoffCaption": "thump thud at 0.966-3.737, 4.659-7.43 and sneeze at 1.51-2.764, 3.704-4.95, 5.513-7.729 and cow mooing at 1.812-6.792", "frequencyCaption": "thump thud two times and sneeze three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3335.wav", "onoffCaption": "whistling at 3.008-8.508", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3385.wav", "onoffCaption": "sneeze at 3.023-7.523", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3510.wav", "onoffCaption": "dog barking at 2.689-4.689, 6.283-8.283", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3678.wav", "onoffCaption": "gunshot at 2.519-4.519 and whistling at 2.939-8.114", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3763.wav", "onoffCaption": "explosion at 2.855-5.573, 6.2-8.918", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3786.wav", "onoffCaption": "door slamming at 2.563-3.854, 5.558-6.363", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3814.wav", "onoffCaption": "whistling at 1.399-9.41", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_64.wav", "onoffCaption": "dog barking at 0.582-2.582, 4.995-6.995 and spraying at 0.846-1.596 and door knocking at 4.54-8.29", "frequencyCaption": "dog barking two times and spraying one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_81.wav", "onoffCaption": "thump thud at 1.031-5.481", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_281.wav", "onoffCaption": "cow mooing at 0.963-3.973, 4.899-7.683 and train horn at 3.96-7.96", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_441.wav", "onoffCaption": "gunshot at 0.774-2.774, 3.707-5.707, 7.02-9.02", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_540.wav", "onoffCaption": "thump thud at 2.697-6.364, 7.409-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_844.wav", "onoffCaption": "tapping clicking clanking at 0.698-4.138, 5.254-8.184", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_945.wav", "onoffCaption": "burping belching at 2.96-7.296", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1070.wav", "onoffCaption": "gunshot at 0.081-2.081, 3.247-5.247, 6.513-8.513", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1095.wav", "onoffCaption": "spraying at 0.659-2.145 and gunshot at 5.491-7.491", "frequencyCaption": "spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1171.wav", "onoffCaption": "dog barking at 0.481-2.481, 3.695-5.695", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1194.wav", "onoffCaption": "sneeze at 1.52-4.134, 5.429-7.668 and door slamming at 2.583-3.886", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1219.wav", "onoffCaption": "sheep goat bleating at 2.316-4.316 and train horn at 7.392-10.0", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1426.wav", "onoffCaption": "sheep goat bleating at 2.896-4.896 and door slamming at 4.863-5.701, 7.543-9.067", "frequencyCaption": "sheep goat bleating one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1527.wav", "onoffCaption": "spraying at 0.031-1.115, 2.086-3.17 and car horn honking at 2.198-5.417", "frequencyCaption": "spraying two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1823.wav", "onoffCaption": "tapping clicking clanking at 0.895-4.335, 5.021-8.461 and cow mooing at 2.805-5.787", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1839.wav", "onoffCaption": "cat meowing at 2.895-4.205, 5.243-6.553, 7.151-8.461", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1922.wav", "onoffCaption": "burping belching at 0.283-3.789, 4.889-8.395", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1938.wav", "onoffCaption": "cat meowing at 2.138-3.15, 4.397-5.409 and sneeze at 7.522-8.625", "frequencyCaption": "cat meowing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3008.wav", "onoffCaption": "gunshot at 0.446-2.616 and burping belching at 4.722-6.753, 7.735-10.0", "frequencyCaption": "gunshot one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3012.wav", "onoffCaption": "train horn at 2.914-7.816", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3109.wav", "onoffCaption": "whistling at 0.123-8.134 and dog barking at 2.714-4.714, 6.881-8.881", "frequencyCaption": "whistling one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3113.wav", "onoffCaption": "thump thud at 0.383-2.611", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3360.wav", "onoffCaption": "dog barking at 0.244-2.244, 4.643-6.643 and sneeze at 1.545-3.228, 4.283-6.202 and woman laughing at 2.999-5.282, 7.507-9.943", "frequencyCaption": "dog barking two times and sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3444.wav", "onoffCaption": "door knocking at 0.208-3.958, 4.484-6.611, 7.771-10.0", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3545.wav", "onoffCaption": "door slamming at 2.58-4.58, 5.255-7.255 and car horn honking at 5.744-7.744", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3736.wav", "onoffCaption": "sheep goat bleating at 0.976-2.976 and gunshot at 1.201-3.201, 4.086-6.086, 7.889-9.889", "frequencyCaption": "sheep goat bleating one times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3841.wav", "onoffCaption": "thump thud at 3.493-7.868 and door knocking at 6.458-8.77", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3940.wav", "onoffCaption": "burping belching at 3.845-6.075 and thump thud at 6.247-8.586", "frequencyCaption": "burping belching one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_31.wav", "onoffCaption": "explosion at 0.114-5.035, 6.617-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_40.wav", "onoffCaption": "cow mooing at 0.732-5.712, 6.321-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_116.wav", "onoffCaption": "burping belching at 2.633-6.656, 7.887-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_143.wav", "onoffCaption": "burping belching at 1.128-3.651 and spraying at 1.561-2.561, 3.832-4.916, 5.585-7.851", "frequencyCaption": "burping belching one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_330.wav", "onoffCaption": "sneeze at 0.333-3.443, 4.453-7.563", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_365.wav", "onoffCaption": "duck quacking at 0.019-2.019 and sheep goat bleating at 0.592-2.592 and explosion at 5.364-10.0", "frequencyCaption": "duck quacking one times and sheep goat bleating one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_380.wav", "onoffCaption": "car horn honking at 2.009-6.916 and tapping clicking clanking at 2.997-6.437", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_515.wav", "onoffCaption": "cow mooing at 0.715-3.684 and thump thud at 1.196-3.424 and cat meowing at 8.242-9.253", "frequencyCaption": "cow mooing one times and thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_581.wav", "onoffCaption": "dog barking at 0.183-2.183", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_628.wav", "onoffCaption": "dog barking at 2.434-4.434", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_698.wav", "onoffCaption": "cat meowing at 1.384-4.414", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_733.wav", "onoffCaption": "duck quacking at 3.485-5.485, 7.687-9.687", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_766.wav", "onoffCaption": "cat meowing at 0.072-1.967, 2.942-4.936 and sneeze at 0.947-2.178, 4.106-6.423", "frequencyCaption": "cat meowing two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_783.wav", "onoffCaption": "explosion at 0.107-3.101, 5.265-8.259", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_811.wav", "onoffCaption": "thump thud at 0.206-2.668, 4.309-6.537", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_860.wav", "onoffCaption": "car horn honking at 1.847-5.434 and burping belching at 2.504-5.504, 6.182-9.182", "frequencyCaption": "car horn honking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_885.wav", "onoffCaption": "door knocking at 3.613-8.613", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1025.wav", "onoffCaption": "tapping clicking clanking at 0.035-3.475, 5.44-7.618 and whistling at 0.845-2.854 and cow mooing at 2.891-7.871", "frequencyCaption": "tapping clicking clanking two times and whistling one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1054.wav", "onoffCaption": "explosion at 2.388-7.388", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1203.wav", "onoffCaption": "sneeze at 0.363-2.359, 3.331-5.327, 6.245-8.241 and whistling at 1.162-5.646 and burping belching at 4.98-7.21", "frequencyCaption": "sneeze three times and whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1256.wav", "onoffCaption": "sheep goat bleating at 1.756-3.756, 6.06-8.06", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1318.wav", "onoffCaption": "dog barking at 2.585-4.585, 5.452-7.452", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1402.wav", "onoffCaption": "car horn honking at 0.149-2.614, 3.555-6.02, 6.806-9.271", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_1473.wav", "onoffCaption": "spraying at 0.374-1.106, 2.858-3.939 and burping belching at 2.746-5.746, 7.61-10.0", "frequencyCaption": "spraying two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1496.wav", "onoffCaption": "burping belching at 1.063-5.063, 6.187-10.0 and train horn at 1.338-3.738", "frequencyCaption": "burping belching two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1519.wav", "onoffCaption": "explosion at 0.451-3.578, 4.185-7.312 and car horn honking at 4.164-7.383", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1568.wav", "onoffCaption": "sneeze at 0.072-4.572 and car horn honking at 2.743-5.208", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1600.wav", "onoffCaption": "tapping clicking clanking at 2.86-6.3, 7.003-9.229 and gunshot at 3.295-5.425", "frequencyCaption": "tapping clicking clanking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1655.wav", "onoffCaption": "woman laughing at 0.265-2.684 and train horn at 0.391-4.459", "frequencyCaption": "woman laughing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1889.wav", "onoffCaption": "door knocking at 3.623-6.086, 7.931-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1906.wav", "onoffCaption": "door slamming at 2.942-3.747, 5.972-6.777", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1977.wav", "onoffCaption": "gunshot at 3.187-5.187 and thump thud at 7.862-10.0", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1992.wav", "onoffCaption": "cat meowing at 0.094-3.335", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3036.wav", "onoffCaption": "cat meowing at 2.02-3.38, 4.785-5.812 and woman laughing at 2.276-5.557, 7.167-10.0 and dog barking at 3.171-5.171, 6.364-8.364", "frequencyCaption": "cat meowing two times and woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3047.wav", "onoffCaption": "car horn honking at 0.568-3.481, 4.098-6.975", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3234.wav", "onoffCaption": "cow mooing at 0.597-5.026, 6.061-8.703", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3261.wav", "onoffCaption": "explosion at 3.738-7.578", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3284.wav", "onoffCaption": "thump thud at 0.415-4.865, 6.186-8.495", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3411.wav", "onoffCaption": "door knocking at 0.393-2.642 and whistling at 5.46-9.944", "frequencyCaption": "door knocking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3460.wav", "onoffCaption": "explosion at 4.039-6.792", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3485.wav", "onoffCaption": "whistling at 1.733-6.217 and cow mooing at 2.774-6.072, 7.142-10.0", "frequencyCaption": "whistling one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3637.wav", "onoffCaption": "sheep goat bleating at 0.468-2.468, 3.357-5.579, 6.358-8.358", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3662.wav", "onoffCaption": "spraying at 0.628-2.114, 3.145-5.537, 6.065-8.117", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3687.wav", "onoffCaption": "cat meowing at 3.218-5.108, 5.993-6.993", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3779.wav", "onoffCaption": "cow mooing at 0.863-3.832", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3915.wav", "onoffCaption": "spraying at 0.661-1.918, 2.505-4.633", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3981.wav", "onoffCaption": "dog barking at 1.467-3.467, 5.657-7.657", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_15.wav", "onoffCaption": "duck quacking at 0.518-2.518, 3.085-5.085, 6.404-8.404 and cow mooing at 2.413-6.842, 7.53-10.0 and car horn honking at 3.791-7.632", "frequencyCaption": "duck quacking three times and cow mooing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_132.wav", "onoffCaption": "whistling at 2.824-5.053, 5.615-7.844", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_167.wav", "onoffCaption": "explosion at 2.758-5.511, 7.256-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_182.wav", "onoffCaption": "door slamming at 2.185-3.664, 5.817-7.12 and car horn honking at 2.817-7.724", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_314.wav", "onoffCaption": "tapping clicking clanking at 0.093-3.533, 4.054-7.494", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_341.wav", "onoffCaption": "car horn honking at 0.634-5.146, 6.739-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_531.wav", "onoffCaption": "explosion at 0.061-5.061, 7.036-9.994", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_564.wav", "onoffCaption": "cow mooing at 3.722-6.704", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_643.wav", "onoffCaption": "spraying at 2.746-3.83, 5.719-6.803", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_659.wav", "onoffCaption": "door slamming at 1.024-1.829, 2.415-3.22, 5.384-6.189", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_717.wav", "onoffCaption": "sheep goat bleating at 1.082-3.082, 3.853-5.853 and thump thud at 7.758-10.0", "frequencyCaption": "sheep goat bleating two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_742.wav", "onoffCaption": "burping belching at 3.38-6.639", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_758.wav", "onoffCaption": "train horn at 1.569-6.646, 7.351-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_835.wav", "onoffCaption": "sheep goat bleating at 1.813-3.813, 5.405-7.405 and burping belching at 2.387-5.931", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1001.wav", "onoffCaption": "sneeze at 0.148-4.204, 5.432-9.488", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1227.wav", "onoffCaption": "tapping clicking clanking at 0.311-3.751", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1268.wav", "onoffCaption": "spraying at 0.814-1.383 and sheep goat bleating at 0.996-2.996 and train horn at 6.867-10.0", "frequencyCaption": "spraying one times and sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1272.wav", "onoffCaption": "door slamming at 1.698-4.659, 6.438-8.801 and cat meowing at 2.838-3.85", "frequencyCaption": "door slamming two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1297.wav", "onoffCaption": "door slamming at 1.376-3.376 and burping belching at 6.395-8.625", "frequencyCaption": "door slamming one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1369.wav", "onoffCaption": "woman laughing at 0.167-3.239, 4.685-7.757", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1373.wav", "onoffCaption": "explosion at 0.042-2.91 and sneeze at 1.567-3.68, 5.675-7.829 and dog barking at 6.99-8.99", "frequencyCaption": "explosion one times and sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1396.wav", "onoffCaption": "sheep goat bleating at 1.953-6.833, 7.799-9.799", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1457.wav", "onoffCaption": "spraying at 2.206-2.728 and sheep goat bleating at 2.967-4.967", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1624.wav", "onoffCaption": "train horn at 0.453-2.893", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1671.wav", "onoffCaption": "sneeze at 0.545-1.648", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1694.wav", "onoffCaption": "sneeze at 0.873-3.983", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1725.wav", "onoffCaption": "train horn at 0.363-4.133 and whistling at 0.652-6.152, 6.997-10.0", "frequencyCaption": "train horn one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_1848.wav", "onoffCaption": "sneeze at 0.755-2.049, 2.763-4.057, 4.699-5.993 and cow mooing at 0.817-3.786", "frequencyCaption": "sneeze three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1953.wav", "onoffCaption": "whistling at 0.392-9.463 and car horn honking at 2.538-4.538", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3063.wav", "onoffCaption": "tapping clicking clanking at 1.047-4.487 and sheep goat bleating at 2.67-4.67", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3086.wav", "onoffCaption": "train horn at 1.792-4.259, 5.145-7.569", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3178.wav", "onoffCaption": "burping belching at 0.665-4.209, 6.437-9.981", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3210.wav", "onoffCaption": "tapping clicking clanking at 2.508-5.948, 7.415-9.678", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3245.wav", "onoffCaption": "tapping clicking clanking at 1.293-4.733, 5.996-9.436 and train horn at 3.296-5.951, 6.594-9.249", "frequencyCaption": "tapping clicking clanking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3311.wav", "onoffCaption": "sneeze at 1.807-3.193, 4.448-5.834, 6.424-7.81", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3435.wav", "onoffCaption": "thump thud at 2.366-4.828, 6.309-8.771", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3613.wav", "onoffCaption": "tapping clicking clanking at 0.319-3.759 and dog barking at 5.607-7.607", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3646.wav", "onoffCaption": "tapping clicking clanking at 1.605-5.045, 6.511-8.868", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3708.wav", "onoffCaption": "dog barking at 0.956-2.956 and sneeze at 1.521-3.466", "frequencyCaption": "dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3747.wav", "onoffCaption": "door slamming at 3.944-5.083, 5.988-7.127", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3931.wav", "onoffCaption": "explosion at 0.693-3.422, 4.387-7.116", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3964.wav", "onoffCaption": "sheep goat bleating at 2.099-4.099, 6.494-8.494", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_128.wav", "onoffCaption": "train horn at 0.374-5.276, 6.495-9.026", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_198.wav", "onoffCaption": "gunshot at 0.784-2.784, 3.801-5.801, 6.312-8.312", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_215.wav", "onoffCaption": "whistling at 0.949-8.699", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_240.wav", "onoffCaption": "door slamming at 1.471-2.371, 3.25-4.15, 5.324-6.224", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_430.wav", "onoffCaption": "spraying at 1.877-2.461, 3.411-3.995, 4.65-5.234", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_465.wav", "onoffCaption": "dog barking at 0.878-6.515, 7.656-10.0 and tapping clicking clanking at 3.261-6.701", "frequencyCaption": "dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_616.wav", "onoffCaption": "sheep goat bleating at 2.186-4.186, 4.941-6.941", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_934.wav", "onoffCaption": "spraying at 1.534-2.709, 5.115-5.69, 7.889-10.0", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_984.wav", "onoffCaption": "door knocking at 3.227-7.737", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1100.wav", "onoffCaption": "sneeze at 2.114-5.189, 6.823-9.898", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1155.wav", "onoffCaption": "explosion at 0.63-3.383, 4.501-7.194", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1326.wav", "onoffCaption": "tapping clicking clanking at 2.131-5.571, 7.638-10.0 and explosion at 2.898-7.898", "frequencyCaption": "tapping clicking clanking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1418.wav", "onoffCaption": "woman laughing at 1.728-3.928, 4.951-7.688", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1503.wav", "onoffCaption": "train horn at 1.911-4.711, 5.396-7.503", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1556.wav", "onoffCaption": "explosion at 1.122-3.875 and dog barking at 7.283-9.283", "frequencyCaption": "explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1770.wav", "onoffCaption": "sneeze at 1.912-4.987, 5.626-7.238", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1795.wav", "onoffCaption": "thump thud at 0.158-3.205, 5.267-7.767", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1852.wav", "onoffCaption": "woman laughing at 0.084-2.2, 2.942-5.058, 5.698-7.814 and duck quacking at 6.149-8.149", "frequencyCaption": "woman laughing three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1949.wav", "onoffCaption": "cat meowing at 2.368-5.272, 6.225-9.129 and sheep goat bleating at 7.594-9.594", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3079.wav", "onoffCaption": "burping belching at 2.466-4.696", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3137.wav", "onoffCaption": "sneeze at 2.975-4.069", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3162.wav", "onoffCaption": "duck quacking at 2.2-4.2 and train horn at 2.651-5.811", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3187.wav", "onoffCaption": "cow mooing at 0.219-4.648, 6.771-9.484", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3344.wav", "onoffCaption": "dog barking at 0.909-2.909", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3534.wav", "onoffCaption": "cat meowing at 3.579-6.609", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3584.wav", "onoffCaption": "explosion at 2.028-7.028, 7.591-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3609.wav", "onoffCaption": "cat meowing at 0.249-1.793, 3.33-4.874, 6.573-8.117 and thump thud at 2.851-5.079", "frequencyCaption": "cat meowing three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3712.wav", "onoffCaption": "tapping clicking clanking at 0.164-3.604 and cat meowing at 1.057-2.332 and explosion at 7.493-10.0", "frequencyCaption": "tapping clicking clanking one times and cat meowing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3830.wav", "onoffCaption": "sneeze at 2.907-4.435, 6.484-8.887", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3865.wav", "onoffCaption": "door knocking at 0.025-2.337 and dog barking at 0.545-2.545 and woman laughing at 6.061-8.166", "frequencyCaption": "door knocking one times and dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_212.wav", "onoffCaption": "woman laughing at 3.453-5.736, 6.978-9.178", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_309.wav", "onoffCaption": "sheep goat bleating at 0.718-2.718, 4.398-6.398, 7.843-9.843", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_437.wav", "onoffCaption": "car horn honking at 2.062-6.311, 6.944-9.159", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_462.wav", "onoffCaption": "duck quacking at 0.153-2.153, 2.773-4.773, 5.307-7.307", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_480.wav", "onoffCaption": "explosion at 1.567-3.631, 4.557-6.621, 7.919-9.983", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_579.wav", "onoffCaption": "cat meowing at 0.489-1.849, 2.754-3.763, 5.08-6.83 and car horn honking at 4.296-6.809 and spraying at 7.112-7.963", "frequencyCaption": "cat meowing three times and car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_644.wav", "onoffCaption": "duck quacking at 0.586-2.586", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_828.wav", "onoffCaption": "spraying at 2.197-4.781", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_933.wav", "onoffCaption": "spraying at 0.268-0.843, 1.489-2.064, 3.367-3.942 and duck quacking at 1.62-3.62", "frequencyCaption": "spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_961.wav", "onoffCaption": "cat meowing at 2.641-3.653, 4.491-5.502", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_983.wav", "onoffCaption": "dog barking at 2.408-4.408, 5.296-7.296", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1107.wav", "onoffCaption": "dog barking at 0.471-2.471, 4.869-6.869, 7.846-9.846", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1374.wav", "onoffCaption": "car horn honking at 2.04-5.215, 6.66-9.835", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1391.wav", "onoffCaption": "cat meowing at 0.847-2.579", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1551.wav", "onoffCaption": "cat meowing at 2.852-5.04, 7.233-8.508", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1639.wav", "onoffCaption": "whistling at 1.187-10.0 and thump thud at 3.478-7.928", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1722.wav", "onoffCaption": "woman laughing at 0.231-7.676 and duck quacking at 3.115-5.115, 6.127-8.127", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1807.wav", "onoffCaption": "burping belching at 0.067-2.888, 4.14-6.961", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1855.wav", "onoffCaption": "thump thud at 3.516-5.744, 6.49-8.718", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3165.wav", "onoffCaption": "thump thud at 0.594-5.044 and whistling at 7.097-10.0", "frequencyCaption": "thump thud one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_3180.wav", "onoffCaption": "woman laughing at 3.115-5.72", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3316.wav", "onoffCaption": "sneeze at 0.781-2.488, 3.315-4.489 and thump thud at 7.487-10.0", "frequencyCaption": "sneeze two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3428.wav", "onoffCaption": "spraying at 3.09-6.109, 7.503-9.198", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3533.wav", "onoffCaption": "spraying at 2.095-3.176, 3.935-5.668 and door knocking at 3.221-5.951, 7.684-10.0", "frequencyCaption": "spraying two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3561.wav", "onoffCaption": "tapping clicking clanking at 2.666-6.106, 7.024-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3583.wav", "onoffCaption": "cow mooing at 1.846-6.275, 7.561-9.783", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3740.wav", "onoffCaption": "duck quacking at 0.909-2.909, 3.443-5.443, 6.475-8.475 and spraying at 3.475-4.079, 4.593-5.197, 6.294-6.898 and sneeze at 4.982-7.443", "frequencyCaption": "duck quacking three times and spraying three times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3837.wav", "onoffCaption": "explosion at 0.018-2.02, 2.817-4.905, 6.975-9.847", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_3862.wav", "onoffCaption": "duck quacking at 0.105-2.105, 2.639-4.639, 5.211-7.211 and sheep goat bleating at 0.824-2.824, 3.832-5.832, 7.049-9.049 and train horn at 1.048-7.108", "frequencyCaption": "duck quacking three times and sheep goat bleating three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3880.wav", "onoffCaption": "explosion at 2.536-5.592, 6.352-9.352", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3979.wav", "onoffCaption": "thump thud at 1.813-4.584 and sheep goat bleating at 2.77-4.77 and cat meowing at 7.863-9.407", "frequencyCaption": "thump thud one times and sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_247.wav", "onoffCaption": "car horn honking at 1.058-5.458 and spraying at 8.191-8.775", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_346.wav", "onoffCaption": "tapping clicking clanking at 3.424-6.864 and gunshot at 4.765-6.765 and sneeze at 4.781-7.094", "frequencyCaption": "tapping clicking clanking one times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_487.wav", "onoffCaption": "gunshot at 2.351-4.521 and door slamming at 6.019-7.252", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_611.wav", "onoffCaption": "train horn at 0.168-10.0 and dog barking at 0.823-2.823, 5.236-7.236", "frequencyCaption": "train horn one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_710.wav", "onoffCaption": "burping belching at 2.117-5.661, 6.597-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_898.wav", "onoffCaption": "duck quacking at 3.594-5.594, 7.372-9.372", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_966.wav", "onoffCaption": "train horn at 3.481-6.801", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1049.wav", "onoffCaption": "woman laughing at 2.805-5.03", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1152.wav", "onoffCaption": "dog barking at 3.6-6.521", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1220.wav", "onoffCaption": "gunshot at 0.582-2.582, 3.6-5.6, 7.272-9.778 and cat meowing at 3.225-4.372", "frequencyCaption": "gunshot three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1321.wav", "onoffCaption": "cow mooing at 0.732-3.701 and gunshot at 2.699-4.699, 5.95-7.95", "frequencyCaption": "cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1504.wav", "onoffCaption": "gunshot at 2.245-4.245, 4.846-6.846", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1676.wav", "onoffCaption": "thump thud at 0.322-4.772, 6.46-9.046 and gunshot at 1.752-3.752", "frequencyCaption": "thump thud two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1689.wav", "onoffCaption": "tapping clicking clanking at 0.371-3.811, 5.342-8.311", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1693.wav", "onoffCaption": "sneeze at 2.165-3.872, 4.533-6.24 and door knocking at 2.828-5.068", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1777.wav", "onoffCaption": "whistling at 3.115-8.29", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1788.wav", "onoffCaption": "door knocking at 0.823-3.932 and train horn at 6.027-9.797", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1792.wav", "onoffCaption": "train horn at 1.612-6.514, 7.852-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1800.wav", "onoffCaption": "duck quacking at 0.833-2.833 and dog barking at 4.411-6.411, 7.889-9.889", "frequencyCaption": "duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3130.wav", "onoffCaption": "gunshot at 3.509-6.015", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3242.wav", "onoffCaption": "whistling at 0.169-3.144, 4.486-7.391", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3258.wav", "onoffCaption": "gunshot at 1.913-3.913 and tapping clicking clanking at 7.042-10.0", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3343.wav", "onoffCaption": "woman laughing at 0.244-2.527", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3359.wav", "onoffCaption": "door slamming at 0.337-2.465", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3498.wav", "onoffCaption": "door slamming at 0.132-2.59, 4.435-7.409 and tapping clicking clanking at 0.201-3.641, 5.333-8.773", "frequencyCaption": "door slamming two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3566.wav", "onoffCaption": "gunshot at 0.056-2.056, 3.398-5.398 and spraying at 2.521-3.029, 4.355-4.855, 5.358-7.942", "frequencyCaption": "gunshot two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3614.wav", "onoffCaption": "spraying at 3.09-3.871, 4.998-5.779", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3715.wav", "onoffCaption": "whistling at 2.984-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3887.wav", "onoffCaption": "dog barking at 1.311-3.311, 4.547-7.468", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_47.wav", "onoffCaption": "sneeze at 2.538-3.702, 5.546-6.8 and duck quacking at 3.634-5.634, 7.978-9.978", "frequencyCaption": "sneeze two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_135.wav", "onoffCaption": "sneeze at 0.624-1.716, 2.751-5.365, 5.998-8.974", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_208.wav", "onoffCaption": "car horn honking at 0.093-4.493, 6.927-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_313.wav", "onoffCaption": "car horn honking at 0.15-2.15 and door knocking at 4.327-6.576, 7.544-10.0", "frequencyCaption": "car horn honking one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_478.wav", "onoffCaption": "burping belching at 0.132-2.362, 3.987-6.313 and woman laughing at 3.778-7.166", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_563.wav", "onoffCaption": "sneeze at 0.216-1.51, 3.715-4.809", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_586.wav", "onoffCaption": "cat meowing at 0.42-2.17 and door knocking at 2.95-6.712 and door slamming at 6.165-7.145", "frequencyCaption": "cat meowing one times and door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_745.wav", "onoffCaption": "whistling at 1.282-9.032 and sneeze at 3.16-5.156, 7.293-9.289", "frequencyCaption": "whistling one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_867.wav", "onoffCaption": "tapping clicking clanking at 0.487-3.927, 6.34-9.002", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_882.wav", "onoffCaption": "gunshot at 0.137-2.137, 4.515-6.515", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_999.wav", "onoffCaption": "thump thud at 0.977-4.024, 5.736-8.075", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1053.wav", "onoffCaption": "sheep goat bleating at 2.992-4.992, 5.618-7.618", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1148.wav", "onoffCaption": "whistling at 1.666-7.607 and cat meowing at 7.548-9.084", "frequencyCaption": "whistling one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1275.wav", "onoffCaption": "duck quacking at 1.585-3.585, 4.114-6.114", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1290.wav", "onoffCaption": "whistling at 1.248-9.259 and door knocking at 1.729-4.576, 5.851-8.698", "frequencyCaption": "whistling one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1405.wav", "onoffCaption": "explosion at 0.883-5.883 and spraying at 6.432-6.94, 8.695-9.203", "frequencyCaption": "explosion one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1623.wav", "onoffCaption": "door knocking at 1.398-7.458", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1738.wav", "onoffCaption": "sneeze at 2.589-3.692, 4.688-6.684, 7.629-9.174", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1901.wav", "onoffCaption": "tapping clicking clanking at 1.129-4.569", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3031.wav", "onoffCaption": "whistling at 3.449-8.949", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3217.wav", "onoffCaption": "door slamming at 1.776-3.167, 5.234-6.134", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3467.wav", "onoffCaption": "cat meowing at 3.211-4.755, 5.561-7.538", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3482.wav", "onoffCaption": "sheep goat bleating at 3.012-5.012, 6.095-8.095", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3599.wav", "onoffCaption": "burping belching at 0.315-3.605, 4.439-6.669 and dog barking at 4.742-6.742 and train horn at 5.884-8.364", "frequencyCaption": "burping belching two times and dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3641.wav", "onoffCaption": "car horn honking at 0.367-3.542, 4.043-7.218 and cow mooing at 0.835-5.815, 7.094-9.789", "frequencyCaption": "car horn honking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3878.wav", "onoffCaption": "cow mooing at 2.415-7.395 and car horn honking at 2.709-7.031", "frequencyCaption": "cow mooing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3963.wav", "onoffCaption": "tapping clicking clanking at 0.761-4.201, 6.468-9.015", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3986.wav", "onoffCaption": "tapping clicking clanking at 1.468-4.908, 6.936-9.688", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_12.wav", "onoffCaption": "car horn honking at 0.54-5.447 and duck quacking at 3.391-5.391, 7.677-9.677", "frequencyCaption": "car horn honking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_63.wav", "onoffCaption": "woman laughing at 1.87-4.436, 6.27-8.553", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_86.wav", "onoffCaption": "train horn at 3.875-7.875", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_111.wav", "onoffCaption": "car horn honking at 0.756-2.756, 4.181-6.708", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_160.wav", "onoffCaption": "cow mooing at 0.675-5.655", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_185.wav", "onoffCaption": "cow mooing at 1.172-4.182 and door knocking at 2.643-6.915", "frequencyCaption": "cow mooing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_279.wav", "onoffCaption": "cat meowing at 2.652-4.84, 5.951-8.087", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_337.wav", "onoffCaption": "spraying at 3.394-3.963", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_362.wav", "onoffCaption": "gunshot at 3.922-5.922", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_387.wav", "onoffCaption": "thump thud at 0.64-2.979, 3.98-6.751", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_536.wav", "onoffCaption": "thump thud at 1.602-6.052 and burping belching at 2.209-4.316, 5.552-7.659", "frequencyCaption": "thump thud one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_547.wav", "onoffCaption": "tapping clicking clanking at 1.737-5.177, 6.879-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_734.wav", "onoffCaption": "sneeze at 1.264-3.51, 5.768-8.014", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_832.wav", "onoffCaption": "door knocking at 0.635-2.699", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_843.wav", "onoffCaption": "dog barking at 0.476-2.476, 3.128-5.128 and duck quacking at 7.098-9.098", "frequencyCaption": "dog barking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_929.wav", "onoffCaption": "burping belching at 1.548-7.492 and cat meowing at 2.021-3.065, 4.919-5.963", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_958.wav", "onoffCaption": "burping belching at 3.236-6.236", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1006.wav", "onoffCaption": "tapping clicking clanking at 0.14-3.58, 5.135-8.575", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1077.wav", "onoffCaption": "tapping clicking clanking at 0.177-3.617, 5.908-7.999 and cat meowing at 0.395-1.581 and whistling at 0.574-8.959", "frequencyCaption": "tapping clicking clanking two times and cat meowing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1092.wav", "onoffCaption": "burping belching at 0.135-4.004, 5.916-8.783", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1189.wav", "onoffCaption": "whistling at 3.358-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1204.wav", "onoffCaption": "sheep goat bleating at 0.234-2.234, 2.756-4.756, 5.332-7.332", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1421.wav", "onoffCaption": "cat meowing at 0.461-1.563 and burping belching at 0.476-5.476, 7.964-10.0", "frequencyCaption": "cat meowing one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1450.wav", "onoffCaption": "gunshot at 0.897-2.897 and cat meowing at 5.661-7.197", "frequencyCaption": "gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1652.wav", "onoffCaption": "door slamming at 0.328-1.179 and cat meowing at 4.049-5.198, 6.096-7.245 and spraying at 5.556-6.497, 8.78-9.955", "frequencyCaption": "door slamming one times and cat meowing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1749.wav", "onoffCaption": "gunshot at 1.998-3.998 and cow mooing at 5.91-8.879", "frequencyCaption": "gunshot one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1925.wav", "onoffCaption": "explosion at 1.732-4.604", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1954.wav", "onoffCaption": "duck quacking at 1.19-3.19, 5.027-7.027", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3015.wav", "onoffCaption": "explosion at 0.061-5.061 and door knocking at 0.093-3.855", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3064.wav", "onoffCaption": "door knocking at 0.013-2.365, 3.513-5.865", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3081.wav", "onoffCaption": "woman laughing at 1.463-4.548, 5.645-7.891", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3266.wav", "onoffCaption": "burping belching at 0.198-3.4 and gunshot at 7.023-9.524", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3283.wav", "onoffCaption": "cow mooing at 3.078-8.058", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3398.wav", "onoffCaption": "sheep goat bleating at 1.518-3.518 and sneeze at 6.834-8.16", "frequencyCaption": "sheep goat bleating one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3432.wav", "onoffCaption": "sneeze at 2.248-3.482, 5.54-6.774", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3443.wav", "onoffCaption": "whistling at 0.159-5.334", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3529.wav", "onoffCaption": "door knocking at 1.623-4.46", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3558.wav", "onoffCaption": "spraying at 1.0-2.486 and cow mooing at 5.754-8.723", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3630.wav", "onoffCaption": "train horn at 0.886-4.286, 6.072-8.872 and tapping clicking clanking at 2.266-5.706", "frequencyCaption": "train horn two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3936.wav", "onoffCaption": "spraying at 0.586-2.281, 4.305-6.0, 6.736-8.431 and sheep goat bleating at 3.311-7.231, 7.902-9.902", "frequencyCaption": "spraying three times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3947.wav", "onoffCaption": "train horn at 0.112-2.912 and whistling at 1.057-4.032 and cow mooing at 5.466-8.476", "frequencyCaption": "train horn one times and whistling one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_36.wav", "onoffCaption": "sheep goat bleating at 2.518-5.518, 6.541-9.541", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_144.wav", "onoffCaption": "cat meowing at 0.305-5.305, 6.539-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_409.wav", "onoffCaption": "explosion at 1.848-4.848, 5.553-8.553", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_413.wav", "onoffCaption": "door slamming at 1.337-1.837, 2.978-3.478", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_508.wav", "onoffCaption": "gunshot at 2.17-4.17, 5.36-7.36", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_512.wav", "onoffCaption": "dog barking at 0.622-2.622", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_660.wav", "onoffCaption": "whistling at 2.177-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_685.wav", "onoffCaption": "cow mooing at 0.06-3.029, 4.064-6.951, 7.727-9.748", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_761.wav", "onoffCaption": "thump thud at 0.843-5.293", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_784.wav", "onoffCaption": "cow mooing at 0.475-3.444, 5.211-7.742 and tapping clicking clanking at 3.309-6.749", "frequencyCaption": "cow mooing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_816.wav", "onoffCaption": "explosion at 1.724-3.726, 5.689-7.691", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_917.wav", "onoffCaption": "explosion at 1.982-6.982 and tapping clicking clanking at 3.567-7.007", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1022.wav", "onoffCaption": "whistling at 0.122-8.133 and door slamming at 0.182-1.321, 2.275-2.775, 4.526-6.526", "frequencyCaption": "whistling one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_1038.wav", "onoffCaption": "train horn at 0.264-2.938, 5.056-7.536 and tapping clicking clanking at 0.704-4.144 and door knocking at 2.107-4.944, 5.54-8.377", "frequencyCaption": "train horn two times and tapping clicking clanking one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1123.wav", "onoffCaption": "tapping clicking clanking at 0.144-3.584, 5.286-7.478", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1139.wav", "onoffCaption": "woman laughing at 0.626-2.82, 5.16-7.354", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1251.wav", "onoffCaption": "thump thud at 2.421-6.796", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1350.wav", "onoffCaption": "door knocking at 0.09-2.927, 5.354-7.487 and woman laughing at 1.506-3.7, 4.763-6.851", "frequencyCaption": "door knocking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1474.wav", "onoffCaption": "gunshot at 2.464-4.464", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1491.wav", "onoffCaption": "tapping clicking clanking at 0.394-3.834, 5.02-8.46", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1575.wav", "onoffCaption": "tapping clicking clanking at 0.154-3.594, 4.396-6.695 and burping belching at 0.413-2.507, 3.041-5.148 and door slamming at 0.419-3.38, 4.399-7.36", "frequencyCaption": "tapping clicking clanking two times and burping belching two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1590.wav", "onoffCaption": "tapping clicking clanking at 2.991-6.431 and sneeze at 4.173-6.132", "frequencyCaption": "tapping clicking clanking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1607.wav", "onoffCaption": "duck quacking at 2.35-4.35, 5.08-7.08", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1706.wav", "onoffCaption": "train horn at 2.838-7.019", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1871.wav", "onoffCaption": "woman laughing at 0.217-2.855", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1894.wav", "onoffCaption": "gunshot at 0.413-2.919, 4.301-6.807", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1970.wav", "onoffCaption": "tapping clicking clanking at 0.321-3.761, 5.461-8.901", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1995.wav", "onoffCaption": "tapping clicking clanking at 0.432-3.872, 4.555-7.09", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3040.wav", "onoffCaption": "duck quacking at 2.579-4.579, 6.775-8.775", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3141.wav", "onoffCaption": "cat meowing at 0.69-1.83, 2.623-3.763, 4.769-5.909 and car horn honking at 7.826-10.0", "frequencyCaption": "cat meowing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3229.wav", "onoffCaption": "woman laughing at 3.984-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3233.wav", "onoffCaption": "thump thud at 2.174-4.674, 6.749-9.249 and door knocking at 2.65-6.266", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3328.wav", "onoffCaption": "whistling at 0.727-7.976 and thump thud at 1.013-5.463", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3332.wav", "onoffCaption": "whistling at 0.582-7.158", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3416.wav", "onoffCaption": "cow mooing at 0.067-3.365, 3.932-6.914 and train horn at 2.227-6.667", "frequencyCaption": "cow mooing two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3517.wav", "onoffCaption": "gunshot at 1.138-3.138", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3665.wav", "onoffCaption": "car horn honking at 0.511-4.833 and whistling at 3.066-5.295, 6.173-8.456", "frequencyCaption": "car horn honking one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3680.wav", "onoffCaption": "sneeze at 0.858-2.386", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3764.wav", "onoffCaption": "tapping clicking clanking at 1.057-4.497, 6.107-8.928 and whistling at 1.373-3.382, 4.394-7.369", "frequencyCaption": "tapping clicking clanking two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_3781.wav", "onoffCaption": "car horn honking at 2.047-5.634 and door slamming at 8.135-8.94", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3809.wav", "onoffCaption": "thump thud at 3.075-5.846, 6.39-9.161", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3813.wav", "onoffCaption": "train horn at 1.522-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3908.wav", "onoffCaption": "spraying at 0.562-2.295 and gunshot at 5.254-7.384", "frequencyCaption": "spraying one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3912.wav", "onoffCaption": "gunshot at 2.345-4.345, 5.839-7.839", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_79.wav", "onoffCaption": "door knocking at 1.555-4.675, 6.777-9.897", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_189.wav", "onoffCaption": "dog barking at 2.828-4.828, 6.53-8.53", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_204.wav", "onoffCaption": "car horn honking at 2.9-5.365, 7.793-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_236.wav", "onoffCaption": "car horn honking at 2.477-4.977, 5.601-8.002", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_263.wav", "onoffCaption": "cow mooing at 0.559-3.857, 5.176-8.474", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_286.wav", "onoffCaption": "spraying at 3.75-6.186, 7.898-10.0", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_378.wav", "onoffCaption": "woman laughing at 0.099-2.894, 4.837-7.297 and explosion at 0.157-3.157", "frequencyCaption": "woman laughing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_421.wav", "onoffCaption": "tapping clicking clanking at 2.375-5.815 and sheep goat bleating at 4.545-6.545", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_446.wav", "onoffCaption": "cat meowing at 0.124-1.337, 2.502-3.715, 4.842-6.055", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_635.wav", "onoffCaption": "sheep goat bleating at 0.123-2.123, 2.76-4.76, 5.892-7.892", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_652.wav", "onoffCaption": "cow mooing at 3.561-6.571, 7.654-10.0 and duck quacking at 4.115-6.115", "frequencyCaption": "cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_859.wav", "onoffCaption": "duck quacking at 0.31-2.31, 4.52-6.52 and thump thud at 4.117-8.035", "frequencyCaption": "duck quacking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_925.wav", "onoffCaption": "whistling at 1.132-8.882", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_942.wav", "onoffCaption": "door slamming at 2.154-3.054", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1088.wav", "onoffCaption": "cow mooing at 0.067-3.365 and explosion at 6.581-10.0", "frequencyCaption": "cow mooing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_1111.wav", "onoffCaption": "whistling at 2.179-9.929", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1176.wav", "onoffCaption": "spraying at 0.051-0.801, 2.592-3.342, 5.836-6.586 and dog barking at 1.423-3.423", "frequencyCaption": "spraying three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1193.wav", "onoffCaption": "cow mooing at 0.157-3.455, 4.239-7.537 and door slamming at 1.499-2.479, 3.672-4.652", "frequencyCaption": "cow mooing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1279.wav", "onoffCaption": "tapping clicking clanking at 0.933-4.373, 4.929-7.481", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1305.wav", "onoffCaption": "woman laughing at 3.907-6.19 and duck quacking at 7.724-9.724", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1362.wav", "onoffCaption": "sheep goat bleating at 2.838-4.838", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1387.wav", "onoffCaption": "dog barking at 1.304-3.304, 3.955-5.955, 6.664-8.664", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1520.wav", "onoffCaption": "tapping clicking clanking at 2.832-6.272, 6.792-9.216", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1547.wav", "onoffCaption": "dog barking at 0.286-2.286, 4.345-6.345", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1648.wav", "onoffCaption": "gunshot at 0.398-2.398 and cat meowing at 2.098-3.11, 5.228-6.24", "frequencyCaption": "gunshot one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1734.wav", "onoffCaption": "explosion at 2.33-5.33, 6.621-8.676 and duck quacking at 3.013-5.013, 5.937-7.937", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1753.wav", "onoffCaption": "thump thud at 1.028-4.695, 5.376-9.043", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1824.wav", "onoffCaption": "dog barking at 0.275-2.275, 4.236-6.636", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1843.wav", "onoffCaption": "train horn at 2.459-5.779 and cow mooing at 2.492-5.502", "frequencyCaption": "train horn one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1958.wav", "onoffCaption": "sheep goat bleating at 2.891-4.891 and spraying at 8.283-9.347", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3068.wav", "onoffCaption": "door knocking at 2.829-6.204, 7.014-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3114.wav", "onoffCaption": "cow mooing at 2.794-5.776, 6.601-9.583", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3173.wav", "onoffCaption": "woman laughing at 3.122-6.222", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3196.wav", "onoffCaption": "duck quacking at 0.461-2.461, 2.976-4.976, 5.619-7.619", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3299.wav", "onoffCaption": "whistling at 0.056-8.441", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3300.wav", "onoffCaption": "spraying at 0.055-1.055 and woman laughing at 0.776-4.057, 4.642-7.923", "frequencyCaption": "spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3367.wav", "onoffCaption": "cow mooing at 0.244-4.673 and thump thud at 2.481-5.528, 7.722-9.95", "frequencyCaption": "cow mooing one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3382.wav", "onoffCaption": "cat meowing at 1.522-6.522, 8.431-9.443", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3459.wav", "onoffCaption": "train horn at 2.054-6.912 and door slamming at 4.29-6.486, 7.105-9.233", "frequencyCaption": "train horn one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3525.wav", "onoffCaption": "dog barking at 2.799-4.799, 6.347-8.347", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3542.wav", "onoffCaption": "car horn honking at 0.751-3.664 and duck quacking at 6.723-8.723", "frequencyCaption": "car horn honking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3731.wav", "onoffCaption": "sheep goat bleating at 0.142-2.142, 3.083-5.083 and burping belching at 7.866-9.897", "frequencyCaption": "sheep goat bleating two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3756.wav", "onoffCaption": "cow mooing at 1.082-6.062, 7.102-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3821.wav", "onoffCaption": "sneeze at 1.225-2.611, 3.422-4.808, 5.872-7.258 and cat meowing at 2.894-4.465, 6.911-7.922", "frequencyCaption": "sneeze three times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3846.wav", "onoffCaption": "car horn honking at 0.208-4.53, 6.828-9.725 and whistling at 1.886-7.888", "frequencyCaption": "car horn honking two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_51.wav", "onoffCaption": "cat meowing at 0.538-5.538, 6.28-8.719 and door knocking at 5.551-7.678", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_139.wav", "onoffCaption": "car horn honking at 0.702-3.628, 4.583-7.509", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_251.wav", "onoffCaption": "whistling at 1.799-6.974, 7.85-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_350.wav", "onoffCaption": "train horn at 0.188-3.668, 5.14-8.62 and gunshot at 2.187-4.187, 5.945-7.945 and door knocking at 2.794-5.914, 6.611-8.799", "frequencyCaption": "train horn two times and gunshot two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_474.wav", "onoffCaption": "dog barking at 0.186-3.107, 4.014-6.014, 6.644-8.644 and duck quacking at 0.882-2.882, 5.239-7.239", "frequencyCaption": "dog barking three times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_491.wav", "onoffCaption": "duck quacking at 3.523-5.523", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_590.wav", "onoffCaption": "cat meowing at 2.657-7.017, 7.526-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_607.wav", "onoffCaption": "whistling at 1.666-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_706.wav", "onoffCaption": "sneeze at 0.5-2.183, 4.167-5.85", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_749.wav", "onoffCaption": "thump thud at 0.647-3.147, 4.243-6.582 and dog barking at 4.148-6.148", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_871.wav", "onoffCaption": "train horn at 1.148-3.822, 5.745-8.419 and sheep goat bleating at 2.434-4.434", "frequencyCaption": "train horn two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_894.wav", "onoffCaption": "cow mooing at 2.365-6.794, 7.82-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_970.wav", "onoffCaption": "door knocking at 2.028-4.796, 6.209-8.977 and whistling at 2.521-5.496", "frequencyCaption": "door knocking two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_995.wav", "onoffCaption": "woman laughing at 3.336-6.131", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1045.wav", "onoffCaption": "woman laughing at 3.529-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1144.wav", "onoffCaption": "thump thud at 0.479-4.929 and tapping clicking clanking at 3.187-6.627", "frequencyCaption": "thump thud one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1236.wav", "onoffCaption": "cow mooing at 2.627-5.596, 6.917-9.886", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1337.wav", "onoffCaption": "burping belching at 2.433-4.636, 5.194-7.367", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1409.wav", "onoffCaption": "sneeze at 3.14-5.601, 7.313-9.789 and thump thud at 3.253-6.3", "frequencyCaption": "sneeze two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1413.wav", "onoffCaption": "sneeze at 0.398-2.105 and spraying at 0.939-3.067", "frequencyCaption": "sneeze one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1508.wav", "onoffCaption": "explosion at 0.103-3.29", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1512.wav", "onoffCaption": "door slamming at 0.683-2.596 and car horn honking at 5.318-9.159", "frequencyCaption": "door slamming one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1660.wav", "onoffCaption": "cow mooing at 0.418-3.428, 3.973-6.583", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1685.wav", "onoffCaption": "sheep goat bleating at 0.076-2.076, 2.698-4.698, 5.861-7.861", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1761.wav", "onoffCaption": "thump thud at 0.738-5.188, 7.319-9.819", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1784.wav", "onoffCaption": "cow mooing at 0.738-5.167, 5.903-8.899", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1816.wav", "onoffCaption": "cat meowing at 0.631-4.764, 6.378-7.527", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1917.wav", "onoffCaption": "burping belching at 0.12-3.664, 4.722-8.266", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3027.wav", "onoffCaption": "explosion at 1.301-6.301, 7.603-9.923 and gunshot at 2.288-4.762, 5.447-7.921", "frequencyCaption": "explosion two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3126.wav", "onoffCaption": "spraying at 0.002-0.906, 1.602-3.335 and gunshot at 6.789-8.789", "frequencyCaption": "spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3254.wav", "onoffCaption": "burping belching at 1.066-4.346 and sheep goat bleating at 2.429-4.429, 5.198-7.198", "frequencyCaption": "burping belching one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3355.wav", "onoffCaption": "cat meowing at 0.682-1.709, 2.888-3.915, 4.78-5.807 and burping belching at 0.736-3.359", "frequencyCaption": "cat meowing three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3471.wav", "onoffCaption": "spraying at 1.916-3.0", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3494.wav", "onoffCaption": "explosion at 2.713-5.441, 7.005-9.733", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3570.wav", "onoffCaption": "gunshot at 1.296-3.296, 4.812-6.812", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3595.wav", "onoffCaption": "cat meowing at 2.868-7.228, 8.142-9.291", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3602.wav", "onoffCaption": "explosion at 0.527-3.527 and train horn at 1.368-3.808, 5.59-8.03", "frequencyCaption": "explosion one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_3618.wav", "onoffCaption": "sheep goat bleating at 3.257-5.257", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3703.wav", "onoffCaption": "explosion at 0.155-5.155 and sheep goat bleating at 0.686-3.686, 4.666-7.666 and duck quacking at 6.741-8.741", "frequencyCaption": "explosion one times and sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3719.wav", "onoffCaption": "sheep goat bleating at 0.275-2.275, 3.223-5.223, 5.77-7.77", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3874.wav", "onoffCaption": "duck quacking at 3.559-5.559, 6.561-8.561", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3891.wav", "onoffCaption": "whistling at 0.043-2.918 and dog barking at 0.052-2.052", "frequencyCaption": "whistling one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3990.wav", "onoffCaption": "tapping clicking clanking at 0.705-4.145, 5.601-7.818", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_90.wav", "onoffCaption": "woman laughing at 0.803-3.408, 3.992-6.787", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_123.wav", "onoffCaption": "sheep goat bleating at 0.202-2.202, 3.375-5.375 and sneeze at 3.476-5.472, 6.037-8.033 and cow mooing at 5.23-8.24", "frequencyCaption": "sheep goat bleating two times and sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_176.wav", "onoffCaption": "cow mooing at 3.536-6.834, 7.601-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_193.wav", "onoffCaption": "explosion at 0.339-3.339, 5.515-7.877", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_305.wav", "onoffCaption": "train horn at 3.158-5.625, 6.413-9.115 and cow mooing at 5.053-8.022", "frequencyCaption": "train horn two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_374.wav", "onoffCaption": "dog barking at 0.106-2.106, 3.22-5.22", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_391.wav", "onoffCaption": "explosion at 0.435-3.164, 4.541-6.834 and spraying at 3.273-6.292", "frequencyCaption": "explosion two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_520.wav", "onoffCaption": "thump thud at 1.037-3.376 and gunshot at 5.909-7.909", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_575.wav", "onoffCaption": "door knocking at 2.665-5.433 and spraying at 7.735-8.339, 8.877-9.481", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_639.wav", "onoffCaption": "dog barking at 0.024-2.024, 2.943-4.943, 5.802-7.802", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_648.wav", "onoffCaption": "car horn honking at 0.672-4.326 and tapping clicking clanking at 6.904-10.0", "frequencyCaption": "car horn honking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_722.wav", "onoffCaption": "whistling at 0.563-8.574 and tapping clicking clanking at 1.14-4.58", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_753.wav", "onoffCaption": "explosion at 0.834-3.834, 5.832-8.832", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_824.wav", "onoffCaption": "burping belching at 2.968-5.198, 6.24-9.24", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_855.wav", "onoffCaption": "sneeze at 1.041-3.655, 5.037-7.651", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1010.wav", "onoffCaption": "cow mooing at 2.924-5.893", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1084.wav", "onoffCaption": "explosion at 2.02-7.02", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1212.wav", "onoffCaption": "cow mooing at 0.686-3.696 and door knocking at 7.844-10.0", "frequencyCaption": "cow mooing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1263.wav", "onoffCaption": "whistling at 2.648-5.523, 6.602-9.287", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1286.wav", "onoffCaption": "sneeze at 2.554-4.513 and thump thud at 6.691-10.0", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1309.wav", "onoffCaption": "cow mooing at 3.532-7.961 and spraying at 4.142-5.389, 6.187-7.128", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1378.wav", "onoffCaption": "door slamming at 0.234-2.234 and explosion at 5.159-7.877 and cat meowing at 5.508-7.502", "frequencyCaption": "door slamming one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1437.wav", "onoffCaption": "car horn honking at 0.038-2.964, 4.164-6.34 and dog barking at 0.425-2.425", "frequencyCaption": "car horn honking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1446.wav", "onoffCaption": "cow mooing at 3.706-6.688, 7.502-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1635.wav", "onoffCaption": "car horn honking at 3.734-6.081, 6.76-8.924", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1644.wav", "onoffCaption": "woman laughing at 4.06-6.479, 7.285-9.704", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1828.wav", "onoffCaption": "dog barking at 0.029-2.029", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1859.wav", "onoffCaption": "sneeze at 0.081-3.156 and door slamming at 1.281-1.962", "frequencyCaption": "sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1933.wav", "onoffCaption": "spraying at 0.054-0.835, 1.62-2.189, 3.452-5.185 and sheep goat bleating at 1.264-3.264, 3.814-5.814, 7.277-9.277", "frequencyCaption": "spraying three times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1942.wav", "onoffCaption": "sheep goat bleating at 1.739-3.739, 6.232-8.232", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3003.wav", "onoffCaption": "whistling at 1.472-9.857 and burping belching at 1.552-4.552", "frequencyCaption": "whistling one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3072.wav", "onoffCaption": "gunshot at 0.892-2.892, 4.482-6.983", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3097.wav", "onoffCaption": "tapping clicking clanking at 0.349-3.789, 4.563-8.003", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3118.wav", "onoffCaption": "door knocking at 1.689-5.457", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3169.wav", "onoffCaption": "train horn at 3.286-8.188", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3201.wav", "onoffCaption": "woman laughing at 0.017-6.751 and sheep goat bleating at 4.882-6.882", "frequencyCaption": "woman laughing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3270.wav", "onoffCaption": "spraying at 0.023-1.718, 3.241-4.936 and burping belching at 2.094-5.374", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3295.wav", "onoffCaption": "gunshot at 0.894-2.894 and tapping clicking clanking at 1.696-5.136, 6.245-8.873 and cat meowing at 3.659-4.744", "frequencyCaption": "gunshot one times and tapping clicking clanking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3424.wav", "onoffCaption": "door slamming at 2.693-3.832, 5.0-6.139, 6.66-7.799", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3455.wav", "onoffCaption": "whistling at 0.61-9.665", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3626.wav", "onoffCaption": "tapping clicking clanking at 3.218-6.658", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3657.wav", "onoffCaption": "gunshot at 0.034-2.034, 3.662-5.662, 7.081-9.081", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3920.wav", "onoffCaption": "door slamming at 2.209-5.17", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3975.wav", "onoffCaption": "whistling at 1.833-7.333 and thump thud at 3.092-7.467 and cat meowing at 4.531-6.115", "frequencyCaption": "whistling one times and thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_20.wav", "onoffCaption": "thump thud at 0.02-4.47", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_75.wav", "onoffCaption": "burping belching at 0.181-2.704, 4.987-7.081 and duck quacking at 1.551-3.551, 5.137-7.137", "frequencyCaption": "burping belching two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_107.wav", "onoffCaption": "cow mooing at 2.043-7.023", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_148.wav", "onoffCaption": "door knocking at 2.621-4.809 and explosion at 6.949-9.821", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_152.wav", "onoffCaption": "burping belching at 0.519-3.778 and tapping clicking clanking at 6.986-10.0", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_321.wav", "onoffCaption": "woman laughing at 0.731-3.336, 4.07-6.675, 7.954-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_405.wav", "onoffCaption": "cow mooing at 0.652-5.632, 6.45-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_504.wav", "onoffCaption": "door knocking at 0.327-5.16, 5.971-8.528 and dog barking at 2.262-4.262 and cat meowing at 3.064-7.424", "frequencyCaption": "door knocking two times and dog barking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_551.wav", "onoffCaption": "cow mooing at 0.409-3.378, 3.897-6.651", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_689.wav", "onoffCaption": "gunshot at 1.993-3.993 and tapping clicking clanking at 6.029-9.469", "frequencyCaption": "gunshot one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_777.wav", "onoffCaption": "burping belching at 0.758-2.961", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_792.wav", "onoffCaption": "dog barking at 0.781-2.781", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_800.wav", "onoffCaption": "spraying at 0.055-0.563, 1.558-2.558, 3.071-3.922 and woman laughing at 5.64-7.84", "frequencyCaption": "spraying three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_901.wav", "onoffCaption": "door slamming at 0.191-1.715, 3.34-4.359 and sneeze at 7.275-10.0", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1034.wav", "onoffCaption": "whistling at 2.152-9.807", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1061.wav", "onoffCaption": "train horn at 0.129-2.266 and sheep goat bleating at 1.564-4.86, 5.722-7.722 and spraying at 6.161-7.408", "frequencyCaption": "train horn one times and sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1135.wav", "onoffCaption": "cat meowing at 0.983-2.123, 2.981-5.67 and thump thud at 1.87-6.32", "frequencyCaption": "cat meowing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1247.wav", "onoffCaption": "burping belching at 3.068-6.612, 7.278-9.991", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1462.wav", "onoffCaption": "explosion at 0.227-3.095, 4.092-6.774, 7.66-9.67", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_1478.wav", "onoffCaption": "explosion at 0.04-2.912, 4.447-7.319", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1487.wav", "onoffCaption": "thump thud at 2.941-7.391", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1563.wav", "onoffCaption": "duck quacking at 2.444-4.444 and woman laughing at 7.55-10.0", "frequencyCaption": "duck quacking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1579.wav", "onoffCaption": "door knocking at 0.166-3.928, 4.859-8.621 and door slamming at 6.402-9.183", "frequencyCaption": "door knocking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1586.wav", "onoffCaption": "door knocking at 1.791-4.031, 5.253-7.493 and train horn at 2.702-5.902", "frequencyCaption": "door knocking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1611.wav", "onoffCaption": "thump thud at 0.696-4.363 and sheep goat bleating at 2.839-5.919, 7.879-9.879", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1867.wav", "onoffCaption": "cow mooing at 2.159-5.141, 5.718-8.591", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1882.wav", "onoffCaption": "gunshot at 0.885-3.125, 5.094-7.334 and sheep goat bleating at 2.826-4.826, 6.678-8.678", "frequencyCaption": "gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1898.wav", "onoffCaption": "duck quacking at 3.245-5.245, 6.684-8.684", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1966.wav", "onoffCaption": "explosion at 0.09-2.097, 4.572-7.1", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1983.wav", "onoffCaption": "dog barking at 0.369-2.369, 4.251-6.251", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1999.wav", "onoffCaption": "cat meowing at 2.229-3.24", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3056.wav", "onoffCaption": "train horn at 1.134-3.774", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3157.wav", "onoffCaption": "train horn at 0.176-4.296 and cat meowing at 7.151-8.706", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3225.wav", "onoffCaption": "car horn honking at 2.801-5.727, 6.289-8.754 and duck quacking at 4.606-6.606", "frequencyCaption": "car horn honking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3400.wav", "onoffCaption": "explosion at 1.732-4.726, 7.146-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3501.wav", "onoffCaption": "duck quacking at 0.252-2.252", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3673.wav", "onoffCaption": "spraying at 0.097-1.161, 2.725-4.853 and gunshot at 0.222-2.222, 4.038-6.038, 7.509-9.509", "frequencyCaption": "spraying two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3696.wav", "onoffCaption": "train horn at 1.872-5.992", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3768.wav", "onoffCaption": "burping belching at 0.085-4.085, 5.224-9.224 and train horn at 5.33-9.53", "frequencyCaption": "burping belching two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3805.wav", "onoffCaption": "explosion at 0.285-2.813, 5.27-8.27 and duck quacking at 1.387-3.387", "frequencyCaption": "explosion two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3904.wav", "onoffCaption": "tapping clicking clanking at 2.524-5.964, 7.219-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3951.wav", "onoffCaption": "burping belching at 2.919-6.121, 6.638-9.84", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_220.wav", "onoffCaption": "cow mooing at 1.917-6.346, 7.969-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_227.wav", "onoffCaption": "cat meowing at 3.979-5.006 and cow mooing at 7.653-10.0", "frequencyCaption": "cat meowing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_275.wav", "onoffCaption": "tapping clicking clanking at 0.18-3.62", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_290.wav", "onoffCaption": "train horn at 0.463-3.703, 5.072-7.818 and whistling at 5.549-7.778", "frequencyCaption": "train horn two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_402.wav", "onoffCaption": "woman laughing at 1.858-4.496, 5.165-7.649 and burping belching at 2.466-5.466", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_450.wav", "onoffCaption": "tapping clicking clanking at 3.642-7.082", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_519.wav", "onoffCaption": "tapping clicking clanking at 2.995-6.435", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_623.wav", "onoffCaption": "dog barking at 0.836-2.836, 3.847-5.847, 6.983-8.983", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_671.wav", "onoffCaption": "explosion at 2.427-4.52, 5.452-8.143", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_676.wav", "onoffCaption": "spraying at 2.443-3.443 and train horn at 2.832-6.192", "frequencyCaption": "spraying one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_693.wav", "onoffCaption": "burping belching at 1.052-6.653, 7.502-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_694.wav", "onoffCaption": "tapping clicking clanking at 0.286-3.726, 4.727-8.167", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_738.wav", "onoffCaption": "burping belching at 2.978-9.955", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_788.wav", "onoffCaption": "woman laughing at 2.872-4.964, 6.418-8.711", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_906.wav", "onoffCaption": "duck quacking at 1.199-3.199, 5.451-7.451 and spraying at 1.857-2.798", "frequencyCaption": "duck quacking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_954.wav", "onoffCaption": "woman laughing at 1.448-3.685, 5.85-8.087", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1029.wav", "onoffCaption": "woman laughing at 0.266-2.852 and spraying at 1.809-2.393, 4.645-5.229", "frequencyCaption": "woman laughing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_1132.wav", "onoffCaption": "train horn at 0.148-3.508, 4.725-8.085", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1160.wav", "onoffCaption": "sheep goat bleating at 0.318-2.318, 3.094-5.094, 7.441-9.441", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1185.wav", "onoffCaption": "thump thud at 0.834-3.605 and duck quacking at 1.364-3.364, 5.81-7.81", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1208.wav", "onoffCaption": "spraying at 1.244-1.752, 4.209-4.959 and sneeze at 7.711-9.037", "frequencyCaption": "spraying two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_1313.wav", "onoffCaption": "cow mooing at 0.194-3.163, 5.187-7.601 and burping belching at 0.273-2.304 and woman laughing at 3.885-6.585", "frequencyCaption": "cow mooing two times and burping belching one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1341.wav", "onoffCaption": "cow mooing at 2.719-5.701, 6.832-9.814", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1346.wav", "onoffCaption": "spraying at 0.168-1.019", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1536.wav", "onoffCaption": "car horn honking at 1.992-5.833, 6.819-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1564.wav", "onoffCaption": "cat meowing at 0.177-2.072, 4.308-6.203 and duck quacking at 0.678-2.678, 4.915-6.915", "frequencyCaption": "cat meowing two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1581.wav", "onoffCaption": "train horn at 2.551-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1710.wav", "onoffCaption": "cat meowing at 0.096-1.198, 2.271-3.373, 4.359-5.461 and explosion at 0.76-2.824, 3.965-6.029 and sheep goat bleating at 4.978-6.978", "frequencyCaption": "cat meowing three times and explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1717.wav", "onoffCaption": "explosion at 0.595-3.595 and woman laughing at 6.279-8.634", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1745.wav", "onoffCaption": "burping belching at 0.264-3.264, 4.921-7.921", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1832.wav", "onoffCaption": "train horn at 2.432-5.072 and cow mooing at 2.509-5.478, 6.707-9.676", "frequencyCaption": "train horn one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1860.wav", "onoffCaption": "cat meowing at 0.869-2.009, 4.238-5.265, 6.706-8.666", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1885.wav", "onoffCaption": "thump thud at 0.058-2.286", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1929.wav", "onoffCaption": "train horn at 1.408-3.563, 4.706-6.861", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3019.wav", "onoffCaption": "tapping clicking clanking at 1.778-5.218, 6.534-9.974", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3102.wav", "onoffCaption": "dog barking at 2.444-4.444 and tapping clicking clanking at 5.38-8.82 and thump thud at 6.34-8.568", "frequencyCaption": "dog barking one times and tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3150.wav", "onoffCaption": "door knocking at 0.459-2.922 and sneeze at 7.356-9.419", "frequencyCaption": "door knocking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3238.wav", "onoffCaption": "whistling at 0.15-9.815 and sheep goat bleating at 1.787-3.787, 4.729-6.729", "frequencyCaption": "whistling one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3323.wav", "onoffCaption": "spraying at 0.067-0.817", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_3324.wav", "onoffCaption": "sneeze at 2.984-7.04", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3371.wav", "onoffCaption": "duck quacking at 2.226-4.226, 5.912-7.912", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3394.wav", "onoffCaption": "explosion at 2.848-6.401, 7.091-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3506.wav", "onoffCaption": "whistling at 0.097-7.847 and door knocking at 0.498-3.266, 4.128-6.508, 7.796-10.0", "frequencyCaption": "whistling one times and door knocking three times"}
+{"filepath": "data/multi_event_train/syn_3554.wav", "onoffCaption": "train horn at 0.045-4.113, 4.781-7.661", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3669.wav", "onoffCaption": "burping belching at 0.131-2.892, 4.13-6.891, 7.797-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3727.wav", "onoffCaption": "sneeze at 1.669-3.665", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3772.wav", "onoffCaption": "door knocking at 2.434-5.501, 6.209-9.276", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3775.wav", "onoffCaption": "car horn honking at 0.698-3.516 and sheep goat bleating at 4.98-6.98", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3790.wav", "onoffCaption": "cow mooing at 3.114-6.083, 7.337-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3797.wav", "onoffCaption": "duck quacking at 0.574-2.574, 3.148-5.148, 5.713-7.713 and dog barking at 2.459-4.459, 6.087-8.087", "frequencyCaption": "duck quacking three times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3802.wav", "onoffCaption": "whistling at 0.586-8.597 and sheep goat bleating at 1.539-3.539, 5.134-7.134", "frequencyCaption": "whistling one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3850.wav", "onoffCaption": "thump thud at 0.06-4.51, 6.619-8.847", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3919.wav", "onoffCaption": "sheep goat bleating at 0.38-2.38", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_68.wav", "onoffCaption": "spraying at 1.145-1.72, 2.499-3.074, 3.755-4.33 and tapping clicking clanking at 1.944-5.384, 6.002-8.278", "frequencyCaption": "spraying three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_72.wav", "onoffCaption": "spraying at 2.92-3.547, 4.121-5.378, 6.063-8.191", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_97.wav", "onoffCaption": "thump thud at 0.697-4.364, 5.526-7.754", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_100.wav", "onoffCaption": "door knocking at 2.269-5.389 and explosion at 7.884-10.0", "frequencyCaption": "door knocking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_272.wav", "onoffCaption": "burping belching at 0.523-4.029, 6.193-9.699", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_297.wav", "onoffCaption": "tapping clicking clanking at 1.345-4.785, 6.911-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_369.wav", "onoffCaption": "burping belching at 0.269-2.367", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_396.wav", "onoffCaption": "cow mooing at 0.073-3.055, 5.173-8.042", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_457.wav", "onoffCaption": "explosion at 1.716-4.903 and door knocking at 3.096-5.16, 6.335-8.399", "frequencyCaption": "explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_556.wav", "onoffCaption": "sneeze at 0.228-3.436, 4.412-5.658 and door slamming at 1.53-4.504, 6.22-9.194", "frequencyCaption": "sneeze two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_624.wav", "onoffCaption": "sneeze at 0.49-2.449 and thump thud at 5.007-9.457", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_848.wav", "onoffCaption": "thump thud at 3.74-6.24, 7.599-9.797", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_852.wav", "onoffCaption": "tapping clicking clanking at 1.814-5.254, 6.511-9.021", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_949.wav", "onoffCaption": "spraying at 2.482-4.215, 6.71-7.794, 8.844-9.445", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_953.wav", "onoffCaption": "car horn honking at 3.959-6.872", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1066.wav", "onoffCaption": "explosion at 2.375-4.695, 5.778-8.098 and woman laughing at 4.233-6.349, 7.227-9.343", "frequencyCaption": "explosion two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1083.wav", "onoffCaption": "woman laughing at 2.835-5.043, 7.399-9.607", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1099.wav", "onoffCaption": "burping belching at 3.336-6.515", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1167.wav", "onoffCaption": "burping belching at 0.344-3.344, 5.134-7.499", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1182.wav", "onoffCaption": "spraying at 1.885-4.277, 5.008-5.508, 7.252-9.014", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1198.wav", "onoffCaption": "door knocking at 1.186-6.019, 7.123-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1215.wav", "onoffCaption": "tapping clicking clanking at 0.347-3.787, 5.034-8.474", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1314.wav", "onoffCaption": "thump thud at 3.539-7.989", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1430.wav", "onoffCaption": "tapping clicking clanking at 2.991-6.431, 7.373-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1531.wav", "onoffCaption": "cat meowing at 0.302-2.296, 3.126-5.12 and car horn honking at 1.986-5.481", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1643.wav", "onoffCaption": "thump thud at 2.695-7.07", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1659.wav", "onoffCaption": "explosion at 3.679-6.551, 7.448-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1742.wav", "onoffCaption": "tapping clicking clanking at 0.157-3.597, 5.964-8.192 and gunshot at 2.244-4.374, 4.978-6.978, 7.529-9.699", "frequencyCaption": "tapping clicking clanking two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1758.wav", "onoffCaption": "burping belching at 0.668-3.191, 4.416-7.208 and car horn honking at 1.278-4.191", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1835.wav", "onoffCaption": "woman laughing at 2.231-4.869, 6.638-9.276", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1934.wav", "onoffCaption": "thump thud at 0.624-4.291 and explosion at 7.861-10.0", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3004.wav", "onoffCaption": "duck quacking at 0.12-2.12, 3.932-5.932 and door slamming at 8.291-9.191", "frequencyCaption": "duck quacking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3105.wav", "onoffCaption": "thump thud at 2.398-4.898, 6.624-9.124", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3277.wav", "onoffCaption": "woman laughing at 0.86-3.143, 4.265-6.548, 7.463-9.746", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_3288.wav", "onoffCaption": "sneeze at 0.323-1.415, 2.705-3.797, 5.231-6.323 and train horn at 2.628-5.068", "frequencyCaption": "sneeze three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3292.wav", "onoffCaption": "burping belching at 1.879-5.879", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3376.wav", "onoffCaption": "train horn at 0.694-4.094, 5.406-8.061", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3389.wav", "onoffCaption": "thump thud at 0.609-5.059", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3393.wav", "onoffCaption": "door slamming at 0.336-1.509", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3448.wav", "onoffCaption": "gunshot at 2.818-4.818 and door knocking at 4.83-6.99, 7.814-9.974", "frequencyCaption": "gunshot one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3452.wav", "onoffCaption": "spraying at 3.48-4.002, 5.912-6.434", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3549.wav", "onoffCaption": "cow mooing at 1.512-4.522 and sheep goat bleating at 6.085-9.381", "frequencyCaption": "cow mooing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3553.wav", "onoffCaption": "explosion at 0.046-3.102, 4.365-7.421 and cow mooing at 1.437-6.417", "frequencyCaption": "explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3621.wav", "onoffCaption": "duck quacking at 0.251-2.251, 3.603-5.603 and door slamming at 3.149-4.914, 5.503-6.503", "frequencyCaption": "duck quacking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3720.wav", "onoffCaption": "gunshot at 0.119-2.119, 3.389-5.389, 6.323-8.323", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3857.wav", "onoffCaption": "car horn honking at 1.094-4.02, 4.596-7.522", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3956.wav", "onoffCaption": "car horn honking at 3.505-8.412", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_27.wav", "onoffCaption": "sheep goat bleating at 3.479-5.479, 6.791-8.791 and cow mooing at 3.583-8.012", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_56.wav", "onoffCaption": "duck quacking at 0.787-2.787 and cat meowing at 1.168-2.18, 3.484-4.496", "frequencyCaption": "duck quacking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_124.wav", "onoffCaption": "explosion at 0.32-2.413 and dog barking at 4.449-6.449, 7.727-9.727 and tapping clicking clanking at 4.578-8.018", "frequencyCaption": "explosion one times and dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_155.wav", "onoffCaption": "sheep goat bleating at 3.874-6.954", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_268.wav", "onoffCaption": "burping belching at 2.076-7.076, 7.833-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_326.wav", "onoffCaption": "sneeze at 0.217-1.32 and duck quacking at 3.736-5.736", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_373.wav", "onoffCaption": "door knocking at 2.926-5.694, 6.774-9.493", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_418.wav", "onoffCaption": "train horn at 0.085-3.619", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_469.wav", "onoffCaption": "train horn at 2.048-5.808, 6.803-9.524", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_503.wav", "onoffCaption": "door slamming at 2.492-4.975 and explosion at 2.856-5.117, 6.1-8.361", "frequencyCaption": "door slamming one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_597.wav", "onoffCaption": "dog barking at 2.154-4.154 and car horn honking at 6.717-9.643", "frequencyCaption": "dog barking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_725.wav", "onoffCaption": "car horn honking at 0.486-3.661, 4.479-6.592", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_770.wav", "onoffCaption": "explosion at 0.494-2.668, 3.329-5.503 and gunshot at 2.729-4.859 and spraying at 9.35-10.0", "frequencyCaption": "explosion two times and gunshot one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_795.wav", "onoffCaption": "woman laughing at 0.279-2.36, 4.524-6.605, 7.54-9.621", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_807.wav", "onoffCaption": "dog barking at 0.412-2.412 and sneeze at 3.819-5.347, 6.581-7.967", "frequencyCaption": "dog barking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_876.wav", "onoffCaption": "whistling at 2.475-4.704, 5.69-7.699 and train horn at 6.152-8.952", "frequencyCaption": "whistling two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_893.wav", "onoffCaption": "explosion at 0.254-3.254 and sneeze at 5.264-7.327", "frequencyCaption": "explosion one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_988.wav", "onoffCaption": "sheep goat bleating at 1.533-3.533 and cat meowing at 6.128-7.488", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1033.wav", "onoffCaption": "gunshot at 1.931-4.437", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1042.wav", "onoffCaption": "whistling at 1.292-9.042", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1128.wav", "onoffCaption": "sheep goat bleating at 1.761-3.761, 5.687-7.687", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1159.wav", "onoffCaption": "burping belching at 0.102-2.517 and door slamming at 0.169-3.048", "frequencyCaption": "burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1240.wav", "onoffCaption": "car horn honking at 3.031-6.618 and spraying at 7.178-8.435", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1414.wav", "onoffCaption": "car horn honking at 2.864-5.79, 7.277-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1465.wav", "onoffCaption": "door knocking at 0.143-3.21, 4.674-7.054", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1480.wav", "onoffCaption": "train horn at 3.285-6.445", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1616.wav", "onoffCaption": "sheep goat bleating at 1.724-3.724, 4.694-6.694, 7.757-9.757", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1910.wav", "onoffCaption": "gunshot at 3.62-5.62, 7.01-9.01", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1961.wav", "onoffCaption": "explosion at 0.83-5.83 and car horn honking at 2.125-6.525, 7.843-10.0", "frequencyCaption": "explosion one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1984.wav", "onoffCaption": "explosion at 0.271-3.271 and spraying at 0.606-3.042, 4.649-7.085 and gunshot at 4.573-6.592", "frequencyCaption": "explosion one times and spraying two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3020.wav", "onoffCaption": "thump thud at 0.274-2.613, 4.736-7.075 and door slamming at 1.616-2.421, 4.032-5.335 and cat meowing at 4.18-7.421", "frequencyCaption": "thump thud two times and door slamming two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3051.wav", "onoffCaption": "explosion at 0.583-5.583, 7.425-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3222.wav", "onoffCaption": "whistling at 2.234-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3339.wav", "onoffCaption": "sneeze at 3.7-5.659 and car horn honking at 7.905-10.0", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3407.wav", "onoffCaption": "cat meowing at 0.704-1.716, 3.124-5.064 and spraying at 1.93-4.322 and woman laughing at 7.89-10.0", "frequencyCaption": "cat meowing two times and spraying one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3476.wav", "onoffCaption": "sneeze at 0.2-1.294, 2.325-4.449", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3493.wav", "onoffCaption": "explosion at 2.754-5.015", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3588.wav", "onoffCaption": "door knocking at 3.157-5.727, 7.114-9.684", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3674.wav", "onoffCaption": "burping belching at 0.178-3.738, 4.276-7.836 and duck quacking at 7.421-9.421", "frequencyCaption": "burping belching two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3691.wav", "onoffCaption": "train horn at 2.007-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3818.wav", "onoffCaption": "train horn at 0.086-2.966, 5.129-7.598", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3869.wav", "onoffCaption": "spraying at 0.068-1.152, 3.159-3.734", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3903.wav", "onoffCaption": "gunshot at 0.325-2.325, 3.275-5.275, 6.343-8.343", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3972.wav", "onoffCaption": "explosion at 2.923-4.93 and cat meowing at 7.772-9.308", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3997.wav", "onoffCaption": "spraying at 0.219-1.086, 2.175-3.042 and gunshot at 1.073-3.073, 4.828-6.828", "frequencyCaption": "spraying two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_171.wav", "onoffCaption": "sheep goat bleating at 0.622-2.622, 3.342-5.342 and explosion at 0.656-5.577, 6.689-10.0", "frequencyCaption": "sheep goat bleating two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_194.wav", "onoffCaption": "cow mooing at 4.267-9.247", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_203.wav", "onoffCaption": "spraying at 0.497-2.192, 2.836-3.777, 4.996-5.847", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_219.wav", "onoffCaption": "burping belching at 0.188-3.188 and door knocking at 0.709-4.471", "frequencyCaption": "burping belching one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_302.wav", "onoffCaption": "spraying at 2.0-2.75", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_318.wav", "onoffCaption": "door knocking at 3.232-6.079, 7.926-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_357.wav", "onoffCaption": "car horn honking at 0.444-3.939", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_527.wav", "onoffCaption": "explosion at 0.505-3.373 and tapping clicking clanking at 6.208-9.648", "frequencyCaption": "explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_572.wav", "onoffCaption": "whistling at 1.684-3.693, 5.499-7.606", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_701.wav", "onoffCaption": "door slamming at 0.027-0.967, 2.312-3.252 and sneeze at 6.756-8.256", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_754.wav", "onoffCaption": "dog barking at 2.725-4.725, 5.904-7.904", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_823.wav", "onoffCaption": "whistling at 0.976-6.151, 7.109-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_938.wav", "onoffCaption": "cat meowing at 0.419-3.323, 4.149-6.126, 7.37-9.364", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1017.wav", "onoffCaption": "sneeze at 1.526-3.765 and dog barking at 3.672-5.672, 6.695-8.695", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1231.wav", "onoffCaption": "door knocking at 0.655-3.03, 4.154-6.529, 7.337-9.712", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1264.wav", "onoffCaption": "sneeze at 2.384-3.912, 4.476-5.802, 6.605-9.066", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1281.wav", "onoffCaption": "burping belching at 0.058-5.058, 7.551-9.892 and door slamming at 1.441-3.637", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1365.wav", "onoffCaption": "train horn at 3.067-9.536", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1380.wav", "onoffCaption": "thump thud at 1.007-3.507, 5.962-8.301", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1441.wav", "onoffCaption": "thump thud at 1.625-4.125, 5.821-7.956", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1628.wav", "onoffCaption": "spraying at 1.654-3.387", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1632.wav", "onoffCaption": "explosion at 0.414-3.142", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1667.wav", "onoffCaption": "train horn at 0.181-5.258, 7.479-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1682.wav", "onoffCaption": "train horn at 3.047-5.202, 5.893-8.048", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1729.wav", "onoffCaption": "burping belching at 2.765-4.968, 6.119-8.322", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_1733.wav", "onoffCaption": "spraying at 2.841-4.098, 6.577-8.339", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1799.wav", "onoffCaption": "cat meowing at 0.443-2.027, 4.267-5.851", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1945.wav", "onoffCaption": "dog barking at 2.352-4.352 and gunshot at 7.525-9.525", "frequencyCaption": "dog barking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3075.wav", "onoffCaption": "gunshot at 0.101-2.101", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3090.wav", "onoffCaption": "sheep goat bleating at 0.67-2.67, 3.618-5.618 and gunshot at 2.733-4.752, 6.302-8.321", "frequencyCaption": "sheep goat bleating two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3206.wav", "onoffCaption": "car horn honking at 0.644-3.109, 5.341-7.806", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3253.wav", "onoffCaption": "cat meowing at 0.657-2.017", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3307.wav", "onoffCaption": "burping belching at 2.496-6.04, 6.739-9.063 and woman laughing at 3.176-5.768", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3348.wav", "onoffCaption": "sheep goat bleating at 1.975-3.975 and burping belching at 2.183-4.944, 5.965-8.088", "frequencyCaption": "sheep goat bleating one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3423.wav", "onoffCaption": "thump thud at 2.28-4.508, 6.92-9.148", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3538.wav", "onoffCaption": "duck quacking at 2.278-4.278, 5.652-7.652", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3605.wav", "onoffCaption": "dog barking at 0.121-2.121", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3650.wav", "onoffCaption": "whistling at 0.406-2.415, 3.048-5.057, 6.24-8.249 and spraying at 4.176-4.826, 5.812-6.876", "frequencyCaption": "whistling three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3751.wav", "onoffCaption": "gunshot at 0.272-2.272, 4.238-6.331, 7.459-9.459", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_3927.wav", "onoffCaption": "door slamming at 0.028-2.028", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_19.wav", "onoffCaption": "gunshot at 2.917-4.917", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_256.wav", "onoffCaption": "sheep goat bleating at 1.495-3.495, 4.567-6.567, 7.123-9.123", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_426.wav", "onoffCaption": "woman laughing at 2.649-4.932, 5.692-7.975", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_432.wav", "onoffCaption": "door knocking at 0.055-3.43 and spraying at 7.599-8.2", "frequencyCaption": "door knocking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_473.wav", "onoffCaption": "gunshot at 0.698-2.698, 3.947-5.947, 6.742-8.742 and explosion at 4.807-7.863", "frequencyCaption": "gunshot three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_496.wav", "onoffCaption": "car horn honking at 2.094-5.313, 5.949-9.168", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_529.wav", "onoffCaption": "door knocking at 3.432-5.592", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_568.wav", "onoffCaption": "dog barking at 0.403-2.403, 4.637-6.637", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_600.wav", "onoffCaption": "train horn at 2.315-8.03", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_641.wav", "onoffCaption": "cow mooing at 0.533-3.515, 4.376-7.358 and duck quacking at 3.39-5.39 and whistling at 5.257-8.232", "frequencyCaption": "cow mooing two times and duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_655.wav", "onoffCaption": "spraying at 0.001-0.852", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_839.wav", "onoffCaption": "door slamming at 2.325-3.442, 4.75-5.867, 6.927-8.044 and sneeze at 3.071-4.359, 4.959-6.247, 7.169-8.457", "frequencyCaption": "door slamming three times and sneeze three times"}
+{"filepath": "data/multi_event_train/syn_889.wav", "onoffCaption": "sheep goat bleating at 3.745-5.745, 7.04-9.04", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_922.wav", "onoffCaption": "door knocking at 2.749-5.596, 6.697-9.593", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_936.wav", "onoffCaption": "gunshot at 2.615-4.656, 5.659-7.7", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_977.wav", "onoffCaption": "car horn honking at 0.024-3.199, 4.429-7.604 and door slamming at 1.132-1.983, 2.772-3.623, 5.279-6.13 and spraying at 3.137-3.918, 5.76-6.541", "frequencyCaption": "car horn honking two times and door slamming three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_992.wav", "onoffCaption": "train horn at 0.88-6.595, 7.622-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1019.wav", "onoffCaption": "door slamming at 0.69-3.664 and duck quacking at 7.922-9.922", "frequencyCaption": "door slamming one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1058.wav", "onoffCaption": "door slamming at 0.388-2.388, 3.626-5.626 and explosion at 2.533-5.286, 6.989-9.497", "frequencyCaption": "door slamming two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1102.wav", "onoffCaption": "gunshot at 3.478-5.608, 6.53-8.77", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1116.wav", "onoffCaption": "door slamming at 2.112-4.24, 5.163-7.291", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1143.wav", "onoffCaption": "sneeze at 2.226-4.185, 5.315-7.274, 7.89-9.849", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_1330.wav", "onoffCaption": "burping belching at 3.154-6.154", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1515.wav", "onoffCaption": "spraying at 0.068-0.669 and woman laughing at 2.812-4.904, 6.245-8.327", "frequencyCaption": "spraying one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1540.wav", "onoffCaption": "sheep goat bleating at 1.643-3.643, 4.39-6.39, 7.124-9.124", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_1554.wav", "onoffCaption": "cat meowing at 2.76-3.781 and door slamming at 5.94-7.079", "frequencyCaption": "cat meowing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1698.wav", "onoffCaption": "spraying at 3.088-5.216", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_1766.wav", "onoffCaption": "dog barking at 2.815-4.815, 5.782-8.22", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1783.wav", "onoffCaption": "door knocking at 2.091-5.644, 6.754-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1811.wav", "onoffCaption": "duck quacking at 0.315-2.315, 3.397-5.397 and dog barking at 0.403-2.403, 3.261-5.261, 6.58-8.58", "frequencyCaption": "duck quacking two times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1844.wav", "onoffCaption": "train horn at 2.463-6.903", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1850.wav", "onoffCaption": "sheep goat bleating at 0.308-2.308, 3.201-5.201, 7.564-10.0", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3121.wav", "onoffCaption": "burping belching at 0.675-4.698, 6.615-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3160.wav", "onoffCaption": "burping belching at 0.221-2.344, 4.641-6.672, 7.976-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_3174.wav", "onoffCaption": "cow mooing at 0.128-3.138, 4.765-7.734 and gunshot at 0.227-2.227, 2.757-4.85, 6.127-8.127 and tapping clicking clanking at 5.439-8.879", "frequencyCaption": "cow mooing two times and gunshot three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3185.wav", "onoffCaption": "gunshot at 1.368-3.368, 4.358-6.358", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3191.wav", "onoffCaption": "explosion at 3.122-6.122", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3249.wav", "onoffCaption": "car horn honking at 0.422-2.887, 3.961-6.426", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3352.wav", "onoffCaption": "dog barking at 0.73-4.05, 6.487-9.807", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3439.wav", "onoffCaption": "sneeze at 0.638-5.638, 6.306-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3489.wav", "onoffCaption": "burping belching at 0.191-2.952 and car horn honking at 6.918-10.0", "frequencyCaption": "burping belching one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3522.wav", "onoffCaption": "tapping clicking clanking at 2.936-6.376 and dog barking at 3.613-5.613 and spraying at 4.705-6.467, 7.598-9.293", "frequencyCaption": "tapping clicking clanking one times and dog barking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3536.wav", "onoffCaption": "gunshot at 1.616-3.616, 5.17-7.17", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3577.wav", "onoffCaption": "car horn honking at 0.724-5.631, 6.312-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3592.wav", "onoffCaption": "sheep goat bleating at 0.138-3.218, 3.865-5.865 and car horn honking at 2.443-6.765", "frequencyCaption": "sheep goat bleating two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3704.wav", "onoffCaption": "sneeze at 1.924-3.027, 4.925-6.028 and cow mooing at 7.95-10.0", "frequencyCaption": "sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3826.wav", "onoffCaption": "spraying at 0.306-2.001 and cat meowing at 6.478-10.0", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3832.wav", "onoffCaption": "burping belching at 0.032-3.032 and explosion at 5.43-10.0", "frequencyCaption": "burping belching one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_3873.wav", "onoffCaption": "thump thud at 0.405-4.855, 5.664-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3896.wav", "onoffCaption": "gunshot at 0.419-2.419, 3.141-5.141, 6.145-8.145 and burping belching at 1.956-5.979", "frequencyCaption": "gunshot three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3929.wav", "onoffCaption": "burping belching at 2.342-4.377, 5.498-7.533", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3968.wav", "onoffCaption": "thump thud at 1.08-5.53, 7.974-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_58.wav", "onoffCaption": "tapping clicking clanking at 2.24-5.68, 6.942-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_217.wav", "onoffCaption": "duck quacking at 1.061-3.061, 4.757-6.757, 7.312-9.312", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_242.wav", "onoffCaption": "woman laughing at 0.008-2.233, 3.414-5.519, 7.665-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_258.wav", "onoffCaption": "door knocking at 2.89-6.506, 7.322-9.482", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_343.wav", "onoffCaption": "duck quacking at 3.426-5.426, 6.592-8.592", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_359.wav", "onoffCaption": "cow mooing at 0.104-3.086 and woman laughing at 5.605-10.0", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_467.wav", "onoffCaption": "thump thud at 1.125-3.896, 5.7-8.471", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_482.wav", "onoffCaption": "door knocking at 2.099-4.723, 6.801-9.425", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_566.wav", "onoffCaption": "dog barking at 2.388-4.388, 6.567-8.567", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_599.wav", "onoffCaption": "whistling at 0.26-5.435, 6.002-8.011", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_614.wav", "onoffCaption": "woman laughing at 0.863-2.968", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_715.wav", "onoffCaption": "woman laughing at 2.516-6.568", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_878.wav", "onoffCaption": "spraying at 0.47-0.992 and whistling at 2.956-5.831", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_963.wav", "onoffCaption": "cow mooing at 0.407-3.705, 4.406-7.704", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_986.wav", "onoffCaption": "duck quacking at 2.219-4.219, 5.442-7.442", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1157.wav", "onoffCaption": "car horn honking at 1.378-4.553", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1225.wav", "onoffCaption": "train horn at 2.208-5.978", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1324.wav", "onoffCaption": "duck quacking at 0.456-2.456, 3.38-5.38, 6.356-8.356 and thump thud at 2.341-4.68, 5.337-7.676", "frequencyCaption": "duck quacking three times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1371.wav", "onoffCaption": "gunshot at 0.281-2.411 and whistling at 4.833-9.317", "frequencyCaption": "gunshot one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1394.wav", "onoffCaption": "sheep goat bleating at 0.453-2.453, 3.709-5.709", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1501.wav", "onoffCaption": "door slamming at 3.171-5.629", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_1669.wav", "onoffCaption": "door slamming at 0.076-0.881, 1.714-2.519 and car horn honking at 5.815-8.28", "frequencyCaption": "door slamming two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1673.wav", "onoffCaption": "train horn at 2.681-5.881, 7.622-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1696.wav", "onoffCaption": "woman laughing at 2.735-9.469", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1727.wav", "onoffCaption": "woman laughing at 2.101-4.295, 5.621-7.848 and whistling at 2.263-5.138", "frequencyCaption": "woman laughing two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_1768.wav", "onoffCaption": "train horn at 0.389-2.789", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1772.wav", "onoffCaption": "cow mooing at 3.636-6.934", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1797.wav", "onoffCaption": "door slamming at 0.328-2.456, 3.815-5.943", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1805.wav", "onoffCaption": "gunshot at 3.296-5.296", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3135.wav", "onoffCaption": "car horn honking at 2.114-6.436, 7.636-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3208.wav", "onoffCaption": "door slamming at 3.143-4.121, 5.763-6.741, 7.247-8.225", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3247.wav", "onoffCaption": "spraying at 1.987-2.487, 4.039-4.539 and door slamming at 7.66-9.425", "frequencyCaption": "spraying two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3313.wav", "onoffCaption": "cow mooing at 0.136-4.565, 6.674-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3346.wav", "onoffCaption": "tapping clicking clanking at 2.996-6.436, 7.718-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3478.wav", "onoffCaption": "door slamming at 2.783-5.757, 6.792-8.988", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_3563.wav", "onoffCaption": "explosion at 0.035-2.763, 3.269-5.997", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_3586.wav", "onoffCaption": "dog barking at 2.001-4.001, 6.436-8.436", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_3611.wav", "onoffCaption": "burping belching at 0.545-3.545", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3710.wav", "onoffCaption": "sheep goat bleating at 0.418-2.418, 3.248-5.248, 6.327-8.327", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3745.wav", "onoffCaption": "whistling at 2.291-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3867.wav", "onoffCaption": "burping belching at 1.62-4.381, 6.604-9.365", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3882.wav", "onoffCaption": "woman laughing at 0.645-3.699 and sneeze at 6.075-7.167, 7.708-8.8", "frequencyCaption": "woman laughing one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3966.wav", "onoffCaption": "thump thud at 0.699-3.746, 6.205-9.122", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3999.wav", "onoffCaption": "tapping clicking clanking at 2.904-6.344, 7.265-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_17.wav", "onoffCaption": "cow mooing at 0.467-5.447, 6.568-8.92", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_42.wav", "onoffCaption": "sheep goat bleating at 3.727-5.727", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_130.wav", "onoffCaption": "thump thud at 0.503-3.274, 5.351-8.122 and cow mooing at 2.094-6.523", "frequencyCaption": "thump thud two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_165.wav", "onoffCaption": "explosion at 0.033-3.033, 4.03-6.902", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_180.wav", "onoffCaption": "door knocking at 1.887-4.199, 5.442-7.754", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_316.wav", "onoffCaption": "cow mooing at 0.186-4.615 and gunshot at 6.965-8.965", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_498.wav", "onoffCaption": "woman laughing at 0.389-3.184 and spraying at 0.479-1.736 and sheep goat bleating at 6.814-8.814", "frequencyCaption": "woman laughing one times and spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_583.wav", "onoffCaption": "cat meowing at 2.992-4.609, 5.325-6.942, 8.008-9.625", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_740.wav", "onoffCaption": "woman laughing at 0.233-2.933, 4.001-6.701, 7.57-10.0", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_837.wav", "onoffCaption": "tapping clicking clanking at 1.154-4.594", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_862.wav", "onoffCaption": "cow mooing at 0.176-3.158", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_887.wav", "onoffCaption": "gunshot at 1.975-3.975 and explosion at 6.851-8.853", "frequencyCaption": "gunshot one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_979.wav", "onoffCaption": "cow mooing at 0.418-5.398, 6.857-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1003.wav", "onoffCaption": "door knocking at 2.873-7.575", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1056.wav", "onoffCaption": "spraying at 0.772-3.791, 4.499-5.126, 6.351-7.432 and door knocking at 2.666-6.816", "frequencyCaption": "spraying three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1118.wav", "onoffCaption": "sneeze at 1.98-4.219, 5.391-8.105", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1270.wav", "onoffCaption": "explosion at 0.304-3.057, 5.349-7.61", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1295.wav", "onoffCaption": "tapping clicking clanking at 0.66-4.1, 5.72-8.425", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1400.wav", "onoffCaption": "cat meowing at 0.025-1.046, 1.59-4.35", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1455.wav", "onoffCaption": "tapping clicking clanking at 0.635-4.075, 5.126-8.566 and car horn honking at 1.938-6.338", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1626.wav", "onoffCaption": "tapping clicking clanking at 2.667-6.107, 7.915-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1904.wav", "onoffCaption": "gunshot at 0.344-2.344, 3.876-5.876", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1951.wav", "onoffCaption": "thump thud at 0.001-2.501, 3.395-5.623", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3034.wav", "onoffCaption": "thump thud at 2.608-5.108, 6.01-8.51", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3061.wav", "onoffCaption": "cow mooing at 1.872-4.841", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3084.wav", "onoffCaption": "whistling at 0.882-3.757, 4.534-6.763 and cat meowing at 1.107-2.651, 3.757-5.023", "frequencyCaption": "whistling two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3212.wav", "onoffCaption": "explosion at 0.067-2.131", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3309.wav", "onoffCaption": "train horn at 0.142-4.142 and duck quacking at 1.183-3.183, 4.552-6.552 and door knocking at 2.065-4.377, 5.166-7.478", "frequencyCaption": "train horn one times and duck quacking two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3437.wav", "onoffCaption": "train horn at 3.389-7.457", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_3462.wav", "onoffCaption": "cat meowing at 1.262-4.503, 6.9-10.0 and spraying at 1.386-3.846 and duck quacking at 2.352-4.352", "frequencyCaption": "cat meowing two times and spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3487.wav", "onoffCaption": "cow mooing at 0.381-3.391, 4.246-7.074 and car horn honking at 2.377-6.889", "frequencyCaption": "cow mooing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3579.wav", "onoffCaption": "door knocking at 0.011-4.544, 5.239-9.772", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3644.wav", "onoffCaption": "dog barking at 1.946-3.946, 4.723-6.723 and sneeze at 4.503-6.906, 7.706-10.0", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3898.wav", "onoffCaption": "explosion at 0.238-3.791", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3933.wav", "onoffCaption": "door slamming at 0.57-3.449, 4.57-6.085, 6.734-7.572", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3983.wav", "onoffCaption": "sneeze at 1.528-3.652, 4.23-5.775, 6.779-9.439", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_66.wav", "onoffCaption": "sheep goat bleating at 3.755-5.755, 6.54-8.54", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_83.wav", "onoffCaption": "burping belching at 1.755-4.957, 6.169-8.93", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_114.wav", "onoffCaption": "thump thud at 0.273-3.32, 4.785-7.832 and whistling at 2.289-7.464", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_299.wav", "onoffCaption": "tapping clicking clanking at 2.854-6.294", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_367.wav", "onoffCaption": "burping belching at 2.702-5.992", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_382.wav", "onoffCaption": "tapping clicking clanking at 3.22-6.66, 7.835-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_428.wav", "onoffCaption": "door knocking at 3.238-5.426, 6.557-8.745", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_459.wav", "onoffCaption": "door slamming at 0.017-1.27 and train horn at 0.31-4.31, 5.293-7.405 and duck quacking at 2.271-4.271, 5.79-7.79", "frequencyCaption": "door slamming one times and train horn two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_533.wav", "onoffCaption": "train horn at 2.06-4.527, 5.17-7.637", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_542.wav", "onoffCaption": "door knocking at 2.58-4.932, 5.435-8.165", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_731.wav", "onoffCaption": "cow mooing at 1.42-4.718", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_781.wav", "onoffCaption": "cat meowing at 2.711-3.924, 4.973-6.186, 7.019-8.232", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_846.wav", "onoffCaption": "cow mooing at 2.238-6.667, 7.302-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1072.wav", "onoffCaption": "train horn at 1.178-7.238", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1097.wav", "onoffCaption": "whistling at 0.339-4.823, 6.015-8.024", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1169.wav", "onoffCaption": "sneeze at 0.092-1.186, 1.806-2.9, 3.554-4.648 and cat meowing at 6.817-8.765", "frequencyCaption": "sneeze three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1201.wav", "onoffCaption": "train horn at 2.979-5.116, 5.771-7.908", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1254.wav", "onoffCaption": "sheep goat bleating at 0.196-2.196, 3.225-5.225 and cow mooing at 7.435-10.0", "frequencyCaption": "sheep goat bleating two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1424.wav", "onoffCaption": "gunshot at 0.785-2.785, 3.691-5.691, 6.961-8.961", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_1602.wav", "onoffCaption": "explosion at 2.752-4.926, 6.126-8.3", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1657.wav", "onoffCaption": "burping belching at 1.032-4.211 and spraying at 7.591-9.077", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1719.wav", "onoffCaption": "car horn honking at 2.991-5.917, 7.165-9.757", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1920.wav", "onoffCaption": "cat meowing at 2.121-5.151, 6.69-8.226", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3010.wav", "onoffCaption": "car horn honking at 0.242-3.896, 5.279-8.933", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3236.wav", "onoffCaption": "burping belching at 1.875-5.875", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3263.wav", "onoffCaption": "dog barking at 3.569-5.569, 6.486-8.486 and thump thud at 4.01-7.928", "frequencyCaption": "dog barking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3286.wav", "onoffCaption": "gunshot at 3.647-5.887, 7.673-9.913", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3378.wav", "onoffCaption": "door knocking at 0.404-3.172, 5.052-7.899 and tapping clicking clanking at 3.784-7.224 and car horn honking at 6.433-8.433", "frequencyCaption": "door knocking two times and tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3446.wav", "onoffCaption": "sheep goat bleating at 0.347-3.987, 5.155-8.795", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_3635.wav", "onoffCaption": "cat meowing at 0.466-2.01 and duck quacking at 5.427-7.427", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3660.wav", "onoffCaption": "tapping clicking clanking at 2.607-6.047, 6.693-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3685.wav", "onoffCaption": "gunshot at 3.213-5.719", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_3828.wav", "onoffCaption": "dog barking at 3.288-5.288 and burping belching at 3.369-5.572, 7.868-10.0", "frequencyCaption": "dog barking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3859.wav", "onoffCaption": "spraying at 0.065-1.065, 2.246-3.246, 5.682-6.682", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_3942.wav", "onoffCaption": "door knocking at 0.673-2.922, 4.302-6.551 and cow mooing at 3.207-6.189", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_29.wav", "onoffCaption": "tapping clicking clanking at 2.876-6.316, 7.077-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_33.wav", "onoffCaption": "sheep goat bleating at 0.232-2.232, 3.068-5.068, 5.691-7.691", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_141.wav", "onoffCaption": "dog barking at 2.679-4.679, 5.988-7.988", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_229.wav", "onoffCaption": "cow mooing at 0.504-3.802 and sheep goat bleating at 6.873-10.0", "frequencyCaption": "cow mooing one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_332.wav", "onoffCaption": "explosion at 2.929-5.658, 6.447-9.447", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_416.wav", "onoffCaption": "cat meowing at 1.978-3.873, 4.787-6.682", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_517.wav", "onoffCaption": "woman laughing at 0.043-7.055 and sheep goat bleating at 1.034-3.034, 4.335-6.335", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_665.wav", "onoffCaption": "thump thud at 0.333-3.104, 4.26-6.811", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_764.wav", "onoffCaption": "explosion at 2.603-5.332", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_809.wav", "onoffCaption": "tapping clicking clanking at 0.597-4.037, 5.572-8.29", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_813.wav", "onoffCaption": "sneeze at 0.338-1.664, 3.859-6.726, 7.402-8.902", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_908.wav", "onoffCaption": "duck quacking at 2.231-4.231, 5.703-7.703", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_912.wav", "onoffCaption": "whistling at 0.37-5.87", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1027.wav", "onoffCaption": "burping belching at 0.405-2.928 and spraying at 7.113-8.808", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1126.wav", "onoffCaption": "cat meowing at 2.066-3.798, 4.339-6.071, 6.707-8.439", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1471.wav", "onoffCaption": "whistling at 2.484-5.459, 7.427-9.449", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1494.wav", "onoffCaption": "whistling at 0.825-3.7, 4.527-7.523", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1570.wav", "onoffCaption": "cow mooing at 0.068-3.078 and dog barking at 5.42-7.858", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1595.wav", "onoffCaption": "door slamming at 1.444-4.444, 6.63-9.63", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1874.wav", "onoffCaption": "gunshot at 0.023-2.023", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1891.wav", "onoffCaption": "burping belching at 0.954-3.157 and tapping clicking clanking at 1.277-4.717, 6.507-9.947", "frequencyCaption": "burping belching one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1975.wav", "onoffCaption": "explosion at 1.831-4.831, 5.888-8.888 and spraying at 2.557-3.307, 4.647-5.514 and tapping clicking clanking at 4.125-7.565", "frequencyCaption": "explosion two times and spraying two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_1990.wav", "onoffCaption": "door slamming at 0.013-1.186, 1.716-2.889, 3.808-4.981", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3045.wav", "onoffCaption": "explosion at 2.17-4.49 and train horn at 6.537-10.0", "frequencyCaption": "explosion one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_3144.wav", "onoffCaption": "gunshot at 0.205-2.205, 4.257-6.257", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3409.wav", "onoffCaption": "door knocking at 0.161-2.382, 3.18-5.984", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3413.wav", "onoffCaption": "explosion at 0.773-3.773, 4.786-7.786 and duck quacking at 1.458-3.458, 5.083-7.083", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3508.wav", "onoffCaption": "cow mooing at 2.491-5.789, 6.803-8.835", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_3512.wav", "onoffCaption": "explosion at 2.04-7.04 and thump thud at 2.667-7.042, 7.744-10.0", "frequencyCaption": "explosion one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3816.wav", "onoffCaption": "train horn at 3.343-6.143, 7.821-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3917.wav", "onoffCaption": "sneeze at 2.112-4.573", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_99.wav", "onoffCaption": "cow mooing at 0.562-3.86, 4.984-8.282", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_233.wav", "onoffCaption": "dog barking at 0.177-2.177 and woman laughing at 0.336-3.036, 4.006-6.644, 7.419-9.665", "frequencyCaption": "dog barking one times and woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_266.wav", "onoffCaption": "cow mooing at 1.671-4.681, 6.283-9.293 and woman laughing at 2.644-5.236, 5.829-7.91", "frequencyCaption": "cow mooing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_283.wav", "onoffCaption": "whistling at 0.42-3.295, 3.993-6.222, 7.259-9.502", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_328.wav", "onoffCaption": "whistling at 1.879-7.054 and tapping clicking clanking at 3.661-7.101, 7.666-9.803", "frequencyCaption": "whistling one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_398.wav", "onoffCaption": "sneeze at 2.719-4.832, 6.205-8.318", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_443.wav", "onoffCaption": "explosion at 0.593-5.593, 7.99-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_558.wav", "onoffCaption": "sheep goat bleating at 2.197-4.197, 4.74-6.74, 7.47-10.0 and door knocking at 2.662-6.215", "frequencyCaption": "sheep goat bleating three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_630.wav", "onoffCaption": "thump thud at 2.609-5.109, 6.484-8.984", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_662.wav", "onoffCaption": "gunshot at 1.884-3.884, 6.184-8.184", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_680.wav", "onoffCaption": "thump thud at 1.832-5.499, 6.258-9.925", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_779.wav", "onoffCaption": "thump thud at 2.493-4.832, 7.066-9.405", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_947.wav", "onoffCaption": "train horn at 0.341-2.821, 3.65-6.05", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1068.wav", "onoffCaption": "car horn honking at 0.181-4.43, 6.693-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1173.wav", "onoffCaption": "explosion at 0.221-2.314, 3.781-5.899", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_1196.wav", "onoffCaption": "sheep goat bleating at 1.568-3.568, 4.446-6.446", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1300.wav", "onoffCaption": "sneeze at 0.212-1.458, 3.742-4.988 and dog barking at 1.234-3.234", "frequencyCaption": "sneeze two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1355.wav", "onoffCaption": "cat meowing at 2.938-4.886, 6.198-8.146", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_1525.wav", "onoffCaption": "train horn at 0.469-3.109, 4.904-7.77", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_1618.wav", "onoffCaption": "cat meowing at 3.514-4.535, 6.027-7.24, 8.265-9.575", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_1703.wav", "onoffCaption": "burping belching at 2.136-7.737 and sheep goat bleating at 4.673-6.673, 7.794-9.794", "frequencyCaption": "burping belching one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1756.wav", "onoffCaption": "car horn honking at 1.746-4.211, 5.004-7.53", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1821.wav", "onoffCaption": "door knocking at 0.277-4.979, 6.493-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3111.wav", "onoffCaption": "thump thud at 1.355-5.805 and door slamming at 1.839-3.839, 4.949-6.862, 8.179-9.328", "frequencyCaption": "thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3279.wav", "onoffCaption": "dog barking at 0.184-3.105 and door knocking at 0.409-2.63 and cat meowing at 5.215-6.236", "frequencyCaption": "dog barking one times and door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3337.wav", "onoffCaption": "dog barking at 0.337-2.337", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3362.wav", "onoffCaption": "train horn at 0.443-2.58, 3.154-5.291, 7.352-9.489", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_3387.wav", "onoffCaption": "sheep goat bleating at 1.837-3.837, 4.859-6.859, 7.452-9.452 and car horn honking at 3.359-6.578", "frequencyCaption": "sheep goat bleating three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3547.wav", "onoffCaption": "gunshot at 3.96-6.053, 7.128-9.128", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_3734.wav", "onoffCaption": "train horn at 2.954-5.434, 7.906-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_3761.wav", "onoffCaption": "tapping clicking clanking at 1.41-4.85, 6.821-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3784.wav", "onoffCaption": "sneeze at 1.346-3.47", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_3843.wav", "onoffCaption": "woman laughing at 0.101-3.173 and spraying at 2.896-3.747", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3958.wav", "onoffCaption": "burping belching at 0.069-7.237", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_109.wav", "onoffCaption": "spraying at 3.713-4.297", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_113.wav", "onoffCaption": "thump thud at 2.654-6.572", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_234.wav", "onoffCaption": "explosion at 3.255-6.255", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_261.wav", "onoffCaption": "car horn honking at 0.263-2.776, 3.88-6.374, 7.098-9.563 and explosion at 0.45-2.452", "frequencyCaption": "car horn honking three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_284.wav", "onoffCaption": "woman laughing at 2.194-4.894, 6.241-8.941 and explosion at 2.707-7.707 and tapping clicking clanking at 3.378-6.818", "frequencyCaption": "woman laughing two times and explosion one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_411.wav", "onoffCaption": "door slamming at 2.769-4.002, 5.099-6.332", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_444.wav", "onoffCaption": "duck quacking at 0.65-2.65, 3.391-5.391, 6.599-8.599", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_545.wav", "onoffCaption": "whistling at 1.186-4.161, 5.356-8.315", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_637.wav", "onoffCaption": "door slamming at 2.817-4.817, 6.014-7.389", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_687.wav", "onoffCaption": "duck quacking at 0.504-2.504, 4.873-6.873 and sheep goat bleating at 5.253-7.253", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_915.wav", "onoffCaption": "cat meowing at 2.392-5.296, 6.243-9.147", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_940.wav", "onoffCaption": "cat meowing at 0.108-1.418, 3.144-4.715 and explosion at 2.253-4.427, 6.188-8.362", "frequencyCaption": "cat meowing two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_1075.wav", "onoffCaption": "door slamming at 3.377-4.668, 7.131-8.27", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_1121.wav", "onoffCaption": "tapping clicking clanking at 0.164-3.604, 4.325-6.915", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1174.wav", "onoffCaption": "burping belching at 2.91-9.59", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1191.wav", "onoffCaption": "dog barking at 0.073-2.073, 2.957-4.957", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1249.wav", "onoffCaption": "thump thud at 0.104-4.479, 5.297-8.068", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_1307.wav", "onoffCaption": "whistling at 3.185-6.16, 7.3-9.569", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_1352.wav", "onoffCaption": "sheep goat bleating at 0.42-2.42, 3.209-5.209, 5.845-7.845 and gunshot at 3.536-5.536", "frequencyCaption": "sheep goat bleating three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1423.wav", "onoffCaption": "door slamming at 2.046-4.409 and cow mooing at 7.179-10.0", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1439.wav", "onoffCaption": "thump thud at 1.856-6.306", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1489.wav", "onoffCaption": "door knocking at 0.208-3.976, 5.205-8.973", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1522.wav", "onoffCaption": "burping belching at 0.811-3.811, 5.905-8.428 and cat meowing at 0.94-4.285", "frequencyCaption": "burping belching two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1577.wav", "onoffCaption": "dog barking at 2.246-4.246", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1592.wav", "onoffCaption": "tapping clicking clanking at 1.029-4.469 and woman laughing at 6.167-9.448", "frequencyCaption": "tapping clicking clanking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_1704.wav", "onoffCaption": "door slamming at 3.323-4.576, 5.335-6.588 and cow mooing at 3.682-6.651", "frequencyCaption": "door slamming two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1751.wav", "onoffCaption": "spraying at 0.112-1.196, 1.83-2.914 and duck quacking at 1.325-3.325 and train horn at 1.764-4.438", "frequencyCaption": "spraying two times and duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_1826.wav", "onoffCaption": "explosion at 2.068-5.068 and cow mooing at 7.619-10.0", "frequencyCaption": "explosion one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_1873.wav", "onoffCaption": "spraying at 0.0-0.867, 1.694-2.561", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1896.wav", "onoffCaption": "cat meowing at 0.271-1.826 and duck quacking at 5.576-7.576", "frequencyCaption": "cat meowing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1968.wav", "onoffCaption": "duck quacking at 0.499-2.499, 4.923-6.923", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_3058.wav", "onoffCaption": "woman laughing at 0.114-6.848, 7.984-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3116.wav", "onoffCaption": "cat meowing at 0.504-1.516, 3.439-4.451, 5.303-6.315 and train horn at 0.902-4.102 and duck quacking at 2.916-4.916", "frequencyCaption": "cat meowing three times and train horn one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3143.wav", "onoffCaption": "duck quacking at 1.039-3.039, 4.428-6.428 and sneeze at 3.454-5.767, 7.844-10.0", "frequencyCaption": "duck quacking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_3330.wav", "onoffCaption": "thump thud at 1.915-6.365", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3365.wav", "onoffCaption": "burping belching at 2.845-6.351, 7.13-9.237 and cow mooing at 6.748-9.717", "frequencyCaption": "burping belching two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3380.wav", "onoffCaption": "tapping clicking clanking at 0.218-3.658, 4.449-7.889 and burping belching at 3.457-5.564, 7.966-10.0", "frequencyCaption": "tapping clicking clanking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3515.wav", "onoffCaption": "car horn honking at 0.091-2.091, 3.107-5.559", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_3540.wav", "onoffCaption": "burping belching at 2.061-5.561", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3628.wav", "onoffCaption": "door slamming at 0.036-1.551, 2.263-3.402, 3.906-6.623", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3698.wav", "onoffCaption": "car horn honking at 0.33-3.243, 4.313-6.313 and door slamming at 8.776-9.276", "frequencyCaption": "car horn honking two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3733.wav", "onoffCaption": "whistling at 1.749-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3766.wav", "onoffCaption": "door slamming at 3.189-4.208, 5.391-7.391, 8.414-9.929", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_3783.wav", "onoffCaption": "woman laughing at 0.348-2.767, 4.374-6.793", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3811.wav", "onoffCaption": "cow mooing at 0.547-3.529 and dog barking at 6.774-8.774", "frequencyCaption": "cow mooing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3844.wav", "onoffCaption": "duck quacking at 0.575-2.575, 3.754-5.754 and spraying at 2.908-4.083, 5.036-6.211, 7.264-8.439", "frequencyCaption": "duck quacking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3945.wav", "onoffCaption": "burping belching at 2.053-8.14 and door knocking at 3.198-6.254", "frequencyCaption": "burping belching one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_34.wav", "onoffCaption": "train horn at 1.481-5.549, 7.631-9.823", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_61.wav", "onoffCaption": "duck quacking at 0.2-2.2", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_84.wav", "onoffCaption": "car horn honking at 0.489-4.143, 5.186-8.038", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_137.wav", "onoffCaption": "door slamming at 0.068-2.849", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_146.wav", "onoffCaption": "tapping clicking clanking at 1.273-4.713, 6.496-9.936 and spraying at 1.889-3.146, 3.685-4.536, 5.216-5.8", "frequencyCaption": "tapping clicking clanking two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_335.wav", "onoffCaption": "whistling at 0.294-8.679", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_360.wav", "onoffCaption": "gunshot at 2.465-4.971, 6.528-9.034", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_385.wav", "onoffCaption": "cow mooing at 3.411-6.421", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_510.wav", "onoffCaption": "whistling at 0.347-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_561.wav", "onoffCaption": "door slamming at 0.024-0.964, 2.21-3.975 and thump thud at 6.63-9.13", "frequencyCaption": "door slamming two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_678.wav", "onoffCaption": "burping belching at 0.375-3.554, 4.177-6.336", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_736.wav", "onoffCaption": "cat meowing at 1.867-2.877, 3.724-5.295", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_763.wav", "onoffCaption": "sheep goat bleating at 2.894-4.894 and cat meowing at 3.637-4.649", "frequencyCaption": "sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_786.wav", "onoffCaption": "door slamming at 3.21-4.734", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_814.wav", "onoffCaption": "spraying at 3.628-4.378, 5.721-6.471", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_841.wav", "onoffCaption": "train horn at 3.28-7.04", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1020.wav", "onoffCaption": "sneeze at 2.574-6.63, 7.459-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1090.wav", "onoffCaption": "door knocking at 3.538-5.602, 7.279-9.343", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1206.wav", "onoffCaption": "sheep goat bleating at 1.226-3.226", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1253.wav", "onoffCaption": "dog barking at 2.277-4.277, 6.028-8.028", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1348.wav", "onoffCaption": "car horn honking at 3.545-7.945", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_1476.wav", "onoffCaption": "cow mooing at 1.587-4.569, 5.899-8.881", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1493.wav", "onoffCaption": "tapping clicking clanking at 0.435-3.875, 5.267-7.646 and spraying at 1.453-2.08", "frequencyCaption": "tapping clicking clanking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1538.wav", "onoffCaption": "door knocking at 0.508-2.91, 5.169-7.571", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1588.wav", "onoffCaption": "cow mooing at 0.188-3.17, 5.076-8.058", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_1605.wav", "onoffCaption": "spraying at 3.885-5.58, 7.472-8.556", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_1650.wav", "onoffCaption": "door knocking at 0.332-3.388, 4.519-7.366", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1869.wav", "onoffCaption": "door knocking at 0.021-2.148, 4.194-6.321, 7.719-9.846", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_1927.wav", "onoffCaption": "duck quacking at 0.096-2.096, 3.435-5.435", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1972.wav", "onoffCaption": "train horn at 0.413-4.853", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1997.wav", "onoffCaption": "spraying at 1.763-3.02 and sheep goat bleating at 4.59-6.59", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3017.wav", "onoffCaption": "duck quacking at 0.399-2.399 and explosion at 0.665-5.665 and cat meowing at 7.386-8.386", "frequencyCaption": "duck quacking one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3042.wav", "onoffCaption": "train horn at 0.39-8.59 and spraying at 2.788-3.692, 4.267-4.836, 5.639-6.147", "frequencyCaption": "train horn one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_3159.wav", "onoffCaption": "tapping clicking clanking at 0.107-3.547", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3231.wav", "onoffCaption": "burping belching at 0.186-5.787, 7.524-9.555 and door slamming at 2.231-3.996", "frequencyCaption": "burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3264.wav", "onoffCaption": "whistling at 2.172-5.047, 6.058-8.933", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_3281.wav", "onoffCaption": "burping belching at 0.199-3.639, 5.384-7.667 and cat meowing at 4.673-6.568, 7.795-9.69", "frequencyCaption": "burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3414.wav", "onoffCaption": "thump thud at 3.291-5.791, 7.021-9.521", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3441.wav", "onoffCaption": "spraying at 0.092-2.552, 4.511-5.161", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3632.wav", "onoffCaption": "thump thud at 1.952-4.452, 6.848-9.348", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_3667.wav", "onoffCaption": "tapping clicking clanking at 3.039-6.479, 7.37-9.609", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3682.wav", "onoffCaption": "burping belching at 1.994-4.755, 5.399-8.16", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_3729.wav", "onoffCaption": "thump thud at 2.368-4.83", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3799.wav", "onoffCaption": "woman laughing at 1.673-4.773, 5.556-8.656 and spraying at 2.654-3.738, 5.882-8.025", "frequencyCaption": "woman laughing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_3910.wav", "onoffCaption": "woman laughing at 0.775-3.194, 5.597-8.563", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_3961.wav", "onoffCaption": "sneeze at 0.001-1.255, 1.77-3.024 and explosion at 6.263-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_10.wav", "onoffCaption": "sneeze at 0.063-1.976, 2.801-4.714 and explosion at 7.673-10.0", "frequencyCaption": "sneeze two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_45.wav", "onoffCaption": "cow mooing at 0.355-3.337, 4.617-7.599", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_162.wav", "onoffCaption": "woman laughing at 1.442-4.496", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_187.wav", "onoffCaption": "burping belching at 0.292-7.269 and spraying at 6.756-7.34", "frequencyCaption": "burping belching one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_311.wav", "onoffCaption": "dog barking at 2.879-4.879, 5.486-7.486", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_344.wav", "onoffCaption": "gunshot at 3.671-5.671, 7.994-9.994", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_534.wav", "onoffCaption": "cow mooing at 0.076-5.056 and door knocking at 0.179-2.559 and woman laughing at 5.2-7.281", "frequencyCaption": "cow mooing one times and door knocking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_584.wav", "onoffCaption": "gunshot at 0.179-2.179 and burping belching at 0.417-2.647 and door slamming at 5.169-7.169", "frequencyCaption": "gunshot one times and burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_609.wav", "onoffCaption": "train horn at 0.029-2.684, 3.212-5.405 and door slamming at 2.16-3.138", "frequencyCaption": "train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_712.wav", "onoffCaption": "burping belching at 0.041-2.076, 2.801-4.836, 5.602-7.637", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_747.wav", "onoffCaption": "duck quacking at 0.484-2.484, 3.334-5.334, 5.954-7.954", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_830.wav", "onoffCaption": "woman laughing at 0.458-2.574 and spraying at 0.798-1.32 and burping belching at 5.364-8.364", "frequencyCaption": "woman laughing one times and spraying one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_865.wav", "onoffCaption": "duck quacking at 2.366-4.366, 5.105-7.105 and cow mooing at 3.278-6.26", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_880.wav", "onoffCaption": "car horn honking at 1.336-4.249 and spraying at 2.806-4.063", "frequencyCaption": "car horn honking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_1004.wav", "onoffCaption": "whistling at 3.77-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1051.wav", "onoffCaption": "car horn honking at 2.969-5.316, 6.834-8.89", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_1222.wav", "onoffCaption": "burping belching at 3.161-7.497", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_1277.wav", "onoffCaption": "explosion at 0.578-5.578", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1292.wav", "onoffCaption": "dog barking at 0.479-2.479, 4.616-6.616", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1339.wav", "onoffCaption": "train horn at 1.157-7.217", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_1389.wav", "onoffCaption": "cat meowing at 0.527-1.674, 2.783-3.93, 4.549-5.696 and gunshot at 2.167-4.167, 6.628-8.628", "frequencyCaption": "cat meowing three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1407.wav", "onoffCaption": "gunshot at 2.758-4.758, 6.25-8.25", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1452.wav", "onoffCaption": "thump thud at 0.107-4.557, 6.901-10.0 and gunshot at 0.577-2.596, 4.566-6.585", "frequencyCaption": "thump thud two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_1549.wav", "onoffCaption": "burping belching at 0.348-3.348 and door knocking at 2.068-4.445, 5.651-8.028", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1621.wav", "onoffCaption": "door knocking at 1.772-6.211", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_1674.wav", "onoffCaption": "thump thud at 3.223-5.685", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1691.wav", "onoffCaption": "woman laughing at 0.558-2.752 and duck quacking at 1.9-3.9, 5.909-7.909", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1818.wav", "onoffCaption": "woman laughing at 1.357-4.452, 5.885-8.98", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1903.wav", "onoffCaption": "spraying at 2.334-2.842, 3.905-4.413, 5.561-6.069", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_1956.wav", "onoffCaption": "sheep goat bleating at 2.9-4.9 and dog barking at 2.959-4.959", "frequencyCaption": "sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3033.wav", "onoffCaption": "thump thud at 1.152-5.527", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_3066.wav", "onoffCaption": "car horn honking at 0.514-3.027 and sheep goat bleating at 0.548-2.548", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3083.wav", "onoffCaption": "whistling at 3.592-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3128.wav", "onoffCaption": "burping belching at 0.775-4.335 and tapping clicking clanking at 2.973-6.413", "frequencyCaption": "burping belching one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_3198.wav", "onoffCaption": "cow mooing at 0.412-5.392, 6.283-10.0 and dog barking at 0.442-2.442, 3.318-5.318, 6.398-8.398", "frequencyCaption": "cow mooing two times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3215.wav", "onoffCaption": "dog barking at 0.091-2.091, 3.213-5.213, 6.426-8.426", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_3240.wav", "onoffCaption": "tapping clicking clanking at 1.525-4.965, 5.786-9.226", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3430.wav", "onoffCaption": "whistling at 1.553-9.491", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_3465.wav", "onoffCaption": "duck quacking at 1.309-3.309", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_3480.wav", "onoffCaption": "spraying at 0.255-2.383, 4.595-6.723", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3616.wav", "onoffCaption": "train horn at 0.002-2.482, 3.51-5.99 and door slamming at 0.294-1.669", "frequencyCaption": "train horn two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_3643.wav", "onoffCaption": "cow mooing at 0.5-3.469", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_3758.wav", "onoffCaption": "door knocking at 0.079-2.847, 4.619-6.84 and dog barking at 5.333-7.333", "frequencyCaption": "door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_3934.wav", "onoffCaption": "door knocking at 0.916-4.684", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_3984.wav", "onoffCaption": "train horn at 1.384-4.744 and thump thud at 2.294-4.794, 5.817-7.852 and car horn honking at 3.846-6.759", "frequencyCaption": "train horn one times and thump thud two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_178.wav", "onoffCaption": "cow mooing at 3.381-7.81", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_210.wav", "onoffCaption": "whistling at 0.071-2.08, 2.921-4.93 and explosion at 0.913-3.913, 5.782-8.782", "frequencyCaption": "whistling two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_245.wav", "onoffCaption": "explosion at 1.121-4.177, 6.294-9.35", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_435.wav", "onoffCaption": "woman laughing at 0.561-3.199, 4.189-6.827 and gunshot at 3.425-5.425", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_460.wav", "onoffCaption": "duck quacking at 3.484-5.484, 6.175-8.175", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_485.wav", "onoffCaption": "tapping clicking clanking at 0.224-3.664 and door knocking at 6.512-9.28", "frequencyCaption": "tapping clicking clanking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_613.wav", "onoffCaption": "whistling at 0.43-8.085 and dog barking at 2.187-4.625, 5.225-7.225 and cow mooing at 3.707-6.689", "frequencyCaption": "whistling one times and dog barking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_646.wav", "onoffCaption": "train horn at 3.064-7.922", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_708.wav", "onoffCaption": "gunshot at 2.261-4.767, 7.175-9.175", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_931.wav", "onoffCaption": "thump thud at 0.167-2.667", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_964.wav", "onoffCaption": "sheep goat bleating at 1.528-3.528, 4.975-6.975", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_981.wav", "onoffCaption": "car horn honking at 0.323-4.645 and gunshot at 7.394-9.394", "frequencyCaption": "car horn honking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_1105.wav", "onoffCaption": "sheep goat bleating at 2.371-4.371, 6.611-8.611", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_1150.wav", "onoffCaption": "duck quacking at 2.106-4.106, 5.268-7.268", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_1238.wav", "onoffCaption": "sheep goat bleating at 3.412-6.492", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_1288.wav", "onoffCaption": "cat meowing at 0.746-2.363", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_1323.wav", "onoffCaption": "dog barking at 0.979-2.979, 4.467-6.467, 7.269-9.269", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_1376.wav", "onoffCaption": "explosion at 2.257-7.257", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_1393.wav", "onoffCaption": "tapping clicking clanking at 0.14-3.58, 5.201-7.253 and duck quacking at 2.426-4.426", "frequencyCaption": "tapping clicking clanking two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_1448.wav", "onoffCaption": "woman laughing at 0.367-2.851, 5.089-7.789", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_1506.wav", "onoffCaption": "dog barking at 0.726-2.726, 4.078-6.078", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_1553.wav", "onoffCaption": "tapping clicking clanking at 0.156-3.596, 4.737-7.641", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_1720.wav", "onoffCaption": "gunshot at 1.155-3.155, 5.398-7.528 and thump thud at 3.363-7.738", "frequencyCaption": "gunshot two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1775.wav", "onoffCaption": "door knocking at 2.969-5.816, 7.75-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_1790.wav", "onoffCaption": "door knocking at 0.011-2.391, 4.036-6.546, 7.09-9.7 and train horn at 1.029-5.469 and thump thud at 1.711-4.211", "frequencyCaption": "door knocking three times and train horn one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_1802.wav", "onoffCaption": "sneeze at 0.384-2.697, 4.475-6.788", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_1857.wav", "onoffCaption": "whistling at 1.733-4.608", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_1919.wav", "onoffCaption": "sheep goat bleating at 0.776-2.776, 3.517-5.517, 6.164-8.164 and car horn honking at 4.011-8.411", "frequencyCaption": "sheep goat bleating three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_3029.wav", "onoffCaption": "cow mooing at 3.488-6.47, 7.941-10.0 and burping belching at 3.869-6.869", "frequencyCaption": "cow mooing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_3099.wav", "onoffCaption": "gunshot at 1.876-3.876, 4.38-6.38 and woman laughing at 2.584-4.778", "frequencyCaption": "gunshot two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3132.wav", "onoffCaption": "door knocking at 0.692-5.525, 6.427-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3167.wav", "onoffCaption": "woman laughing at 3.333-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_3182.wav", "onoffCaption": "tapping clicking clanking at 1.809-5.249, 6.999-9.867", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_3314.wav", "onoffCaption": "sheep goat bleating at 1.93-3.93", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_3341.wav", "onoffCaption": "spraying at 3.082-5.666, 7.209-9.793", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_3531.wav", "onoffCaption": "sneeze at 0.185-3.26, 3.84-6.316, 7.242-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3564.wav", "onoffCaption": "sheep goat bleating at 0.809-2.809, 4.118-6.65, 7.416-9.416", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_3581.wav", "onoffCaption": "explosion at 0.101-2.189", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_3659.wav", "onoffCaption": "sneeze at 0.638-2.751, 3.548-5.661, 6.812-8.925", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_3717.wav", "onoffCaption": "door knocking at 1.199-3.359, 4.572-7.531", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_3742.wav", "onoffCaption": "cat meowing at 0.275-1.286", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_3835.wav", "onoffCaption": "cat meowing at 2.698-3.707, 4.225-6.202", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_3860.wav", "onoffCaption": "duck quacking at 0.859-2.859, 3.637-5.637, 6.544-8.544", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_3885.wav", "onoffCaption": "cow mooing at 3.53-8.51", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2120.wav", "onoffCaption": "spraying at 0.057-2.449, 3.255-3.763", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2175.wav", "onoffCaption": "spraying at 2.071-3.071 and sheep goat bleating at 3.431-7.351", "frequencyCaption": "spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2190.wav", "onoffCaption": "train horn at 1.771-5.305", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2248.wav", "onoffCaption": "explosion at 1.654-4.372, 5.029-7.747 and car horn honking at 2.113-4.626, 5.21-7.675 and burping belching at 5.214-7.54", "frequencyCaption": "explosion two times and car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2306.wav", "onoffCaption": "cow mooing at 0.725-3.694, 6.08-8.942", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2353.wav", "onoffCaption": "car horn honking at 1.862-3.862, 4.452-6.452, 7.24-9.24", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_2438.wav", "onoffCaption": "sheep goat bleating at 1.107-4.107 and duck quacking at 3.052-5.052", "frequencyCaption": "sheep goat bleating one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2477.wav", "onoffCaption": "whistling at 2.127-5.002, 5.545-8.42", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2488.wav", "onoffCaption": "sheep goat bleating at 0.735-2.735, 3.386-5.386 and door slamming at 1.8-3.033, 3.723-4.956 and tapping clicking clanking at 2.198-5.638, 6.329-9.769", "frequencyCaption": "sheep goat bleating two times and door slamming two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2523.wav", "onoffCaption": "woman laughing at 2.767-4.967, 5.806-8.006", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2576.wav", "onoffCaption": "explosion at 0.277-2.868 and cat meowing at 6.026-10.0", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2593.wav", "onoffCaption": "dog barking at 1.825-3.825, 4.517-6.517", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2705.wav", "onoffCaption": "woman laughing at 0.971-4.071 and explosion at 7.935-10.0", "frequencyCaption": "woman laughing one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2750.wav", "onoffCaption": "gunshot at 2.845-4.845, 5.516-7.516", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2827.wav", "onoffCaption": "burping belching at 0.845-3.171", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2872.wav", "onoffCaption": "door slamming at 3.272-6.233", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2897.wav", "onoffCaption": "car horn honking at 1.909-5.75, 7.713-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2969.wav", "onoffCaption": "duck quacking at 0.892-2.892", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2996.wav", "onoffCaption": "dog barking at 2.902-4.902, 7.003-9.003", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4078.wav", "onoffCaption": "door slamming at 0.238-0.738", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4136.wav", "onoffCaption": "burping belching at 3.19-7.213 and duck quacking at 3.224-5.224, 5.953-7.953", "frequencyCaption": "burping belching one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4163.wav", "onoffCaption": "train horn at 1.452-4.332 and cow mooing at 6.448-9.746", "frequencyCaption": "train horn one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4186.wav", "onoffCaption": "car horn honking at 1.257-3.604 and dog barking at 6.705-10.0", "frequencyCaption": "car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4310.wav", "onoffCaption": "train horn at 2.161-6.342, 6.935-9.277", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4345.wav", "onoffCaption": "gunshot at 0.772-2.772", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4535.wav", "onoffCaption": "sheep goat bleating at 2.485-6.125", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4560.wav", "onoffCaption": "burping belching at 0.436-3.197 and gunshot at 1.84-4.346", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4585.wav", "onoffCaption": "thump thud at 0.62-4.538, 6.81-9.31", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4608.wav", "onoffCaption": "whistling at 3.318-7.802", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4713.wav", "onoffCaption": "train horn at 0.267-3.427, 5.079-8.239", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4746.wav", "onoffCaption": "gunshot at 1.968-3.968, 4.801-6.801", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4831.wav", "onoffCaption": "tapping clicking clanking at 0.749-4.189", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4864.wav", "onoffCaption": "sheep goat bleating at 0.308-2.308, 4.509-6.509", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4881.wav", "onoffCaption": "sneeze at 0.109-2.512, 3.965-6.397 and whistling at 2.282-7.782", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_6.wav", "onoffCaption": "thump thud at 0.14-3.807, 5.41-9.077 and spraying at 0.454-0.954, 1.799-2.65, 4.17-4.745", "frequencyCaption": "thump thud two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2021.wav", "onoffCaption": "door knocking at 0.429-5.429, 6.778-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2074.wav", "onoffCaption": "whistling at 0.727-5.211, 6.879-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2091.wav", "onoffCaption": "spraying at 1.831-4.85, 6.142-7.837, 8.47-9.12", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2207.wav", "onoffCaption": "thump thud at 2.65-5.15", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2252.wav", "onoffCaption": "explosion at 0.012-3.565, 5.39-7.843", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2349.wav", "onoffCaption": "thump thud at 0.309-2.809, 3.923-6.423, 7.405-9.905", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2422.wav", "onoffCaption": "car horn honking at 2.361-6.761, 7.274-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2453.wav", "onoffCaption": "thump thud at 3.225-5.687, 6.839-9.605", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2492.wav", "onoffCaption": "explosion at 0.455-5.455 and burping belching at 2.437-6.306, 7.467-9.59", "frequencyCaption": "explosion one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2539.wav", "onoffCaption": "sneeze at 0.488-2.612, 3.718-6.686, 7.684-9.212", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2548.wav", "onoffCaption": "door knocking at 2.153-4.616, 5.711-8.174", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2589.wav", "onoffCaption": "gunshot at 3.892-5.892", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2604.wav", "onoffCaption": "thump thud at 0.045-3.092, 3.677-5.905 and train horn at 1.656-4.856, 5.734-8.934", "frequencyCaption": "thump thud two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2651.wav", "onoffCaption": "explosion at 2.423-5.151, 6.028-8.756", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2868.wav", "onoffCaption": "thump thud at 0.855-5.23, 6.386-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2926.wav", "onoffCaption": "dog barking at 0.219-3.14, 4.211-7.132", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2973.wav", "onoffCaption": "tapping clicking clanking at 2.703-6.143, 7.187-9.19", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4037.wav", "onoffCaption": "train horn at 1.92-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4062.wav", "onoffCaption": "cat meowing at 0.012-1.596, 2.868-3.877, 5.167-7.299 and dog barking at 6.274-8.274", "frequencyCaption": "cat meowing three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4087.wav", "onoffCaption": "explosion at 2.553-6.106, 7.68-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4179.wav", "onoffCaption": "duck quacking at 2.786-4.786, 5.324-7.324", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4211.wav", "onoffCaption": "thump thud at 0.315-2.654, 4.121-6.645 and spraying at 5.969-6.75", "frequencyCaption": "thump thud two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4244.wav", "onoffCaption": "car horn honking at 0.049-3.636 and dog barking at 0.885-2.885 and spraying at 1.371-2.435", "frequencyCaption": "car horn honking one times and dog barking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4434.wav", "onoffCaption": "sneeze at 3.28-5.519, 7.824-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4461.wav", "onoffCaption": "cat meowing at 3.162-4.311, 5.935-6.946", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4484.wav", "onoffCaption": "door knocking at 3.211-5.674, 6.972-9.435", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4612.wav", "onoffCaption": "door knocking at 0.089-4.528, 5.195-9.634", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4647.wav", "onoffCaption": "sneeze at 0.569-2.156, 3.144-4.731, 5.782-7.369 and gunshot at 5.878-7.878", "frequencyCaption": "sneeze three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4709.wav", "onoffCaption": "whistling at 0.206-6.583, 7.637-10.0 and door knocking at 2.165-4.735", "frequencyCaption": "whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4930.wav", "onoffCaption": "explosion at 2.446-7.446", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4965.wav", "onoffCaption": "explosion at 3.189-6.245, 7.568-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4980.wav", "onoffCaption": "cat meowing at 2.685-3.712 and door slamming at 2.893-4.032, 6.014-7.153", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2005.wav", "onoffCaption": "thump thud at 0.343-4.01, 4.582-7.082 and car horn honking at 1.183-5.583, 7.486-10.0", "frequencyCaption": "thump thud two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2050.wav", "onoffCaption": "sneeze at 2.987-4.09 and spraying at 4.169-5.073, 5.726-6.327, 7.545-8.629", "frequencyCaption": "sneeze one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2223.wav", "onoffCaption": "duck quacking at 0.332-2.332, 2.942-4.942, 6.627-8.627", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2239.wav", "onoffCaption": "sheep goat bleating at 0.339-2.339, 3.156-5.156", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2276.wav", "onoffCaption": "thump thud at 0.322-4.697, 6.983-9.599", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2293.wav", "onoffCaption": "car horn honking at 0.996-3.461, 4.536-7.001, 7.588-10.0 and tapping clicking clanking at 2.689-6.129", "frequencyCaption": "car horn honking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2322.wav", "onoffCaption": "thump thud at 0.418-3.465, 4.569-7.161", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2338.wav", "onoffCaption": "spraying at 0.027-1.513, 3.975-5.232, 7.548-8.723 and burping belching at 2.07-4.593", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2388.wav", "onoffCaption": "woman laughing at 0.794-3.002, 4.227-6.435", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2406.wav", "onoffCaption": "car horn honking at 0.603-2.603 and train horn at 3.813-6.973", "frequencyCaption": "car horn honking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2620.wav", "onoffCaption": "cat meowing at 2.446-4.582, 5.348-7.484", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2675.wav", "onoffCaption": "dog barking at 2.962-4.962, 7.387-9.387", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2690.wav", "onoffCaption": "duck quacking at 0.393-2.393, 3.405-5.405, 5.915-7.915", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2774.wav", "onoffCaption": "sheep goat bleating at 2.619-4.619, 5.242-7.242", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2819.wav", "onoffCaption": "sheep goat bleating at 0.032-2.032, 3.521-5.521 and door knocking at 0.864-3.984, 6.406-9.526", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2902.wav", "onoffCaption": "sneeze at 0.001-2.462, 4.092-6.553", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2957.wav", "onoffCaption": "door slamming at 0.986-3.865, 4.823-7.702", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4013.wav", "onoffCaption": "tapping clicking clanking at 0.088-3.528, 5.482-8.922", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4046.wav", "onoffCaption": "cow mooing at 0.641-5.07 and door slamming at 1.475-3.24, 4.431-6.196, 6.906-8.671", "frequencyCaption": "cow mooing one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4108.wav", "onoffCaption": "spraying at 1.708-2.649, 4.881-5.508", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4112.wav", "onoffCaption": "sneeze at 3.141-4.418", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4235.wav", "onoffCaption": "car horn honking at 0.77-5.092, 6.546-9.404", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4260.wav", "onoffCaption": "woman laughing at 0.463-3.517, 5.483-8.537", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4285.wav", "onoffCaption": "cow mooing at 0.753-3.722, 4.97-7.106", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4410.wav", "onoffCaption": "woman laughing at 0.635-7.369, 7.95-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4445.wav", "onoffCaption": "whistling at 2.422-8.914", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4544.wav", "onoffCaption": "spraying at 1.576-2.326, 2.849-3.599, 4.646-5.396 and gunshot at 1.949-3.949, 5.622-7.622", "frequencyCaption": "spraying three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4636.wav", "onoffCaption": "train horn at 0.178-2.852", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4663.wav", "onoffCaption": "sneeze at 2.824-5.484, 6.738-7.832 and duck quacking at 2.914-4.914", "frequencyCaption": "sneeze two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4686.wav", "onoffCaption": "car horn honking at 1.868-5.363, 6.193-9.688", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4778.wav", "onoffCaption": "dog barking at 1.856-3.856, 4.606-6.606, 7.924-9.924 and tapping clicking clanking at 2.775-6.215", "frequencyCaption": "dog barking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4840.wav", "onoffCaption": "whistling at 0.068-4.552", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4914.wav", "onoffCaption": "dog barking at 2.546-4.546, 5.133-7.484", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4941.wav", "onoffCaption": "duck quacking at 0.364-2.364, 4.648-6.648", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2104.wav", "onoffCaption": "car horn honking at 2.211-4.558, 6.798-9.145", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2151.wav", "onoffCaption": "gunshot at 2.256-4.256, 4.891-6.891", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2156.wav", "onoffCaption": "dog barking at 0.052-2.052, 2.948-4.948, 5.728-7.728 and door slamming at 1.308-4.282, 5.326-8.3", "frequencyCaption": "dog barking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2289.wav", "onoffCaption": "spraying at 1.58-2.312, 3.778-4.51, 5.034-5.766 and gunshot at 2.716-4.716, 5.541-7.541", "frequencyCaption": "spraying three times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2377.wav", "onoffCaption": "cat meowing at 3.482-7.842", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2392.wav", "onoffCaption": "sneeze at 3.873-5.869, 6.876-8.872", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2449.wav", "onoffCaption": "train horn at 0.152-2.632 and spraying at 0.716-1.583 and woman laughing at 5.347-7.702", "frequencyCaption": "train horn one times and spraying one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2500.wav", "onoffCaption": "door slamming at 1.297-2.135 and spraying at 4.361-4.869, 7.135-7.704", "frequencyCaption": "door slamming one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2507.wav", "onoffCaption": "door knocking at 0.199-2.387", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2552.wav", "onoffCaption": "sheep goat bleating at 0.008-2.008 and woman laughing at 0.341-3.441, 4.758-7.858", "frequencyCaption": "sheep goat bleating one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2668.wav", "onoffCaption": "tapping clicking clanking at 2.077-5.517, 6.443-8.613 and cow mooing at 5.63-8.612", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2721.wav", "onoffCaption": "gunshot at 3.487-5.487", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2773.wav", "onoffCaption": "door knocking at 0.433-2.896 and door slamming at 3.612-4.987, 7.366-8.741", "frequencyCaption": "door knocking one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2791.wav", "onoffCaption": "door knocking at 0.652-4.402", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2803.wav", "onoffCaption": "thump thud at 0.011-2.35 and spraying at 4.888-6.145", "frequencyCaption": "thump thud one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2804.wav", "onoffCaption": "cow mooing at 0.049-4.478, 6.606-9.378 and thump thud at 2.151-4.922, 6.328-8.828 and burping belching at 3.01-5.108", "frequencyCaption": "cow mooing two times and thump thud two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2856.wav", "onoffCaption": "duck quacking at 1.872-3.872, 6.151-8.151", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2918.wav", "onoffCaption": "gunshot at 3.13-5.13, 6.539-8.539", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4009.wav", "onoffCaption": "whistling at 1.047-6.547, 7.7-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4147.wav", "onoffCaption": "whistling at 1.949-7.124", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4298.wav", "onoffCaption": "train horn at 0.227-2.667, 4.148-6.296", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4334.wav", "onoffCaption": "explosion at 0.551-3.551 and cat meowing at 0.943-3.079", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4361.wav", "onoffCaption": "cow mooing at 2.731-5.7, 6.649-9.618", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4366.wav", "onoffCaption": "duck quacking at 0.872-2.872, 4.328-6.328 and explosion at 1.918-4.79, 5.543-8.271", "frequencyCaption": "duck quacking two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_4383.wav", "onoffCaption": "car horn honking at 1.956-5.175, 5.822-8.12 and burping belching at 4.017-7.017", "frequencyCaption": "car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4384.wav", "onoffCaption": "door knocking at 0.945-4.001 and train horn at 6.737-8.892", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4511.wav", "onoffCaption": "car horn honking at 3.701-8.213", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4679.wav", "onoffCaption": "thump thud at 0.164-4.614, 5.439-9.889", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4730.wav", "onoffCaption": "train horn at 1.755-4.429, 5.275-7.949", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4737.wav", "onoffCaption": "gunshot at 1.048-3.048, 4.748-6.748 and spraying at 3.137-4.201 and sheep goat bleating at 3.936-5.936", "frequencyCaption": "gunshot two times and spraying one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4762.wav", "onoffCaption": "door slamming at 2.032-4.49, 6.927-8.927", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4787.wav", "onoffCaption": "woman laughing at 0.311-2.511, 3.934-6.526 and sneeze at 2.213-3.8, 6.089-7.676", "frequencyCaption": "woman laughing two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4815.wav", "onoffCaption": "sneeze at 0.265-1.499 and whistling at 5.673-7.902", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4847.wav", "onoffCaption": "whistling at 0.013-7.763 and tapping clicking clanking at 0.111-3.551 and sneeze at 1.548-3.135, 4.059-5.646", "frequencyCaption": "whistling one times and tapping clicking clanking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2002.wav", "onoffCaption": "woman laughing at 3.333-10.0", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2018.wav", "onoffCaption": "spraying at 0.405-1.032, 1.643-2.27, 3.029-3.656 and explosion at 0.527-5.527, 6.694-10.0", "frequencyCaption": "spraying three times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2103.wav", "onoffCaption": "sneeze at 1.671-3.984, 5.21-7.523 and burping belching at 1.76-3.99, 5.309-7.346 and door slamming at 4.12-6.837", "frequencyCaption": "sneeze two times and burping belching two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2119.wav", "onoffCaption": "dog barking at 2.203-4.203, 6.323-8.323", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2271.wav", "onoffCaption": "tapping clicking clanking at 0.592-4.032, 5.924-9.364", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2294.wav", "onoffCaption": "gunshot at 0.764-2.764 and spraying at 5.053-6.134, 8.451-9.532", "frequencyCaption": "gunshot one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2325.wav", "onoffCaption": "dog barking at 0.642-2.642, 3.414-5.414, 6.681-8.681 and cow mooing at 4.632-7.601", "frequencyCaption": "dog barking three times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2370.wav", "onoffCaption": "thump thud at 0.04-2.502", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2395.wav", "onoffCaption": "spraying at 0.189-0.697, 1.468-1.976, 2.924-3.432 and burping belching at 1.098-3.328", "frequencyCaption": "spraying three times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2454.wav", "onoffCaption": "cat meowing at 0.477-2.425, 3.605-5.553, 7.46-9.408", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2555.wav", "onoffCaption": "cat meowing at 0.068-1.484, 3.808-5.768", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2627.wav", "onoffCaption": "burping belching at 0.031-4.031, 5.198-9.198", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2697.wav", "onoffCaption": "whistling at 2.205-4.214, 4.925-6.934 and sneeze at 4.976-6.15, 8.57-9.744", "frequencyCaption": "whistling two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2726.wav", "onoffCaption": "sneeze at 1.043-3.446 and thump thud at 6.486-8.825", "frequencyCaption": "sneeze one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2796.wav", "onoffCaption": "spraying at 0.031-1.206, 2.852-4.099, 5.641-8.225 and tapping clicking clanking at 6.006-9.446", "frequencyCaption": "spraying three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2851.wav", "onoffCaption": "burping belching at 0.624-4.13, 5.824-9.33", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2950.wav", "onoffCaption": "explosion at 3.244-5.973 and sheep goat bleating at 3.421-5.421", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4014.wav", "onoffCaption": "door knocking at 2.683-5.063, 7.022-9.402", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4041.wav", "onoffCaption": "explosion at 2.544-7.544", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4115.wav", "onoffCaption": "cow mooing at 3.295-8.275", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4140.wav", "onoffCaption": "door slamming at 1.027-3.906 and whistling at 6.755-9.73", "frequencyCaption": "door slamming one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4228.wav", "onoffCaption": "woman laughing at 0.009-2.107, 4.329-6.427", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4232.wav", "onoffCaption": "cat meowing at 0.174-1.449, 2.265-3.54, 4.479-5.754", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4329.wav", "onoffCaption": "burping belching at 2.373-5.873", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4333.wav", "onoffCaption": "door knocking at 0.096-4.629 and cow mooing at 2.461-5.471, 6.445-9.405", "frequencyCaption": "door knocking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4417.wav", "onoffCaption": "explosion at 0.436-3.027, 4.242-6.244, 7.375-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_4458.wav", "onoffCaption": "door slamming at 2.041-3.806", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4516.wav", "onoffCaption": "thump thud at 3.061-6.108, 6.938-9.985", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4543.wav", "onoffCaption": "dog barking at 3.213-5.213 and sheep goat bleating at 7.736-9.736", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4664.wav", "onoffCaption": "car horn honking at 0.073-2.586 and cat meowing at 0.304-1.304, 2.382-3.382, 4.485-5.485", "frequencyCaption": "car horn honking one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4681.wav", "onoffCaption": "explosion at 1.201-6.201, 7.318-10.0 and train horn at 3.436-6.636", "frequencyCaption": "explosion two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4765.wav", "onoffCaption": "sheep goat bleating at 0.926-2.926, 3.878-6.477, 7.176-9.176", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4780.wav", "onoffCaption": "cat meowing at 0.028-1.303, 2.305-3.876, 5.456-6.483", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4808.wav", "onoffCaption": "duck quacking at 2.642-4.642, 6.256-8.256", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4812.wav", "onoffCaption": "burping belching at 3.173-7.509", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4909.wav", "onoffCaption": "train horn at 0.691-4.891, 5.967-8.26", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4913.wav", "onoffCaption": "sneeze at 0.276-1.51 and whistling at 5.819-10.0", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2026.wav", "onoffCaption": "sheep goat bleating at 0.154-2.154, 3.022-5.022, 6.172-8.172", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2057.wav", "onoffCaption": "burping belching at 2.683-6.227, 6.907-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2200.wav", "onoffCaption": "whistling at 2.022-9.772", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2224.wav", "onoffCaption": "burping belching at 2.351-4.382", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2255.wav", "onoffCaption": "tapping clicking clanking at 2.125-5.565 and sneeze at 2.99-5.466, 6.608-7.772", "frequencyCaption": "tapping clicking clanking one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2401.wav", "onoffCaption": "cow mooing at 0.674-5.103, 6.783-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2470.wav", "onoffCaption": "door slamming at 0.396-1.771, 2.767-4.142", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2495.wav", "onoffCaption": "door slamming at 0.565-2.928, 3.911-4.762, 6.499-7.439", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2603.wav", "onoffCaption": "sheep goat bleating at 1.829-3.829 and cow mooing at 6.879-9.848", "frequencyCaption": "sheep goat bleating one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2672.wav", "onoffCaption": "cat meowing at 1.868-3.485 and door slamming at 2.29-4.516, 6.876-9.102", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2718.wav", "onoffCaption": "sneeze at 0.124-1.581, 3.955-5.483", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2769.wav", "onoffCaption": "car horn honking at 0.692-4.941", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2905.wav", "onoffCaption": "burping belching at 0.043-4.442, 6.234-10.0 and sheep goat bleating at 0.792-2.792, 3.816-5.816, 6.776-8.776", "frequencyCaption": "burping belching two times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2974.wav", "onoffCaption": "sheep goat bleating at 3.18-6.5", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2991.wav", "onoffCaption": "thump thud at 0.605-4.272 and tapping clicking clanking at 0.78-4.22 and sheep goat bleating at 1.8-3.8, 4.865-6.865, 7.674-9.674", "frequencyCaption": "thump thud one times and tapping clicking clanking one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4030.wav", "onoffCaption": "sheep goat bleating at 2.192-4.192, 4.895-6.895 and whistling at 3.502-9.002", "frequencyCaption": "sheep goat bleating two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4065.wav", "onoffCaption": "thump thud at 0.945-3.992, 5.083-8.13", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4080.wav", "onoffCaption": "whistling at 1.188-3.197, 5.237-7.246 and car horn honking at 3.639-7.961", "frequencyCaption": "whistling two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4216.wav", "onoffCaption": "woman laughing at 0.33-3.611, 4.272-6.404", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4267.wav", "onoffCaption": "tapping clicking clanking at 2.832-6.272 and car horn honking at 5.367-7.832", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4282.wav", "onoffCaption": "dog barking at 1.217-3.217 and train horn at 6.035-9.515", "frequencyCaption": "dog barking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4399.wav", "onoffCaption": "door knocking at 1.174-3.637 and sheep goat bleating at 6.536-10.0", "frequencyCaption": "door knocking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4433.wav", "onoffCaption": "woman laughing at 0.612-2.98, 5.346-7.714", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4442.wav", "onoffCaption": "door knocking at 2.41-4.631, 5.512-7.733", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4466.wav", "onoffCaption": "thump thud at 0.065-2.565, 3.598-5.778, 6.412-8.912 and gunshot at 2.169-4.169, 4.865-6.865, 7.796-9.796", "frequencyCaption": "thump thud three times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4483.wav", "onoffCaption": "cow mooing at 0.202-3.171, 4.097-7.066, 7.97-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_4528.wav", "onoffCaption": "burping belching at 0.152-3.152, 5.315-8.315 and woman laughing at 3.008-5.613 and door slamming at 3.552-4.785, 6.323-7.556", "frequencyCaption": "burping belching two times and woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4559.wav", "onoffCaption": "tapping clicking clanking at 0.302-3.742, 4.486-6.714 and burping belching at 1.799-6.799", "frequencyCaption": "tapping clicking clanking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4598.wav", "onoffCaption": "duck quacking at 1.01-3.01, 3.848-5.848, 6.532-8.532", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4631.wav", "onoffCaption": "burping belching at 0.8-3.621, 5.735-8.514", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4640.wav", "onoffCaption": "cat meowing at 2.515-3.825, 4.843-6.593", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4937.wav", "onoffCaption": "spraying at 0.644-1.891, 3.986-4.986", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4946.wav", "onoffCaption": "car horn honking at 0.421-3.347, 3.927-6.853, 7.568-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4987.wav", "onoffCaption": "gunshot at 0.526-2.526 and dog barking at 5.505-7.505", "frequencyCaption": "gunshot one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_1.wav", "onoffCaption": "burping belching at 1.324-8.492", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2069.wav", "onoffCaption": "sheep goat bleating at 0.568-2.568, 3.305-5.305 and door slamming at 7.71-8.561", "frequencyCaption": "sheep goat bleating two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2073.wav", "onoffCaption": "dog barking at 0.81-2.81, 4.698-6.698 and sneeze at 1.441-3.687, 5.397-7.873", "frequencyCaption": "dog barking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2096.wav", "onoffCaption": "sneeze at 0.54-1.643 and car horn honking at 0.905-5.305, 6.304-10.0", "frequencyCaption": "sneeze one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2168.wav", "onoffCaption": "woman laughing at 2.695-5.976, 6.949-9.534", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2172.wav", "onoffCaption": "woman laughing at 2.653-6.229", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2197.wav", "onoffCaption": "woman laughing at 0.631-3.05, 5.474-7.715", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2301.wav", "onoffCaption": "door slamming at 0.391-2.309, 3.454-5.372 and sneeze at 2.098-5.208", "frequencyCaption": "door slamming two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2425.wav", "onoffCaption": "car horn honking at 0.069-3.564, 5.485-8.276", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2524.wav", "onoffCaption": "cow mooing at 3.119-7.548", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2656.wav", "onoffCaption": "tapping clicking clanking at 0.232-3.672, 4.981-8.421", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2757.wav", "onoffCaption": "car horn honking at 1.995-6.395, 7.171-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2820.wav", "onoffCaption": "burping belching at 0.103-2.138, 3.472-5.507, 6.832-8.867", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2875.wav", "onoffCaption": "door knocking at 1.909-6.442", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2921.wav", "onoffCaption": "sheep goat bleating at 1.122-3.122, 4.877-6.877", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4131.wav", "onoffCaption": "train horn at 0.546-3.22 and sheep goat bleating at 1.516-3.516, 4.159-6.159, 7.47-9.47", "frequencyCaption": "train horn one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4243.wav", "onoffCaption": "woman laughing at 1.329-3.411, 5.369-7.451", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4259.wav", "onoffCaption": "tapping clicking clanking at 0.446-3.886, 5.188-7.422 and spraying at 5.404-6.404", "frequencyCaption": "tapping clicking clanking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4342.wav", "onoffCaption": "woman laughing at 1.285-4.357 and sneeze at 1.929-4.405, 5.129-7.605", "frequencyCaption": "woman laughing one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4358.wav", "onoffCaption": "dog barking at 1.161-3.161, 5.169-7.169", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4499.wav", "onoffCaption": "car horn honking at 0.823-5.73 and door slamming at 2.357-3.748, 5.822-7.213 and duck quacking at 3.094-5.094, 7.242-9.242", "frequencyCaption": "car horn honking one times and door slamming two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4567.wav", "onoffCaption": "dog barking at 2.991-4.991, 7.131-9.131", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4582.wav", "onoffCaption": "explosion at 0.407-3.125 and gunshot at 3.084-5.084, 6.253-8.253 and burping belching at 4.468-7.647", "frequencyCaption": "explosion one times and gunshot two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4615.wav", "onoffCaption": "tapping clicking clanking at 0.349-3.789, 5.332-7.939 and cow mooing at 2.213-5.511, 7.329-10.0", "frequencyCaption": "tapping clicking clanking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4714.wav", "onoffCaption": "thump thud at 1.441-4.488, 5.626-8.673", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4863.wav", "onoffCaption": "sheep goat bleating at 2.651-5.731, 7.327-9.327", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4879.wav", "onoffCaption": "sneeze at 1.717-3.005, 4.541-5.829", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4886.wav", "onoffCaption": "woman laughing at 2.456-5.048, 5.701-8.293", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4962.wav", "onoffCaption": "spraying at 0.625-2.32, 2.931-3.798", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4978.wav", "onoffCaption": "door knocking at 1.907-4.095", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2098.wav", "onoffCaption": "duck quacking at 1.532-3.532", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2127.wav", "onoffCaption": "explosion at 3.111-5.983, 6.942-9.582", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2183.wav", "onoffCaption": "thump thud at 2.161-5.208, 6.064-8.472", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2315.wav", "onoffCaption": "explosion at 1.424-6.424 and cat meowing at 2.368-4.316 and car horn honking at 2.424-4.424", "frequencyCaption": "explosion one times and cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2354.wav", "onoffCaption": "door knocking at 0.652-4.152, 5.012-7.135, 7.746-9.81", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2571.wav", "onoffCaption": "spraying at 2.456-4.584, 5.711-7.473", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2594.wav", "onoffCaption": "car horn honking at 2.033-4.533, 5.2-7.7", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2619.wav", "onoffCaption": "duck quacking at 1.522-3.522 and cat meowing at 1.603-3.791", "frequencyCaption": "duck quacking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2658.wav", "onoffCaption": "explosion at 0.811-5.677, 7.813-10.0 and door knocking at 2.02-5.77", "frequencyCaption": "explosion two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2702.wav", "onoffCaption": "whistling at 0.515-3.39, 3.92-6.795 and explosion at 2.138-5.01, 6.901-9.773", "frequencyCaption": "whistling two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2743.wav", "onoffCaption": "tapping clicking clanking at 2.399-5.839, 6.987-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2834.wav", "onoffCaption": "cow mooing at 2.952-5.962, 7.639-10.0 and woman laughing at 3.706-6.406", "frequencyCaption": "cow mooing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2890.wav", "onoffCaption": "duck quacking at 2.845-4.845", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4125.wav", "onoffCaption": "thump thud at 1.182-3.682 and door knocking at 6.723-9.57", "frequencyCaption": "thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4164.wav", "onoffCaption": "whistling at 3.425-5.654", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4181.wav", "onoffCaption": "car horn honking at 0.313-3.488, 5.773-8.948", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4317.wav", "onoffCaption": "sneeze at 0.088-2.201, 3.06-5.377, 6.262-7.496 and door slamming at 0.488-2.708, 3.344-4.244 and train horn at 0.872-6.587, 7.647-10.0", "frequencyCaption": "sneeze three times and door slamming two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4356.wav", "onoffCaption": "train horn at 2.962-6.722", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4429.wav", "onoffCaption": "duck quacking at 0.833-2.833, 3.531-5.531, 6.133-8.133 and woman laughing at 5.524-8.624", "frequencyCaption": "duck quacking three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4468.wav", "onoffCaption": "door slamming at 0.051-2.509", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4532.wav", "onoffCaption": "door slamming at 2.31-4.793, 5.41-6.701", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4573.wav", "onoffCaption": "burping belching at 1.245-5.114 and door slamming at 2.463-4.463, 5.179-7.179", "frequencyCaption": "burping belching one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4596.wav", "onoffCaption": "dog barking at 0.023-2.023, 2.648-4.648, 6.804-8.804", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4700.wav", "onoffCaption": "door slamming at 1.79-4.01 and door knocking at 7.65-10.0", "frequencyCaption": "door slamming one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4741.wav", "onoffCaption": "sneeze at 3.285-7.785", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4836.wav", "onoffCaption": "car horn honking at 0.067-2.993, 3.783-6.709, 7.913-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4877.wav", "onoffCaption": "explosion at 0.204-2.524, 3.236-5.41, 6.621-8.623", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_4892.wav", "onoffCaption": "thump thud at 0.249-3.916, 5.15-8.817 and cat meowing at 0.688-1.697, 2.233-3.242, 3.778-4.787", "frequencyCaption": "thump thud two times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4989.wav", "onoffCaption": "sheep goat bleating at 0.79-2.79 and woman laughing at 1.659-4.251, 4.803-7.395 and gunshot at 3.514-5.514, 6.423-8.696", "frequencyCaption": "sheep goat bleating one times and woman laughing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2028.wav", "onoffCaption": "tapping clicking clanking at 3.282-6.722", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2032.wav", "onoffCaption": "thump thud at 1.878-4.217", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2129.wav", "onoffCaption": "cat meowing at 2.578-3.663, 4.726-6.036, 7.089-8.089", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2133.wav", "onoffCaption": "train horn at 3.33-6.57, 7.334-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2166.wav", "onoffCaption": "cow mooing at 2.414-6.843, 7.633-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2241.wav", "onoffCaption": "whistling at 3.139-8.314", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2340.wav", "onoffCaption": "car horn honking at 0.508-4.349 and explosion at 7.279-9.281", "frequencyCaption": "car horn honking one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2464.wav", "onoffCaption": "tapping clicking clanking at 0.896-4.336 and duck quacking at 1.153-3.153, 4.089-6.089, 6.905-8.905", "frequencyCaption": "tapping clicking clanking one times and duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2481.wav", "onoffCaption": "spraying at 1.831-2.331 and door slamming at 4.61-6.61, 7.521-9.521", "frequencyCaption": "spraying one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2530.wav", "onoffCaption": "train horn at 3.401-6.641", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2565.wav", "onoffCaption": "sneeze at 0.005-1.505", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2580.wav", "onoffCaption": "gunshot at 3.815-5.815, 7.3-9.3", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2617.wav", "onoffCaption": "duck quacking at 1.56-3.56, 4.821-6.821", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2716.wav", "onoffCaption": "sheep goat bleating at 0.826-2.826 and thump thud at 5.91-8.41", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2861.wav", "onoffCaption": "sneeze at 2.056-7.056 and tapping clicking clanking at 4.55-7.99 and thump thud at 5.027-7.798", "frequencyCaption": "sneeze one times and tapping clicking clanking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2884.wav", "onoffCaption": "door knocking at 0.648-2.951, 4.123-6.605, 7.909-9.931", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2960.wav", "onoffCaption": "woman laughing at 0.175-2.257 and door slamming at 0.823-2.823 and sneeze at 6.102-7.559", "frequencyCaption": "woman laughing one times and door slamming one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2985.wav", "onoffCaption": "thump thud at 0.261-2.6, 3.314-5.653 and sneeze at 4.087-6.172", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4071.wav", "onoffCaption": "cat meowing at 2.995-4.097, 4.741-5.843", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4094.wav", "onoffCaption": "whistling at 3.031-9.342", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4170.wav", "onoffCaption": "thump thud at 1.319-5.237 and cat meowing at 4.051-5.587", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4195.wav", "onoffCaption": "whistling at 2.822-5.051, 5.722-8.376 and explosion at 6.118-8.836", "frequencyCaption": "whistling two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4202.wav", "onoffCaption": "whistling at 2.434-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4218.wav", "onoffCaption": "cow mooing at 0.267-3.565 and woman laughing at 1.052-3.657", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4303.wav", "onoffCaption": "tapping clicking clanking at 0.66-4.1, 6.49-9.93 and cat meowing at 1.059-2.603, 4.596-6.14 and sheep goat bleating at 3.085-5.085", "frequencyCaption": "tapping clicking clanking two times and cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4319.wav", "onoffCaption": "door slamming at 3.854-5.003, 5.659-7.885", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4427.wav", "onoffCaption": "sheep goat bleating at 3.094-5.094, 6.139-8.139", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4526.wav", "onoffCaption": "spraying at 0.254-1.335, 3.355-5.05", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4654.wav", "onoffCaption": "cat meowing at 3.661-6.565, 7.709-8.895", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4755.wav", "onoffCaption": "cow mooing at 4.05-7.019 and spraying at 6.688-7.769, 9.086-9.713", "frequencyCaption": "cow mooing one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4822.wav", "onoffCaption": "train horn at 3.056-5.523 and car horn honking at 3.381-5.846 and burping belching at 4.458-7.458", "frequencyCaption": "train horn one times and car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4838.wav", "onoffCaption": "door knocking at 0.085-2.306 and spraying at 5.037-5.621 and duck quacking at 5.649-7.649", "frequencyCaption": "door knocking one times and spraying one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4923.wav", "onoffCaption": "gunshot at 0.444-2.444, 3.118-5.248 and explosion at 2.474-7.474", "frequencyCaption": "gunshot two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4939.wav", "onoffCaption": "gunshot at 2.558-5.032, 5.941-7.941", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2016.wav", "onoffCaption": "whistling at 0.036-5.211 and burping belching at 2.935-5.058, 6.513-9.513", "frequencyCaption": "whistling one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2067.wav", "onoffCaption": "sneeze at 0.627-2.69", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2082.wav", "onoffCaption": "spraying at 2.94-5.959", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2158.wav", "onoffCaption": "whistling at 0.357-8.012 and gunshot at 3.468-5.468, 7.306-9.306", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2199.wav", "onoffCaption": "thump thud at 3.283-6.95", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2214.wav", "onoffCaption": "cat meowing at 0.151-1.417, 3.482-4.748", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2265.wav", "onoffCaption": "whistling at 1.369-5.853 and tapping clicking clanking at 2.623-6.063, 7.495-10.0", "frequencyCaption": "whistling one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2280.wav", "onoffCaption": "burping belching at 0.359-5.359, 6.238-8.564 and tapping clicking clanking at 3.033-6.473", "frequencyCaption": "burping belching two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2415.wav", "onoffCaption": "car horn honking at 0.386-3.881, 5.676-9.171", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2431.wav", "onoffCaption": "explosion at 0.062-5.062, 7.284-9.469", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2440.wav", "onoffCaption": "door slamming at 0.123-0.623, 2.142-2.642 and sheep goat bleating at 1.348-3.348, 3.893-5.893, 6.837-8.837", "frequencyCaption": "door slamming two times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2633.wav", "onoffCaption": "whistling at 0.759-3.634, 4.275-6.577", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2642.wav", "onoffCaption": "tapping clicking clanking at 1.34-4.78, 7.05-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2728.wav", "onoffCaption": "cat meowing at 0.493-3.523 and woman laughing at 0.638-2.73", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2759.wav", "onoffCaption": "gunshot at 1.592-3.611", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2935.wav", "onoffCaption": "door knocking at 0.728-4.478 and duck quacking at 2.452-4.452, 5.784-7.784", "frequencyCaption": "door knocking one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2944.wav", "onoffCaption": "door slamming at 3.33-5.33", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4024.wav", "onoffCaption": "car horn honking at 2.651-5.164, 5.914-8.427 and thump thud at 3.286-6.057, 7.195-9.966", "frequencyCaption": "car horn honking two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4055.wav", "onoffCaption": "tapping clicking clanking at 2.017-5.457, 7.274-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4226.wav", "onoffCaption": "tapping clicking clanking at 0.981-4.421, 5.748-8.421", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4257.wav", "onoffCaption": "cow mooing at 2.78-5.762", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4296.wav", "onoffCaption": "cat meowing at 0.666-2.626, 3.341-5.301", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4403.wav", "onoffCaption": "door slamming at 2.323-3.626, 4.97-6.273, 7.601-8.904", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4472.wav", "onoffCaption": "door knocking at 2.045-4.422, 5.41-7.787", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4497.wav", "onoffCaption": "whistling at 2.277-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4518.wav", "onoffCaption": "spraying at 3.256-4.503, 5.811-7.058, 8.236-9.483", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4569.wav", "onoffCaption": "train horn at 2.431-5.965, 7.28-9.954", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4601.wav", "onoffCaption": "cow mooing at 0.4-5.38, 7.824-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4670.wav", "onoffCaption": "whistling at 3.544-9.044", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4695.wav", "onoffCaption": "burping belching at 0.668-4.668, 6.056-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4888.wav", "onoffCaption": "burping belching at 3.083-6.083", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4907.wav", "onoffCaption": "whistling at 1.293-9.304", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4976.wav", "onoffCaption": "tapping clicking clanking at 0.552-3.992, 5.863-9.303", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4993.wav", "onoffCaption": "train horn at 0.195-2.869 and cat meowing at 6.247-7.291, 8.178-9.488", "frequencyCaption": "train horn one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2043.wav", "onoffCaption": "cat meowing at 3.812-5.752", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2059.wav", "onoffCaption": "burping belching at 0.09-2.197 and door knocking at 0.721-3.788, 5.023-8.09", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2142.wav", "onoffCaption": "door slamming at 0.363-1.616, 2.579-4.805, 6.932-8.223", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2230.wav", "onoffCaption": "train horn at 0.572-6.632", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2331.wav", "onoffCaption": "thump thud at 0.794-3.022, 3.604-5.832, 6.585-8.813", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2364.wav", "onoffCaption": "explosion at 2.299-5.293, 6.682-9.682", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2381.wav", "onoffCaption": "woman laughing at 0.245-2.327, 2.828-5.334, 7.526-9.642", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_2514.wav", "onoffCaption": "gunshot at 3.315-5.315, 7.012-9.012", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2629.wav", "onoffCaption": "spraying at 0.982-2.715, 4.089-5.822 and cow mooing at 1.108-6.088, 7.379-10.0", "frequencyCaption": "spraying two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2666.wav", "onoffCaption": "thump thud at 0.125-2.896, 4.014-6.785, 7.495-10.0 and woman laughing at 2.778-4.978", "frequencyCaption": "thump thud three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2683.wav", "onoffCaption": "train horn at 2.278-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2699.wav", "onoffCaption": "sheep goat bleating at 3.383-5.383, 7.401-9.401", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2732.wav", "onoffCaption": "cow mooing at 3.624-8.604", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2767.wav", "onoffCaption": "dog barking at 2.432-4.432, 5.383-7.383", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2782.wav", "onoffCaption": "door knocking at 0.89-4.01, 5.434-7.971 and burping belching at 4.746-7.072", "frequencyCaption": "door knocking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2798.wav", "onoffCaption": "thump thud at 0.059-2.559, 3.231-5.68, 6.442-8.781 and spraying at 2.588-3.096", "frequencyCaption": "thump thud three times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2810.wav", "onoffCaption": "door slamming at 2.837-3.642", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2911.wav", "onoffCaption": "gunshot at 1.534-3.704, 6.11-8.11", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4000.wav", "onoffCaption": "burping belching at 2.612-6.118, 6.622-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4101.wav", "onoffCaption": "whistling at 0.049-7.142 and door slamming at 1.786-3.786, 5.681-7.681", "frequencyCaption": "whistling one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4269.wav", "onoffCaption": "car horn honking at 0.1-2.565", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4273.wav", "onoffCaption": "woman laughing at 3.317-5.672", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4368.wav", "onoffCaption": "sneeze at 3.702-5.615, 6.946-8.859", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4372.wav", "onoffCaption": "cat meowing at 1.592-2.592, 3.323-4.333, 4.964-7.647", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4397.wav", "onoffCaption": "dog barking at 0.044-2.444 and cow mooing at 6.758-10.0", "frequencyCaption": "dog barking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4419.wav", "onoffCaption": "car horn honking at 3.16-5.946, 6.874-9.66", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4456.wav", "onoffCaption": "cat meowing at 3.282-4.592, 6.073-8.963 and woman laughing at 3.328-5.933", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4557.wav", "onoffCaption": "car horn honking at 0.072-3.913, 6.076-9.917 and duck quacking at 0.139-2.139, 3.74-5.74", "frequencyCaption": "car horn honking two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4625.wav", "onoffCaption": "woman laughing at 0.112-3.5, 4.993-7.276 and cow mooing at 2.294-7.274", "frequencyCaption": "woman laughing two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4724.wav", "onoffCaption": "whistling at 3.107-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4806.wav", "onoffCaption": "explosion at 1.533-6.399", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4849.wav", "onoffCaption": "burping belching at 2.476-5.237, 6.847-9.668", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4853.wav", "onoffCaption": "car horn honking at 2.537-4.884, 6.616-8.963", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4948.wav", "onoffCaption": "woman laughing at 0.01-3.11 and door slamming at 0.11-2.593, 4.13-7.091", "frequencyCaption": "woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4952.wav", "onoffCaption": "tapping clicking clanking at 2.225-5.665", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2110.wav", "onoffCaption": "whistling at 0.11-8.121 and gunshot at 0.66-2.66, 4.32-6.32", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2117.wav", "onoffCaption": "duck quacking at 2.303-4.303, 4.821-6.821, 7.375-9.375", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2145.wav", "onoffCaption": "car horn honking at 2.249-5.067, 6.86-9.325", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2278.wav", "onoffCaption": "car horn honking at 0.005-3.18 and cat meowing at 0.584-2.532 and sneeze at 1.507-2.784", "frequencyCaption": "car horn honking one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2336.wav", "onoffCaption": "burping belching at 2.727-6.75, 7.974-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2363.wav", "onoffCaption": "explosion at 0.209-2.937, 4.197-6.925", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2408.wav", "onoffCaption": "duck quacking at 0.919-2.919, 4.192-6.192, 6.918-8.918", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2513.wav", "onoffCaption": "tapping clicking clanking at 1.575-5.015, 5.631-7.968", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2541.wav", "onoffCaption": "door knocking at 2.431-7.431 and spraying at 3.38-3.984, 6.273-8.401", "frequencyCaption": "door knocking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2546.wav", "onoffCaption": "car horn honking at 0.134-3.629, 4.572-6.776", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2735.wav", "onoffCaption": "train horn at 2.719-7.049", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2760.wav", "onoffCaption": "woman laughing at 0.243-2.829 and whistling at 6.242-10.0", "frequencyCaption": "woman laughing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2785.wav", "onoffCaption": "sheep goat bleating at 0.364-3.364 and gunshot at 2.49-4.49, 5.326-7.326 and thump thud at 4.036-6.375, 7.809-10.0", "frequencyCaption": "sheep goat bleating one times and gunshot two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2817.wav", "onoffCaption": "door slamming at 0.034-2.162, 3.44-6.414", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2842.wav", "onoffCaption": "woman laughing at 3.832-6.398, 7.087-9.653 and thump thud at 3.997-7.664", "frequencyCaption": "woman laughing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2845.wav", "onoffCaption": "burping belching at 0.066-2.189 and dog barking at 4.94-6.94, 7.927-9.927", "frequencyCaption": "burping belching one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2959.wav", "onoffCaption": "thump thud at 2.464-6.131", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4106.wav", "onoffCaption": "burping belching at 2.396-5.396, 6.29-9.29", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4154.wav", "onoffCaption": "burping belching at 0.1-3.38 and duck quacking at 4.7-6.7", "frequencyCaption": "burping belching one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4320.wav", "onoffCaption": "woman laughing at 1.992-4.347 and sheep goat bleating at 2.326-4.326, 6.491-8.491", "frequencyCaption": "woman laughing one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4327.wav", "onoffCaption": "tapping clicking clanking at 0.526-3.966, 4.575-8.015", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4375.wav", "onoffCaption": "car horn honking at 1.515-5.837, 7.812-9.812", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4390.wav", "onoffCaption": "door knocking at 0.626-2.978, 3.775-6.087, 6.804-9.392", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4502.wav", "onoffCaption": "duck quacking at 2.496-4.496 and car horn honking at 2.541-5.041, 5.542-8.042", "frequencyCaption": "duck quacking one times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4550.wav", "onoffCaption": "gunshot at 0.379-2.379, 3.425-5.425 and cat meowing at 6.709-9.613", "frequencyCaption": "gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4638.wav", "onoffCaption": "tapping clicking clanking at 2.149-5.589, 6.582-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4688.wav", "onoffCaption": "train horn at 1.924-4.564", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4723.wav", "onoffCaption": "train horn at 0.301-8.501 and spraying at 1.872-2.813, 3.956-5.718, 6.281-7.062", "frequencyCaption": "train horn one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4771.wav", "onoffCaption": "thump thud at 2.044-6.494, 7.6-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4776.wav", "onoffCaption": "burping belching at 1.744-5.288, 7.129-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4793.wav", "onoffCaption": "duck quacking at 0.789-2.789, 3.91-5.91", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4794.wav", "onoffCaption": "thump thud at 3.096-7.471", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4801.wav", "onoffCaption": "sheep goat bleating at 2.842-4.842, 5.542-7.542", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4854.wav", "onoffCaption": "sneeze at 0.166-1.711, 3.091-4.185", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2011.wav", "onoffCaption": "cow mooing at 3.455-7.884", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2044.wav", "onoffCaption": "cow mooing at 2.752-6.05", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2237.wav", "onoffCaption": "spraying at 2.505-3.68, 5.61-8.07 and burping belching at 4.712-7.335", "frequencyCaption": "spraying two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2246.wav", "onoffCaption": "train horn at 2.386-4.866, 5.514-7.994", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2262.wav", "onoffCaption": "door slamming at 0.296-1.82, 2.403-3.927, 4.953-6.477", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2287.wav", "onoffCaption": "sneeze at 2.468-4.08, 4.643-6.255, 7.115-8.727", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2379.wav", "onoffCaption": "thump thud at 1.14-3.602, 4.287-6.749", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2386.wav", "onoffCaption": "sneeze at 0.998-3.401, 4.07-5.347", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2412.wav", "onoffCaption": "whistling at 2.899-8.074", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2447.wav", "onoffCaption": "explosion at 0.429-5.429, 6.245-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2509.wav", "onoffCaption": "car horn honking at 0.405-3.9, 5.336-7.484", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2610.wav", "onoffCaption": "spraying at 0.285-2.721, 3.874-6.31, 7.014-9.45", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2634.wav", "onoffCaption": "cow mooing at 1.887-6.316 and dog barking at 2.493-4.493, 5.504-7.504", "frequencyCaption": "cow mooing one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2661.wav", "onoffCaption": "spraying at 0.128-1.823, 2.851-4.337, 5.4-6.05 and duck quacking at 1.565-3.565, 5.072-7.072 and dog barking at 1.874-3.874", "frequencyCaption": "spraying three times and duck quacking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2684.wav", "onoffCaption": "sheep goat bleating at 0.422-2.422", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2858.wav", "onoffCaption": "spraying at 1.549-2.281, 3.504-4.236, 4.962-5.694", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2916.wav", "onoffCaption": "train horn at 1.358-4.598, 5.88-9.12", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2943.wav", "onoffCaption": "door slamming at 2.913-4.913, 6.059-8.059", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4007.wav", "onoffCaption": "train horn at 0.741-4.221, 6.423-9.063 and cat meowing at 2.233-3.593, 5.797-7.692", "frequencyCaption": "train horn two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4048.wav", "onoffCaption": "woman laughing at 3.569-6.364 and sneeze at 4.859-6.855", "frequencyCaption": "woman laughing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4052.wav", "onoffCaption": "duck quacking at 2.19-4.19, 6.433-8.433", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4076.wav", "onoffCaption": "thump thud at 0.403-3.174 and sheep goat bleating at 2.8-4.8, 6.068-8.068", "frequencyCaption": "thump thud one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4093.wav", "onoffCaption": "door knocking at 3.703-5.924, 7.166-9.387", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4149.wav", "onoffCaption": "whistling at 3.415-9.124", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4153.wav", "onoffCaption": "door knocking at 0.839-3.569, 4.834-7.564 and sneeze at 1.952-4.037", "frequencyCaption": "door knocking two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4188.wav", "onoffCaption": "duck quacking at 2.14-4.14", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4221.wav", "onoffCaption": "whistling at 0.086-8.471", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4274.wav", "onoffCaption": "cat meowing at 0.042-1.144, 2.589-3.598 and woman laughing at 0.866-3.452, 4.245-6.831", "frequencyCaption": "cat meowing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4291.wav", "onoffCaption": "thump thud at 1.138-5.588 and whistling at 1.843-6.327 and woman laughing at 5.378-7.797", "frequencyCaption": "thump thud one times and whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4404.wav", "onoffCaption": "burping belching at 2.51-5.51, 7.101-9.541 and gunshot at 2.836-4.836", "frequencyCaption": "burping belching two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4420.wav", "onoffCaption": "duck quacking at 2.843-4.843, 5.486-7.486 and thump thud at 5.54-8.311", "frequencyCaption": "duck quacking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4451.wav", "onoffCaption": "train horn at 0.239-3.039, 5.026-7.426", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4505.wav", "onoffCaption": "door knocking at 1.084-3.654", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4622.wav", "onoffCaption": "whistling at 2.926-5.155, 7.141-9.447", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4677.wav", "onoffCaption": "cow mooing at 3.047-7.476 and woman laughing at 3.474-6.269", "frequencyCaption": "cow mooing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4692.wav", "onoffCaption": "duck quacking at 0.013-2.013", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4739.wav", "onoffCaption": "door slamming at 1.895-3.813", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4789.wav", "onoffCaption": "sneeze at 1.002-5.058", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4900.wav", "onoffCaption": "cow mooing at 0.223-3.233, 4.638-7.648 and door knocking at 4.601-6.761", "frequencyCaption": "cow mooing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4924.wav", "onoffCaption": "burping belching at 0.009-5.953", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4955.wav", "onoffCaption": "door knocking at 0.394-5.096, 6.053-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_8.wav", "onoffCaption": "sheep goat bleating at 0.236-2.236, 3.143-5.143, 6.147-8.147", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2035.wav", "onoffCaption": "thump thud at 3.809-6.148, 7.831-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2060.wav", "onoffCaption": "sheep goat bleating at 0.166-5.046, 6.154-8.154 and explosion at 0.486-2.66, 3.478-6.478", "frequencyCaption": "sheep goat bleating two times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2085.wav", "onoffCaption": "burping belching at 3.03-5.395, 5.899-8.216", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2161.wav", "onoffCaption": "door knocking at 2.239-6.511", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2184.wav", "onoffCaption": "woman laughing at 2.236-4.473, 6.682-8.919 and sneeze at 2.47-6.97 and sheep goat bleating at 6.116-8.116", "frequencyCaption": "woman laughing two times and sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2213.wav", "onoffCaption": "duck quacking at 2.521-4.521, 6.283-8.283", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2308.wav", "onoffCaption": "cow mooing at 0.549-3.531, 5.902-8.609", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2436.wav", "onoffCaption": "thump thud at 3.198-6.865", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2463.wav", "onoffCaption": "whistling at 1.38-9.391", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2486.wav", "onoffCaption": "gunshot at 1.07-3.343", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2537.wav", "onoffCaption": "train horn at 2.728-7.586", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2578.wav", "onoffCaption": "gunshot at 2.509-4.509, 6.605-8.605", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2645.wav", "onoffCaption": "sheep goat bleating at 0.42-2.42, 4.869-6.869 and cat meowing at 2.605-3.915, 4.495-6.631", "frequencyCaption": "sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2829.wav", "onoffCaption": "whistling at 0.202-5.377", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2833.wav", "onoffCaption": "sneeze at 2.403-6.051, 7.994-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2899.wav", "onoffCaption": "door knocking at 0.036-2.66, 4.527-7.364 and dog barking at 0.341-2.341, 2.973-4.973", "frequencyCaption": "door knocking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2928.wav", "onoffCaption": "burping belching at 0.293-6.973 and woman laughing at 2.69-4.806, 5.463-7.568", "frequencyCaption": "burping belching one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2932.wav", "onoffCaption": "spraying at 0.047-0.651, 1.26-2.164, 3.361-4.608 and dog barking at 0.972-2.972, 3.608-5.608", "frequencyCaption": "spraying three times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2967.wav", "onoffCaption": "cow mooing at 2.971-5.94, 7.066-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2982.wav", "onoffCaption": "door knocking at 2.719-5.099", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4023.wav", "onoffCaption": "door knocking at 1.077-4.577, 5.819-7.908", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4039.wav", "onoffCaption": "thump thud at 2.056-4.556, 6.038-8.538", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4122.wav", "onoffCaption": "cat meowing at 3.331-4.342, 4.92-5.931, 6.61-7.621", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4138.wav", "onoffCaption": "door knocking at 2.83-5.182, 6.312-8.664", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4205.wav", "onoffCaption": "spraying at 0.092-2.484, 4.849-7.241", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4250.wav", "onoffCaption": "burping belching at 0.753-3.932", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4351.wav", "onoffCaption": "dog barking at 0.601-2.601, 5.002-7.002", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4475.wav", "onoffCaption": "burping belching at 1.004-7.684 and whistling at 1.432-5.916, 6.702-9.65 and gunshot at 2.55-4.55, 5.582-7.582", "frequencyCaption": "burping belching one times and whistling two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4490.wav", "onoffCaption": "door knocking at 2.648-4.775 and tapping clicking clanking at 7.088-10.0", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4574.wav", "onoffCaption": "sneeze at 4.1-6.346", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4606.wav", "onoffCaption": "thump thud at 1.232-4.003, 6.177-8.891", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4653.wav", "onoffCaption": "tapping clicking clanking at 2.647-6.087", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4707.wav", "onoffCaption": "explosion at 0.051-5.051", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4748.wav", "onoffCaption": "thump thud at 1.279-3.779, 5.309-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4895.wav", "onoffCaption": "gunshot at 3.74-6.214, 7.578-9.578", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4971.wav", "onoffCaption": "cat meowing at 0.262-2.256", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4994.wav", "onoffCaption": "cat meowing at 3.746-4.746, 5.536-6.536 and sheep goat bleating at 4.329-6.329", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2048.wav", "onoffCaption": "burping belching at 0.76-4.304, 5.102-8.646", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2134.wav", "onoffCaption": "spraying at 0.015-1.19, 2.932-5.368", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2153.wav", "onoffCaption": "spraying at 0.046-1.046 and explosion at 0.666-5.666, 6.451-10.0", "frequencyCaption": "spraying one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2209.wav", "onoffCaption": "gunshot at 2.013-4.013", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2312.wav", "onoffCaption": "explosion at 2.43-7.351", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2347.wav", "onoffCaption": "explosion at 0.017-3.017, 3.938-6.938 and dog barking at 1.463-3.463", "frequencyCaption": "explosion two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2479.wav", "onoffCaption": "burping belching at 1.689-7.776", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2505.wav", "onoffCaption": "thump thud at 1.734-6.109, 6.786-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2562.wav", "onoffCaption": "car horn honking at 0.787-5.694, 7.691-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2587.wav", "onoffCaption": "sheep goat bleating at 3.286-5.286", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2711.wav", "onoffCaption": "train horn at 2.669-4.824, 6.196-8.66", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2744.wav", "onoffCaption": "car horn honking at 0.047-2.973 and cat meowing at 6.678-8.219", "frequencyCaption": "car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2801.wav", "onoffCaption": "burping belching at 1.71-5.0", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2866.wav", "onoffCaption": "cat meowing at 0.382-2.57, 3.17-4.181, 4.892-6.078 and whistling at 0.515-8.9 and duck quacking at 5.413-7.413", "frequencyCaption": "cat meowing three times and whistling one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2883.wav", "onoffCaption": "sheep goat bleating at 3.204-5.204", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2998.wav", "onoffCaption": "spraying at 0.088-1.345, 3.612-4.869", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4089.wav", "onoffCaption": "sneeze at 0.337-1.591 and train horn at 3.809-9.524", "frequencyCaption": "sneeze one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4177.wav", "onoffCaption": "dog barking at 0.177-2.177 and sneeze at 0.871-3.485, 5.324-7.465 and woman laughing at 3.376-6.014, 7.273-10.0", "frequencyCaption": "dog barking one times and sneeze two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4192.wav", "onoffCaption": "woman laughing at 0.638-3.433, 5.851-8.646", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4278.wav", "onoffCaption": "gunshot at 1.428-3.521, 4.261-6.354, 7.169-9.262 and door slamming at 5.994-7.369", "frequencyCaption": "gunshot three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4304.wav", "onoffCaption": "train horn at 2.458-6.639, 7.492-9.932", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4363.wav", "onoffCaption": "thump thud at 0.301-2.64 and duck quacking at 0.65-2.65, 4.552-6.552", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4386.wav", "onoffCaption": "sneeze at 0.984-4.059, 5.924-8.999", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4521.wav", "onoffCaption": "sneeze at 0.704-2.663, 3.814-5.773", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4546.wav", "onoffCaption": "dog barking at 0.715-2.715, 5.2-7.2", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4591.wav", "onoffCaption": "sheep goat bleating at 2.507-4.507, 5.973-7.973", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4649.wav", "onoffCaption": "spraying at 0.22-1.087, 2.362-3.229 and cat meowing at 6.586-8.526", "frequencyCaption": "spraying two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4735.wav", "onoffCaption": "train horn at 0.231-5.308", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4752.wav", "onoffCaption": "cow mooing at 1.023-3.992, 4.756-7.738 and cat meowing at 2.326-3.87, 4.908-6.452", "frequencyCaption": "cow mooing two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4825.wav", "onoffCaption": "burping belching at 0.564-5.564, 7.134-9.46", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4870.wav", "onoffCaption": "thump thud at 1.655-6.03, 6.852-9.508", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2106.wav", "onoffCaption": "car horn honking at 1.789-4.302, 5.357-7.87", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2274.wav", "onoffCaption": "explosion at 0.13-5.051, 6.09-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2291.wav", "onoffCaption": "cat meowing at 1.616-2.628, 3.395-6.299, 7.041-8.143", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2320.wav", "onoffCaption": "gunshot at 0.53-2.803, 3.85-5.85, 7.066-9.066", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2375.wav", "onoffCaption": "door slamming at 1.481-2.872, 3.714-5.714, 6.695-7.844 and explosion at 5.251-8.438", "frequencyCaption": "door slamming three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2390.wav", "onoffCaption": "duck quacking at 0.077-2.077, 4.153-6.153", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2451.wav", "onoffCaption": "burping belching at 0.021-3.311 and cat meowing at 5.102-7.042", "frequencyCaption": "burping belching one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2550.wav", "onoffCaption": "train horn at 2.524-8.268", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2622.wav", "onoffCaption": "tapping clicking clanking at 0.613-4.053 and train horn at 0.753-3.393, 4.437-7.077", "frequencyCaption": "tapping clicking clanking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2638.wav", "onoffCaption": "gunshot at 0.324-2.324, 3.995-6.501 and whistling at 6.819-8.828", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2688.wav", "onoffCaption": "dog barking at 0.336-2.336, 3.264-5.264, 5.967-7.967 and tapping clicking clanking at 2.286-5.726", "frequencyCaption": "dog barking three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2723.wav", "onoffCaption": "spraying at 0.799-1.974, 2.594-3.178, 4.428-6.19 and woman laughing at 2.694-4.977, 6.495-9.48", "frequencyCaption": "spraying three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2739.wav", "onoffCaption": "spraying at 1.824-2.451, 3.203-4.898, 7.269-8.516", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2776.wav", "onoffCaption": "sheep goat bleating at 2.702-4.702, 6.792-8.792", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2793.wav", "onoffCaption": "spraying at 2.441-3.068, 5.222-5.849 and explosion at 7.549-10.0", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2854.wav", "onoffCaption": "cow mooing at 2.28-7.26", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2955.wav", "onoffCaption": "sneeze at 3.434-4.82, 5.419-6.805", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4044.wav", "onoffCaption": "duck quacking at 1.493-3.493, 4.339-6.339, 7.456-9.456", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4110.wav", "onoffCaption": "tapping clicking clanking at 1.135-4.575, 5.943-8.406", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4145.wav", "onoffCaption": "sheep goat bleating at 3.411-5.411, 6.238-8.238", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4237.wav", "onoffCaption": "spraying at 0.24-1.726", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4336.wav", "onoffCaption": "burping belching at 2.645-5.824, 7.891-9.922", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4408.wav", "onoffCaption": "explosion at 1.757-4.757, 5.479-8.479 and car horn honking at 6.199-8.199", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4412.wav", "onoffCaption": "cat meowing at 0.27-1.63", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4509.wav", "onoffCaption": "woman laughing at 0.787-4.363, 6.706-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4513.wav", "onoffCaption": "gunshot at 0.752-2.752, 3.852-5.852, 6.881-8.881", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4661.wav", "onoffCaption": "thump thud at 1.443-3.782, 5.47-7.809", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4684.wav", "onoffCaption": "door slamming at 2.704-4.079, 6.287-7.811", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4760.wav", "onoffCaption": "tapping clicking clanking at 0.533-3.973, 5.147-8.587", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4785.wav", "onoffCaption": "whistling at 0.652-8.663", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4817.wav", "onoffCaption": "train horn at 0.488-4.488 and sheep goat bleating at 0.833-2.833, 3.373-5.373, 5.952-7.952", "frequencyCaption": "train horn one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4842.wav", "onoffCaption": "tapping clicking clanking at 2.652-6.092 and whistling at 2.885-8.385 and sheep goat bleating at 6.014-8.014", "frequencyCaption": "tapping clicking clanking one times and whistling one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4916.wav", "onoffCaption": "duck quacking at 2.204-4.204, 6.278-8.278", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4959.wav", "onoffCaption": "cat meowing at 0.048-1.06, 1.852-2.864, 4.697-5.709 and train horn at 0.069-3.549, 4.338-7.276", "frequencyCaption": "cat meowing three times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4.wav", "onoffCaption": "spraying at 0.375-2.503 and car horn honking at 4.549-8.136", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2007.wav", "onoffCaption": "train horn at 3.118-6.278, 7.28-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2023.wav", "onoffCaption": "car horn honking at 4.143-7.638", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2052.wav", "onoffCaption": "spraying at 2.342-2.992 and cat meowing at 3.473-4.483, 5.231-6.258, 7.243-8.553", "frequencyCaption": "spraying one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2076.wav", "onoffCaption": "spraying at 0.727-1.508 and thump thud at 5.495-9.413", "frequencyCaption": "spraying one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2089.wav", "onoffCaption": "duck quacking at 0.81-2.81 and spraying at 3.644-4.144, 5.805-6.656", "frequencyCaption": "duck quacking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2093.wav", "onoffCaption": "sneeze at 0.898-2.175, 3.979-5.305 and spraying at 3.016-3.524, 4.306-4.814", "frequencyCaption": "sneeze two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2138.wav", "onoffCaption": "burping belching at 1.702-6.038, 7.211-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2149.wav", "onoffCaption": "door knocking at 1.85-6.683, 7.825-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2188.wav", "onoffCaption": "train horn at 3.407-9.876", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2205.wav", "onoffCaption": "thump thud at 3.709-6.171", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2221.wav", "onoffCaption": "woman laughing at 2.818-5.89, 7.129-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2250.wav", "onoffCaption": "explosion at 2.937-7.937", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2404.wav", "onoffCaption": "tapping clicking clanking at 2.396-5.836, 7.121-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2420.wav", "onoffCaption": "cat meowing at 0.79-2.15 and woman laughing at 5.494-7.61", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2475.wav", "onoffCaption": "door knocking at 3.062-5.374, 6.46-8.772 and woman laughing at 3.618-6.037, 7.169-9.518", "frequencyCaption": "door knocking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2490.wav", "onoffCaption": "cow mooing at 2.202-5.171, 6.377-8.411", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2606.wav", "onoffCaption": "door knocking at 2.842-5.61", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2653.wav", "onoffCaption": "cat meowing at 3.516-4.601", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2677.wav", "onoffCaption": "whistling at 0.283-4.767, 5.551-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2692.wav", "onoffCaption": "explosion at 0.361-3.417, 5.348-8.404", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2748.wav", "onoffCaption": "thump thud at 0.302-2.802, 4.125-6.625 and dog barking at 0.683-2.683", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2789.wav", "onoffCaption": "explosion at 1.018-4.571", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2825.wav", "onoffCaption": "door knocking at 0.964-4.732", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2900.wav", "onoffCaption": "train horn at 0.446-3.12, 4.402-6.539 and thump thud at 3.737-7.404", "frequencyCaption": "train horn two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2924.wav", "onoffCaption": "burping belching at 0.202-6.882", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2971.wav", "onoffCaption": "burping belching at 1.991-6.39, 7.29-9.52", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2994.wav", "onoffCaption": "tapping clicking clanking at 2.3-5.74, 6.598-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4011.wav", "onoffCaption": "sneeze at 2.458-3.561, 4.16-6.297", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4035.wav", "onoffCaption": "car horn honking at 3.885-6.703, 7.76-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4060.wav", "onoffCaption": "gunshot at 2.22-4.22, 4.777-6.777, 7.486-9.486", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4085.wav", "onoffCaption": "dog barking at 0.613-2.613, 3.151-5.151, 6.447-8.447", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4213.wav", "onoffCaption": "explosion at 0.298-5.298, 5.925-8.075", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4246.wav", "onoffCaption": "duck quacking at 1.745-3.745, 4.943-6.943", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4262.wav", "onoffCaption": "car horn honking at 0.025-2.025 and spraying at 3.726-4.353, 6.476-7.103, 8.766-9.393", "frequencyCaption": "car horn honking one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4287.wav", "onoffCaption": "door knocking at 2.855-5.207", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4308.wav", "onoffCaption": "explosion at 0.564-5.564, 7.629-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4379.wav", "onoffCaption": "dog barking at 0.247-2.247", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4436.wav", "onoffCaption": "gunshot at 3.193-5.433, 7.188-9.188", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4447.wav", "onoffCaption": "thump thud at 3.806-7.724", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4463.wav", "onoffCaption": "tapping clicking clanking at 0.406-3.846, 6.106-8.202", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4486.wav", "onoffCaption": "car horn honking at 3.658-6.584, 7.528-9.535", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4578.wav", "onoffCaption": "car horn honking at 1.804-5.023", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4610.wav", "onoffCaption": "sneeze at 0.396-3.604, 4.394-7.602", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4634.wav", "onoffCaption": "door slamming at 0.545-1.778 and duck quacking at 4.483-6.483", "frequencyCaption": "door slamming one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4645.wav", "onoffCaption": "door slamming at 0.033-1.324, 2.029-3.32, 4.632-5.923", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4829.wav", "onoffCaption": "car horn honking at 0.514-3.427, 4.395-7.321, 7.878-10.0", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_4858.wav", "onoffCaption": "train horn at 0.62-4.688, 5.25-9.318", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4899.wav", "onoffCaption": "explosion at 2.455-7.455", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4932.wav", "onoffCaption": "burping belching at 0.383-3.927, 5.815-9.359", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4943.wav", "onoffCaption": "explosion at 0.369-3.363, 4.428-7.086 and cow mooing at 1.762-5.06", "frequencyCaption": "explosion two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4967.wav", "onoffCaption": "door knocking at 1.44-5.59, 6.272-8.575 and cat meowing at 2.072-3.285, 3.8-5.013", "frequencyCaption": "door knocking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4982.wav", "onoffCaption": "door knocking at 0.035-2.535, 3.47-5.97, 6.806-9.306 and door slamming at 0.149-0.83, 1.334-2.185", "frequencyCaption": "door knocking three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2039.wav", "onoffCaption": "explosion at 3.189-7.029", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2122.wav", "onoffCaption": "spraying at 2.788-3.639, 4.382-5.233", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2177.wav", "onoffCaption": "cat meowing at 0.579-2.715 and door knocking at 1.119-4.672, 5.375-8.928", "frequencyCaption": "cat meowing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2192.wav", "onoffCaption": "whistling at 2.771-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2304.wav", "onoffCaption": "cat meowing at 0.014-1.55, 2.249-3.785, 5.054-6.59 and door knocking at 0.12-2.341, 4.211-6.432", "frequencyCaption": "cat meowing three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2351.wav", "onoffCaption": "train horn at 0.696-5.136 and door knocking at 6.603-9.978", "frequencyCaption": "train horn one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2521.wav", "onoffCaption": "door knocking at 0.54-6.6, 7.245-9.422", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2574.wav", "onoffCaption": "train horn at 0.214-5.072, 7.224-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2591.wav", "onoffCaption": "car horn honking at 0.359-3.145", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2649.wav", "onoffCaption": "car horn honking at 0.111-4.511 and thump thud at 2.579-5.626 and door slamming at 2.596-3.596, 4.32-5.3, 6.247-7.622", "frequencyCaption": "car horn honking one times and thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2707.wav", "onoffCaption": "whistling at 1.983-6.467 and gunshot at 3.437-5.437, 5.946-7.946", "frequencyCaption": "whistling one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2752.wav", "onoffCaption": "burping belching at 0.679-6.28", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2870.wav", "onoffCaption": "door knocking at 2.203-5.259, 7.501-9.947", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2895.wav", "onoffCaption": "cow mooing at 0.668-3.65, 4.571-7.553 and dog barking at 2.804-4.804, 6.928-9.366", "frequencyCaption": "cow mooing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4134.wav", "onoffCaption": "tapping clicking clanking at 2.236-5.676", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4184.wav", "onoffCaption": "gunshot at 2.648-4.648 and woman laughing at 7.575-10.0", "frequencyCaption": "gunshot one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4209.wav", "onoffCaption": "door slamming at 2.279-3.532 and thump thud at 6.078-8.306 and door knocking at 7.164-9.352", "frequencyCaption": "door slamming one times and thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4312.wav", "onoffCaption": "tapping clicking clanking at 0.624-4.064, 4.836-8.276", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4347.wav", "onoffCaption": "cat meowing at 1.85-4.754, 6.696-7.708", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4479.wav", "onoffCaption": "duck quacking at 3.224-5.224 and whistling at 7.322-10.0", "frequencyCaption": "duck quacking one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4562.wav", "onoffCaption": "sheep goat bleating at 0.675-2.675, 5.049-7.049 and spraying at 2.84-4.968", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4587.wav", "onoffCaption": "door slamming at 1.539-4.256, 5.225-7.942 and gunshot at 5.959-7.959", "frequencyCaption": "door slamming two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4711.wav", "onoffCaption": "cow mooing at 2.135-5.145, 6.111-8.674 and explosion at 3.265-8.265", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4744.wav", "onoffCaption": "explosion at 2.169-4.43, 5.09-7.351", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4833.wav", "onoffCaption": "duck quacking at 1.645-3.645, 5.743-7.743", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4866.wav", "onoffCaption": "sneeze at 0.481-5.01, 6.331-8.038", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4883.wav", "onoffCaption": "door slamming at 2.32-4.233, 5.384-7.384", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4928.wav", "onoffCaption": "thump thud at 0.16-2.622, 3.878-6.34", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4998.wav", "onoffCaption": "train horn at 0.32-2.787, 4.37-7.25 and sheep goat bleating at 3.461-6.781, 7.415-9.415", "frequencyCaption": "train horn two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2125.wav", "onoffCaption": "burping belching at 3.079-6.079, 7.222-9.897", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2170.wav", "onoffCaption": "spraying at 0.023-2.607, 3.651-4.255, 4.818-6.065 and duck quacking at 0.269-2.269", "frequencyCaption": "spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2195.wav", "onoffCaption": "burping belching at 0.087-3.346, 4.593-6.616 and explosion at 4.732-7.732", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2218.wav", "onoffCaption": "tapping clicking clanking at 2.113-5.553, 6.579-9.274", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2303.wav", "onoffCaption": "burping belching at 2.154-4.975, 5.67-8.491", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2356.wav", "onoffCaption": "cow mooing at 2.672-5.641, 7.662-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2526.wav", "onoffCaption": "cow mooing at 0.244-3.226, 4.079-6.625, 7.594-10.0 and car horn honking at 3.697-6.872", "frequencyCaption": "cow mooing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2700.wav", "onoffCaption": "dog barking at 1.524-3.524, 4.817-6.817", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2755.wav", "onoffCaption": "car horn honking at 1.663-5.317, 6.053-8.202", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2822.wav", "onoffCaption": "whistling at 2.19-5.065, 6.03-8.905", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2892.wav", "onoffCaption": "car horn honking at 0.074-3.728 and door slamming at 6.804-8.328", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2939.wav", "onoffCaption": "cow mooing at 1.111-4.093, 4.704-6.768, 7.68-10.0", "frequencyCaption": "cow mooing three times"}
+{"filepath": "data/multi_event_train/syn_2989.wav", "onoffCaption": "cat meowing at 2.02-3.575, 4.742-6.286 and door knocking at 2.836-6.452", "frequencyCaption": "cat meowing two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4028.wav", "onoffCaption": "whistling at 2.589-7.073", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4098.wav", "onoffCaption": "sheep goat bleating at 1.437-3.437, 3.974-6.969 and door knocking at 1.638-6.638, 7.29-10.0", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4133.wav", "onoffCaption": "explosion at 0.035-2.209, 4.651-7.242 and door slamming at 0.721-1.86, 2.689-4.454, 5.508-7.728", "frequencyCaption": "explosion two times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4161.wav", "onoffCaption": "sneeze at 2.452-4.928, 5.8-8.276", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4166.wav", "onoffCaption": "cat meowing at 1.548-3.438, 4.507-6.397, 7.079-8.969", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4183.wav", "onoffCaption": "burping belching at 1.619-5.125, 6.442-9.948", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4340.wav", "onoffCaption": "cow mooing at 0.835-3.817, 5.853-8.84", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4530.wav", "onoffCaption": "whistling at 1.37-10.0 and gunshot at 3.876-5.876", "frequencyCaption": "whistling one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4537.wav", "onoffCaption": "burping belching at 1.095-4.639, 5.725-8.07 and spraying at 1.935-2.802, 3.928-4.555", "frequencyCaption": "burping belching two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4565.wav", "onoffCaption": "burping belching at 1.951-4.712, 5.866-8.138", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4580.wav", "onoffCaption": "thump thud at 0.07-2.841, 5.122-7.893 and sheep goat bleating at 1.296-3.296", "frequencyCaption": "thump thud two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4716.wav", "onoffCaption": "burping belching at 3.69-7.25", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4834.wav", "onoffCaption": "door slamming at 2.952-5.952, 7.214-10.0", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4861.wav", "onoffCaption": "duck quacking at 0.25-2.25 and thump thud at 0.885-3.224 and door knocking at 6.572-10.0", "frequencyCaption": "duck quacking one times and thump thud one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4884.wav", "onoffCaption": "cat meowing at 0.347-1.357 and spraying at 0.846-2.021", "frequencyCaption": "cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_3.wav", "onoffCaption": "duck quacking at 0.203-2.203, 2.739-4.739 and train horn at 1.131-5.461, 6.713-9.193", "frequencyCaption": "duck quacking two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2000.wav", "onoffCaption": "burping belching at 0.839-3.839, 5.914-8.914", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2024.wav", "onoffCaption": "thump thud at 0.692-5.142", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2071.wav", "onoffCaption": "woman laughing at 2.728-5.147, 6.23-8.649", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2094.wav", "onoffCaption": "burping belching at 0.22-2.843, 3.448-5.813, 7.816-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2202.wav", "onoffCaption": "tapping clicking clanking at 0.947-4.387, 5.953-9.393 and woman laughing at 1.93-4.167", "frequencyCaption": "tapping clicking clanking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2257.wav", "onoffCaption": "door knocking at 0.06-3.676 and sneeze at 0.255-3.365 and door slamming at 2.1-3.119", "frequencyCaption": "door knocking one times and sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2296.wav", "onoffCaption": "gunshot at 0.013-2.013, 3.206-5.206, 6.18-8.18", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2319.wav", "onoffCaption": "thump thud at 0.838-3.3 and woman laughing at 2.775-5.57, 6.239-9.034", "frequencyCaption": "thump thud one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2427.wav", "onoffCaption": "sneeze at 2.616-3.79, 4.47-6.177 and cat meowing at 7.473-9.017", "frequencyCaption": "sneeze two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2456.wav", "onoffCaption": "door slamming at 0.534-3.017, 4.185-6.668, 7.849-10.0", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2468.wav", "onoffCaption": "dog barking at 0.272-2.272, 4.17-6.17, 7.135-9.135", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2472.wav", "onoffCaption": "woman laughing at 2.006-8.74", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2497.wav", "onoffCaption": "sneeze at 1.87-4.109, 4.688-6.927", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2569.wav", "onoffCaption": "car horn honking at 0.611-5.518, 7.265-10.0 and sheep goat bleating at 0.81-2.81, 4.698-6.698", "frequencyCaption": "car horn honking two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2573.wav", "onoffCaption": "dog barking at 0.815-2.815, 3.608-5.608 and explosion at 6.902-9.493", "frequencyCaption": "dog barking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2596.wav", "onoffCaption": "duck quacking at 0.302-2.302 and door slamming at 6.473-9.254", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2601.wav", "onoffCaption": "burping belching at 2.939-5.037 and spraying at 4.187-4.837, 5.94-7.949, 9.176-9.78", "frequencyCaption": "burping belching one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_2654.wav", "onoffCaption": "cat meowing at 0.362-1.778, 2.393-3.809, 4.914-6.33 and car horn honking at 3.445-6.231", "frequencyCaption": "cat meowing three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2838.wav", "onoffCaption": "door slamming at 1.159-3.522 and dog barking at 2.177-4.177, 4.702-6.702, 7.836-9.836", "frequencyCaption": "door slamming one times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2849.wav", "onoffCaption": "door knocking at 2.197-4.965, 5.543-8.39", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2877.wav", "onoffCaption": "woman laughing at 1.891-4.457, 5.083-7.451 and dog barking at 4.627-7.065, 7.622-10.0", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2888.wav", "onoffCaption": "gunshot at 3.402-5.402, 6.683-8.683", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2923.wav", "onoffCaption": "spraying at 2.881-3.785", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2952.wav", "onoffCaption": "duck quacking at 1.431-3.431, 5.499-7.499", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2976.wav", "onoffCaption": "dog barking at 0.106-2.106, 3.066-5.066, 7.405-9.405", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2993.wav", "onoffCaption": "spraying at 0.875-2.637, 3.389-4.389, 5.063-6.004 and door slamming at 2.355-4.268", "frequencyCaption": "spraying three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4032.wav", "onoffCaption": "car horn honking at 0.015-2.515, 3.464-5.964", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4067.wav", "onoffCaption": "cow mooing at 1.537-5.966 and burping belching at 3.187-5.81", "frequencyCaption": "cow mooing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4082.wav", "onoffCaption": "door knocking at 0.068-4.443 and duck quacking at 0.264-2.264 and dog barking at 2.234-4.234, 4.989-6.989", "frequencyCaption": "door knocking one times and duck quacking one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4129.wav", "onoffCaption": "tapping clicking clanking at 1.318-4.758 and sheep goat bleating at 3.041-5.041", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4199.wav", "onoffCaption": "dog barking at 0.158-2.158, 3.19-5.19, 7.467-9.467", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4214.wav", "onoffCaption": "train horn at 1.922-5.456", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4230.wav", "onoffCaption": "train horn at 0.323-3.563, 5.508-8.471", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4241.wav", "onoffCaption": "whistling at 0.723-3.698 and tapping clicking clanking at 5.638-9.078", "frequencyCaption": "whistling one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4315.wav", "onoffCaption": "burping belching at 1.063-7.743 and cat meowing at 6.9-7.921, 8.476-9.52", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4415.wav", "onoffCaption": "car horn honking at 0.981-5.381, 6.324-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4431.wav", "onoffCaption": "woman laughing at 0.175-6.909, 7.845-9.926", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4464.wav", "onoffCaption": "door slamming at 2.516-3.494, 4.587-5.527, 6.595-7.97", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4481.wav", "onoffCaption": "door knocking at 2.762-5.492, 6.815-8.975", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4617.wav", "onoffCaption": "gunshot at 1.616-3.616, 4.192-6.192, 7.032-9.032", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4642.wav", "onoffCaption": "explosion at 3.239-8.105", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4658.wav", "onoffCaption": "train horn at 1.537-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4666.wav", "onoffCaption": "cow mooing at 1.708-5.006, 6.168-9.466", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4683.wav", "onoffCaption": "tapping clicking clanking at 3.129-6.569", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4743.wav", "onoffCaption": "gunshot at 2.914-4.914, 7.35-9.35", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4759.wav", "onoffCaption": "sneeze at 3.36-5.473, 6.677-8.79", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4798.wav", "onoffCaption": "tapping clicking clanking at 0.768-4.208, 5.52-8.96", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4935.wav", "onoffCaption": "cow mooing at 1.516-4.814, 7.173-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4960.wav", "onoffCaption": "spraying at 1.552-2.799, 4.582-7.042 and train horn at 2.416-5.896", "frequencyCaption": "spraying two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4985.wav", "onoffCaption": "train horn at 0.388-4.388 and sneeze at 1.181-2.458, 4.879-6.113", "frequencyCaption": "train horn one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2055.wav", "onoffCaption": "door slamming at 0.237-1.215, 1.804-3.345, 4.557-7.436 and duck quacking at 0.321-2.321", "frequencyCaption": "door slamming three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2101.wav", "onoffCaption": "gunshot at 1.697-3.827", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2154.wav", "onoffCaption": "dog barking at 0.723-2.723, 3.316-5.316, 6.483-8.483", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2226.wav", "onoffCaption": "door slamming at 0.091-2.091, 2.603-3.454, 4.045-4.883", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2269.wav", "onoffCaption": "door slamming at 1.059-2.972 and gunshot at 2.674-4.674, 5.625-7.625", "frequencyCaption": "door slamming one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2273.wav", "onoffCaption": "woman laughing at 1.117-3.354, 3.88-6.117", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2327.wav", "onoffCaption": "door slamming at 1.389-4.363, 6.253-9.227", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2368.wav", "onoffCaption": "car horn honking at 2.647-7.159", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2372.wav", "onoffCaption": "whistling at 0.111-2.986", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2397.wav", "onoffCaption": "tapping clicking clanking at 0.796-4.236", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2403.wav", "onoffCaption": "explosion at 2.342-5.469 and gunshot at 7.933-10.0", "frequencyCaption": "explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2419.wav", "onoffCaption": "gunshot at 0.477-2.477, 3.149-5.149 and door knocking at 1.493-4.613", "frequencyCaption": "gunshot two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2502.wav", "onoffCaption": "gunshot at 0.827-2.827, 4.117-6.117", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2518.wav", "onoffCaption": "cow mooing at 0.012-2.981", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2557.wav", "onoffCaption": "sneeze at 3.357-5.818, 7.048-8.576", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2625.wav", "onoffCaption": "thump thud at 1.569-6.019, 6.528-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2670.wav", "onoffCaption": "cow mooing at 1.159-4.128, 5.227-8.196 and explosion at 2.413-7.413", "frequencyCaption": "cow mooing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2695.wav", "onoffCaption": "train horn at 0.6-4.781 and spraying at 0.721-1.453", "frequencyCaption": "train horn one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2724.wav", "onoffCaption": "woman laughing at 0.35-7.362 and tapping clicking clanking at 3.906-7.346", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2771.wav", "onoffCaption": "door slamming at 0.63-3.604, 4.655-7.629 and cat meowing at 2.295-4.235, 5.62-6.629 and duck quacking at 6.304-8.304", "frequencyCaption": "door slamming two times and cat meowing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2794.wav", "onoffCaption": "gunshot at 1.926-4.199, 5.42-7.42", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2806.wav", "onoffCaption": "burping belching at 0.386-2.589, 3.628-5.831, 6.503-8.706 and door knocking at 2.198-7.031", "frequencyCaption": "burping belching three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2853.wav", "onoffCaption": "thump thud at 1.981-4.443, 6.001-8.463", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2907.wav", "onoffCaption": "cow mooing at 0.749-3.759 and sheep goat bleating at 2.754-7.474 and dog barking at 3.083-5.083, 6.184-8.184", "frequencyCaption": "cow mooing one times and sheep goat bleating one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2948.wav", "onoffCaption": "cat meowing at 0.559-2.291 and car horn honking at 6.429-10.0", "frequencyCaption": "cat meowing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4016.wav", "onoffCaption": "sneeze at 2.058-3.289, 5.747-6.978 and door knocking at 4.891-7.194", "frequencyCaption": "sneeze two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4043.wav", "onoffCaption": "door knocking at 0.075-4.225 and train horn at 6.32-10.0", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4059.wav", "onoffCaption": "gunshot at 0.745-2.745, 4.879-6.879", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4117.wav", "onoffCaption": "cat meowing at 0.091-1.112, 2.04-4.57, 5.573-7.533", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4142.wav", "onoffCaption": "dog barking at 0.414-2.414, 3.557-5.557, 6.334-8.334 and train horn at 1.087-3.242", "frequencyCaption": "dog barking three times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4158.wav", "onoffCaption": "burping belching at 0.49-4.49, 5.449-9.449", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4265.wav", "onoffCaption": "dog barking at 0.175-5.812, 6.78-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4280.wav", "onoffCaption": "dog barking at 2.669-4.669 and sheep goat bleating at 7.512-9.512", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4331.wav", "onoffCaption": "cow mooing at 1.208-6.188 and duck quacking at 4.284-6.284", "frequencyCaption": "cow mooing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4364.wav", "onoffCaption": "burping belching at 0.19-3.19, 5.308-8.308 and dog barking at 0.267-2.267 and gunshot at 1.186-3.186, 5.249-7.249", "frequencyCaption": "burping belching two times and dog barking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4381.wav", "onoffCaption": "spraying at 0.23-1.925 and car horn honking at 4.134-8.646", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4440.wav", "onoffCaption": "cow mooing at 0.007-2.976, 5.371-7.46", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4514.wav", "onoffCaption": "tapping clicking clanking at 1.267-4.707, 5.587-9.027", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4541.wav", "onoffCaption": "dog barking at 2.832-4.832, 5.772-7.772", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4629.wav", "onoffCaption": "duck quacking at 0.526-2.526, 3.179-5.179", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4633.wav", "onoffCaption": "duck quacking at 0.812-2.812, 3.631-5.631, 6.45-8.45 and cat meowing at 2.522-4.077 and burping belching at 2.965-6.965", "frequencyCaption": "duck quacking three times and cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4699.wav", "onoffCaption": "door knocking at 3.099-5.121, 5.729-7.751", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4728.wav", "onoffCaption": "explosion at 0.713-5.634, 6.498-9.498", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4732.wav", "onoffCaption": "tapping clicking clanking at 2.353-5.793, 6.59-9.076", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4767.wav", "onoffCaption": "dog barking at 2.492-4.492, 6.811-8.811", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4782.wav", "onoffCaption": "whistling at 3.517-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4810.wav", "onoffCaption": "cat meowing at 2.171-3.171", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4845.wav", "onoffCaption": "door knocking at 2.751-5.321, 6.377-8.947", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4911.wav", "onoffCaption": "thump thud at 1.568-3.907, 4.45-6.789, 7.746-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4944.wav", "onoffCaption": "woman laughing at 0.34-2.54, 4.896-7.096", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2115.wav", "onoffCaption": "sheep goat bleating at 0.902-2.902, 4.596-6.596 and cow mooing at 1.699-4.709, 5.792-8.247", "frequencyCaption": "sheep goat bleating two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2140.wav", "onoffCaption": "burping belching at 0.728-3.728, 4.523-7.523 and car horn honking at 2.581-7.488", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2228.wav", "onoffCaption": "whistling at 1.17-6.67", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2298.wav", "onoffCaption": "car horn honking at 0.039-2.857, 3.365-6.183, 6.974-9.792", "frequencyCaption": "car horn honking three times"}
+{"filepath": "data/multi_event_train/syn_2333.wav", "onoffCaption": "gunshot at 3.701-5.701, 6.615-8.615", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2366.wav", "onoffCaption": "woman laughing at 2.103-4.328, 4.915-7.115", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2383.wav", "onoffCaption": "door slamming at 3.631-6.06, 6.913-7.764", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2458.wav", "onoffCaption": "duck quacking at 1.91-3.91, 5.691-7.691", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2516.wav", "onoffCaption": "tapping clicking clanking at 2.089-5.529 and train horn at 2.833-7.014", "frequencyCaption": "tapping clicking clanking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2543.wav", "onoffCaption": "cow mooing at 3.059-6.028, 7.252-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2730.wav", "onoffCaption": "dog barking at 0.421-2.421, 4.285-6.285", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2765.wav", "onoffCaption": "whistling at 0.918-3.793, 4.747-7.622", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2780.wav", "onoffCaption": "sneeze at 3.002-4.166, 4.977-7.343", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2812.wav", "onoffCaption": "train horn at 1.579-5.579 and sheep goat bleating at 3.344-5.344", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2847.wav", "onoffCaption": "dog barking at 0.439-2.439 and tapping clicking clanking at 5.168-8.608", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2909.wav", "onoffCaption": "duck quacking at 0.518-2.518, 4.844-6.844", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4018.wav", "onoffCaption": "spraying at 0.372-0.976, 2.675-3.302, 5.276-5.851 and car horn honking at 3.509-7.831", "frequencyCaption": "spraying three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4103.wav", "onoffCaption": "cat meowing at 2.286-5.316, 6.353-8.313 and woman laughing at 5.253-7.453 and sneeze at 5.34-6.666", "frequencyCaption": "cat meowing two times and woman laughing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4156.wav", "onoffCaption": "explosion at 0.661-3.661", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4325.wav", "onoffCaption": "gunshot at 0.211-2.211", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4370.wav", "onoffCaption": "gunshot at 1.844-3.844, 5.097-7.097, 7.762-9.762", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4395.wav", "onoffCaption": "dog barking at 0.207-2.607, 4.311-6.311", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4500.wav", "onoffCaption": "explosion at 0.364-5.23, 7.676-10.0 and burping belching at 1.243-5.579", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4555.wav", "onoffCaption": "door knocking at 0.82-5.353, 6.393-8.795", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4668.wav", "onoffCaption": "duck quacking at 1.603-3.603, 5.311-7.311", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4726.wav", "onoffCaption": "thump thud at 2.21-5.877", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4773.wav", "onoffCaption": "sheep goat bleating at 1.094-3.094, 4.244-6.244, 7.536-9.536", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4796.wav", "onoffCaption": "sneeze at 0.242-1.568, 2.708-4.034, 4.92-6.246", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4804.wav", "onoffCaption": "duck quacking at 0.366-2.366, 3.404-5.404 and woman laughing at 7.732-10.0", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4851.wav", "onoffCaption": "door slamming at 0.634-1.937, 3.148-4.451, 6.589-7.892", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4905.wav", "onoffCaption": "burping belching at 1.445-7.389 and cow mooing at 4.409-7.707", "frequencyCaption": "burping belching one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2014.wav", "onoffCaption": "duck quacking at 0.1-2.1, 2.681-4.681, 5.3-7.3 and woman laughing at 0.878-3.103, 5.052-7.548", "frequencyCaption": "duck quacking three times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2041.wav", "onoffCaption": "gunshot at 0.756-2.756, 4.93-6.93", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2232.wav", "onoffCaption": "cow mooing at 0.046-3.015 and sneeze at 0.117-1.704", "frequencyCaption": "cow mooing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2267.wav", "onoffCaption": "thump thud at 0.246-4.621 and sheep goat bleating at 1.704-3.704, 5.086-8.086 and dog barking at 4.822-6.822", "frequencyCaption": "thump thud one times and sheep goat bleating two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2282.wav", "onoffCaption": "spraying at 0.196-0.797", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2329.wav", "onoffCaption": "sneeze at 0.061-2.378 and gunshot at 0.877-2.877 and dog barking at 1.829-3.829", "frequencyCaption": "sneeze one times and gunshot one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2399.wav", "onoffCaption": "cow mooing at 0.806-5.235", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2417.wav", "onoffCaption": "gunshot at 2.087-4.087, 4.956-6.956, 7.941-9.941", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2442.wav", "onoffCaption": "door knocking at 3.887-6.289, 7.303-9.705", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2559.wav", "onoffCaption": "car horn honking at 1.034-4.529 and dog barking at 6.57-8.57", "frequencyCaption": "car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2631.wav", "onoffCaption": "burping belching at 0.394-3.596 and door slamming at 1.021-3.021 and thump thud at 5.105-9.555", "frequencyCaption": "burping belching one times and door slamming one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2664.wav", "onoffCaption": "car horn honking at 0.045-2.831 and thump thud at 6.644-9.415", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2681.wav", "onoffCaption": "woman laughing at 0.0-3.085 and door knocking at 5.304-8.804", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2808.wav", "onoffCaption": "cat meowing at 0.026-1.026, 1.84-2.84 and woman laughing at 4.173-6.456", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2913.wav", "onoffCaption": "woman laughing at 3.473-6.268", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2946.wav", "onoffCaption": "burping belching at 2.011-7.612", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4002.wav", "onoffCaption": "gunshot at 0.13-2.13 and woman laughing at 1.245-3.445, 5.933-8.133", "frequencyCaption": "gunshot one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4057.wav", "onoffCaption": "door knocking at 1.551-3.854, 6.338-8.641", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4119.wav", "onoffCaption": "gunshot at 4.016-6.016", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4224.wav", "onoffCaption": "duck quacking at 0.733-2.733, 3.793-5.793, 6.393-8.393 and sheep goat bleating at 2.124-5.204 and cat meowing at 6.595-8.543", "frequencyCaption": "duck quacking three times and sheep goat bleating one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4271.wav", "onoffCaption": "cat meowing at 2.651-3.736, 4.47-6.63, 7.481-8.502", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4294.wav", "onoffCaption": "whistling at 1.904-4.779, 6.225-8.454", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4401.wav", "onoffCaption": "cow mooing at 0.424-3.406, 5.35-7.852 and car horn honking at 0.85-5.099, 5.722-8.254", "frequencyCaption": "cow mooing two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4454.wav", "onoffCaption": "door knocking at 2.973-5.436 and spraying at 7.871-8.621", "frequencyCaption": "door knocking one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4627.wav", "onoffCaption": "cat meowing at 0.803-1.814, 2.735-4.045 and woman laughing at 3.119-6.406", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4672.wav", "onoffCaption": "car horn honking at 0.057-3.552, 4.582-7.095", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4697.wav", "onoffCaption": "gunshot at 0.129-2.129, 2.887-4.887, 5.731-8.232 and door knocking at 5.47-8.2", "frequencyCaption": "gunshot three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4769.wav", "onoffCaption": "woman laughing at 2.871-5.29 and duck quacking at 3.644-5.644, 7.393-9.393", "frequencyCaption": "woman laughing one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4921.wav", "onoffCaption": "dog barking at 0.349-2.787, 4.269-6.269", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4950.wav", "onoffCaption": "cat meowing at 0.651-2.268, 3.577-4.588, 5.723-7.267 and duck quacking at 3.073-5.073", "frequencyCaption": "cat meowing three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2030.wav", "onoffCaption": "thump thud at 3.336-7.786", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2065.wav", "onoffCaption": "sneeze at 0.087-1.375, 2.994-5.307, 6.757-8.702", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2080.wav", "onoffCaption": "dog barking at 2.86-4.86, 6.259-8.259", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2164.wav", "onoffCaption": "duck quacking at 1.029-3.029, 4.586-6.586, 7.745-9.745", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2216.wav", "onoffCaption": "car horn honking at 1.183-4.77 and cat meowing at 3.31-4.927, 6.66-8.277", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2243.wav", "onoffCaption": "sneeze at 0.224-1.327 and duck quacking at 0.322-2.322", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2358.wav", "onoffCaption": "burping belching at 2.599-5.801, 7.699-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2433.wav", "onoffCaption": "gunshot at 0.442-2.442", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2466.wav", "onoffCaption": "whistling at 1.816-6.3", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2483.wav", "onoffCaption": "dog barking at 1.83-3.83, 5.526-7.526", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2528.wav", "onoffCaption": "explosion at 1.045-3.798, 4.36-6.68", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2598.wav", "onoffCaption": "burping belching at 2.37-6.37, 7.357-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2615.wav", "onoffCaption": "duck quacking at 3.183-5.183", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2640.wav", "onoffCaption": "dog barking at 0.901-2.901, 5.102-7.102", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2879.wav", "onoffCaption": "spraying at 0.77-3.354, 5.572-6.353 and woman laughing at 3.378-6.473, 7.664-10.0 and sheep goat bleating at 4.777-6.777", "frequencyCaption": "spraying two times and woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2937.wav", "onoffCaption": "door knocking at 0.913-3.98, 5.526-7.766", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2962.wav", "onoffCaption": "burping belching at 0.503-3.503, 5.097-8.097 and door knocking at 7.1-9.502", "frequencyCaption": "burping belching two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2987.wav", "onoffCaption": "dog barking at 0.633-2.633, 4.02-6.02, 7.631-9.742", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4026.wav", "onoffCaption": "door slamming at 0.769-1.269 and dog barking at 4.994-6.994", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4073.wav", "onoffCaption": "duck quacking at 0.686-2.686, 4.676-6.676", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4096.wav", "onoffCaption": "tapping clicking clanking at 0.677-4.117", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4168.wav", "onoffCaption": "door knocking at 2.99-5.342 and tapping clicking clanking at 7.271-10.0", "frequencyCaption": "door knocking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4200.wav", "onoffCaption": "whistling at 2.405-5.38 and car horn honking at 3.25-7.65", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4255.wav", "onoffCaption": "door slamming at 2.703-3.82, 4.869-6.16, 6.968-8.509", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4425.wav", "onoffCaption": "whistling at 1.576-7.683", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4470.wav", "onoffCaption": "door slamming at 0.129-0.81, 1.333-3.559, 4.597-5.097 and door knocking at 7.313-10.0", "frequencyCaption": "door slamming three times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4495.wav", "onoffCaption": "sneeze at 3.101-5.186, 7.527-9.234", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4603.wav", "onoffCaption": "cow mooing at 1.547-4.557, 5.816-8.826 and cat meowing at 2.421-4.415", "frequencyCaption": "cow mooing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4656.wav", "onoffCaption": "burping belching at 1.681-4.681, 5.243-8.243", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4718.wav", "onoffCaption": "train horn at 2.361-4.801 and sheep goat bleating at 7.701-9.701", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4974.wav", "onoffCaption": "cow mooing at 0.276-3.574 and burping belching at 0.774-3.774 and thump thud at 1.698-4.16", "frequencyCaption": "cow mooing one times and burping belching one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4991.wav", "onoffCaption": "explosion at 3.281-8.281", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2131.wav", "onoffCaption": "spraying at 2.591-3.113 and duck quacking at 4.191-6.191 and cow mooing at 4.489-7.499", "frequencyCaption": "spraying one times and duck quacking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2181.wav", "onoffCaption": "door knocking at 2.812-7.345, 7.981-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2259.wav", "onoffCaption": "dog barking at 0.445-2.445, 3.999-5.999 and gunshot at 4.324-6.324", "frequencyCaption": "dog barking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2317.wav", "onoffCaption": "tapping clicking clanking at 0.145-3.585", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2342.wav", "onoffCaption": "burping belching at 1.788-6.124, 7.024-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2429.wav", "onoffCaption": "duck quacking at 2.986-4.986 and door slamming at 6.513-9.23", "frequencyCaption": "duck quacking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2499.wav", "onoffCaption": "sheep goat bleating at 0.024-3.32 and dog barking at 1.92-3.92", "frequencyCaption": "sheep goat bleating one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2532.wav", "onoffCaption": "explosion at 2.04-4.631, 7.094-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2567.wav", "onoffCaption": "door slamming at 0.156-2.074", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2582.wav", "onoffCaption": "car horn honking at 3.286-7.535", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2714.wav", "onoffCaption": "sneeze at 2.176-4.121, 6.135-8.08", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2741.wav", "onoffCaption": "sneeze at 0.258-1.422, 3.657-4.821 and woman laughing at 7.507-10.0", "frequencyCaption": "sneeze two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2836.wav", "onoffCaption": "thump thud at 2.554-5.054", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2863.wav", "onoffCaption": "duck quacking at 0.365-2.365 and car horn honking at 4.688-8.183 and cat meowing at 5.873-7.148", "frequencyCaption": "duck quacking one times and car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2886.wav", "onoffCaption": "burping belching at 0.982-4.542", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2978.wav", "onoffCaption": "cat meowing at 0.357-1.901, 4.148-5.692, 6.877-8.421", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4069.wav", "onoffCaption": "train horn at 0.267-3.747, 4.969-7.643", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4127.wav", "onoffCaption": "explosion at 1.4-3.407, 5.53-7.544", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4172.wav", "onoffCaption": "thump thud at 3.642-6.104", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4197.wav", "onoffCaption": "car horn honking at 0.548-2.895, 4.036-6.962 and duck quacking at 2.823-4.823, 5.992-7.992", "frequencyCaption": "car horn honking two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4301.wav", "onoffCaption": "door knocking at 0.963-3.315, 4.703-7.055", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4354.wav", "onoffCaption": "car horn honking at 2.552-5.465", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4488.wav", "onoffCaption": "gunshot at 0.894-3.064, 4.684-6.684 and spraying at 1.03-1.762, 3.015-3.747", "frequencyCaption": "gunshot two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4524.wav", "onoffCaption": "explosion at 1.076-3.083, 5.017-7.024", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4571.wav", "onoffCaption": "thump thud at 0.313-2.541 and door slamming at 4.244-5.082, 5.615-6.453, 7.056-7.894", "frequencyCaption": "thump thud one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4594.wav", "onoffCaption": "woman laughing at 3.108-5.713, 6.503-8.987 and door slamming at 5.806-7.197", "frequencyCaption": "woman laughing two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4619.wav", "onoffCaption": "door knocking at 0.244-2.814, 4.673-7.012 and gunshot at 0.491-2.491 and sneeze at 3.661-4.895", "frequencyCaption": "door knocking two times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4702.wav", "onoffCaption": "thump thud at 2.28-6.73, 7.816-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4757.wav", "onoffCaption": "dog barking at 0.149-2.149 and tapping clicking clanking at 0.464-3.904 and burping belching at 5.381-8.925", "frequencyCaption": "dog barking one times and tapping clicking clanking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4820.wav", "onoffCaption": "cow mooing at 2.125-5.135 and gunshot at 3.983-5.983, 7.818-9.818", "frequencyCaption": "cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4872.wav", "onoffCaption": "thump thud at 2.268-6.718", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4875.wav", "onoffCaption": "door slamming at 0.609-2.609, 4.435-6.435 and sheep goat bleating at 2.272-4.272, 5.291-7.291", "frequencyCaption": "door slamming two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4890.wav", "onoffCaption": "sheep goat bleating at 3.192-5.192, 5.962-7.962 and spraying at 3.225-5.353, 6.646-8.774", "frequencyCaption": "sheep goat bleating two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4969.wav", "onoffCaption": "sneeze at 0.02-1.979 and door slamming at 1.078-2.453", "frequencyCaption": "sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2037.wav", "onoffCaption": "cow mooing at 1.731-4.713, 6.368-9.35 and spraying at 2.99-4.247, 5.36-6.617", "frequencyCaption": "cow mooing two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2078.wav", "onoffCaption": "cat meowing at 0.134-3.038, 4.259-7.163 and explosion at 0.649-2.742, 3.601-5.694 and whistling at 1.104-5.588, 6.512-9.029", "frequencyCaption": "cat meowing two times and explosion two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_2136.wav", "onoffCaption": "spraying at 0.061-0.665, 2.012-2.793 and explosion at 6.775-8.839", "frequencyCaption": "spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2163.wav", "onoffCaption": "train horn at 1.284-4.084, 5.069-7.95 and gunshot at 2.44-4.44", "frequencyCaption": "train horn two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2186.wav", "onoffCaption": "train horn at 2.174-7.918 and door knocking at 2.306-4.494, 6.83-9.018", "frequencyCaption": "train horn one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2310.wav", "onoffCaption": "cow mooing at 0.644-5.073, 7.545-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2345.wav", "onoffCaption": "cow mooing at 0.329-5.309, 6.179-9.161", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2461.wav", "onoffCaption": "explosion at 1.207-3.3, 3.876-5.969 and car horn honking at 2.517-5.43, 7.292-10.0", "frequencyCaption": "explosion two times and car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2535.wav", "onoffCaption": "cow mooing at 0.06-3.07, 5.163-8.173 and thump thud at 0.481-4.148, 4.943-7.171", "frequencyCaption": "cow mooing two times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2560.wav", "onoffCaption": "thump thud at 3.064-6.111", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2585.wav", "onoffCaption": "cow mooing at 0.75-5.73, 6.405-8.768", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2608.wav", "onoffCaption": "train horn at 1.016-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2713.wav", "onoffCaption": "explosion at 1.31-4.304, 5.298-8.298", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2746.wav", "onoffCaption": "explosion at 0.892-3.764, 4.383-6.907, 7.915-10.0", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_2831.wav", "onoffCaption": "gunshot at 2.394-4.394, 6.612-9.086", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2864.wav", "onoffCaption": "train horn at 0.519-5.377, 6.619-9.02", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2881.wav", "onoffCaption": "dog barking at 0.363-2.363, 4.329-6.329, 7.313-9.313", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2980.wav", "onoffCaption": "door slamming at 2.98-4.495, 5.851-7.366", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4120.wav", "onoffCaption": "sneeze at 0.206-2.452, 4.198-5.372, 6.923-9.537", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4175.wav", "onoffCaption": "duck quacking at 2.753-4.753, 6.321-8.321", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4190.wav", "onoffCaption": "sheep goat bleating at 0.805-2.805", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4248.wav", "onoffCaption": "duck quacking at 0.163-2.163 and tapping clicking clanking at 5.673-9.113", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4306.wav", "onoffCaption": "tapping clicking clanking at 3.279-6.719 and sheep goat bleating at 7.072-9.072", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4353.wav", "onoffCaption": "duck quacking at 0.65-2.65, 4.02-6.02", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4438.wav", "onoffCaption": "thump thud at 0.905-4.572 and cat meowing at 6.995-8.566", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4523.wav", "onoffCaption": "duck quacking at 2.504-4.504 and train horn at 6.766-10.0", "frequencyCaption": "duck quacking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4576.wav", "onoffCaption": "gunshot at 0.73-2.73, 4.574-6.574 and train horn at 3.274-6.594", "frequencyCaption": "gunshot two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4593.wav", "onoffCaption": "car horn honking at 0.119-5.026 and sneeze at 0.407-1.638 and door slamming at 0.512-2.64", "frequencyCaption": "car horn honking one times and sneeze one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4705.wav", "onoffCaption": "sheep goat bleating at 0.586-4.226, 6.002-9.642", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4750.wav", "onoffCaption": "tapping clicking clanking at 1.972-5.412, 6.469-9.909 and woman laughing at 2.03-4.128", "frequencyCaption": "tapping clicking clanking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4827.wav", "onoffCaption": "dog barking at 0.577-2.577 and train horn at 0.636-4.756, 5.88-10.0", "frequencyCaption": "dog barking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4897.wav", "onoffCaption": "cow mooing at 2.385-5.367 and spraying at 8.29-8.917", "frequencyCaption": "cow mooing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2013.wav", "onoffCaption": "dog barking at 1.338-3.338, 4.879-6.879", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2062.wav", "onoffCaption": "car horn honking at 1.119-3.632, 5.271-7.784 and dog barking at 4.978-6.978, 7.545-9.545", "frequencyCaption": "car horn honking two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2087.wav", "onoffCaption": "door knocking at 0.82-5.259 and burping belching at 1.654-7.255 and door slamming at 6.373-9.347", "frequencyCaption": "door knocking one times and burping belching one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2108.wav", "onoffCaption": "cow mooing at 3.083-6.093", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2179.wav", "onoffCaption": "gunshot at 0.366-2.366 and door slamming at 4.91-5.761, 7.513-8.364", "frequencyCaption": "gunshot one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2211.wav", "onoffCaption": "sneeze at 0.014-4.514, 5.427-7.903", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2244.wav", "onoffCaption": "train horn at 1.183-4.583, 5.705-9.105", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2434.wav", "onoffCaption": "dog barking at 0.142-2.142, 3.472-5.472, 6.656-8.656", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2484.wav", "onoffCaption": "cat meowing at 0.11-3.351, 4.446-7.687", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2612.wav", "onoffCaption": "cat meowing at 4.007-5.562, 6.692-7.967", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2647.wav", "onoffCaption": "spraying at 0.301-0.876, 3.183-3.758", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2709.wav", "onoffCaption": "train horn at 1.594-5.594", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2930.wav", "onoffCaption": "tapping clicking clanking at 0.463-3.903, 4.444-6.621, 7.812-10.0", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_2965.wav", "onoffCaption": "door knocking at 0.177-3.552", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4021.wav", "onoffCaption": "dog barking at 2.558-4.558 and sheep goat bleating at 6.288-8.288", "frequencyCaption": "dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4074.wav", "onoffCaption": "gunshot at 1.364-3.364, 4.698-6.698, 7.739-9.739", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4091.wav", "onoffCaption": "train horn at 3.3-6.18, 6.958-9.838", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4207.wav", "onoffCaption": "spraying at 0.091-0.995", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4252.wav", "onoffCaption": "duck quacking at 0.773-2.773, 4.659-6.659 and cat meowing at 6.245-7.257, 8.139-9.683", "frequencyCaption": "duck quacking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4349.wav", "onoffCaption": "whistling at 0.804-2.813, 4.259-7.09 and sheep goat bleating at 1.958-6.678", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4422.wav", "onoffCaption": "door knocking at 3.405-5.785", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4477.wav", "onoffCaption": "cat meowing at 0.083-1.349, 2.041-3.625 and gunshot at 2.761-4.761", "frequencyCaption": "cat meowing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4492.wav", "onoffCaption": "gunshot at 0.382-2.382, 3.831-5.831", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4539.wav", "onoffCaption": "duck quacking at 3.101-5.101 and thump thud at 7.995-10.0", "frequencyCaption": "duck quacking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4589.wav", "onoffCaption": "burping belching at 0.511-3.511, 4.303-7.246, 7.967-10.0", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_4604.wav", "onoffCaption": "thump thud at 3.427-5.927 and tapping clicking clanking at 3.66-7.1", "frequencyCaption": "thump thud one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4651.wav", "onoffCaption": "gunshot at 1.084-3.084, 3.952-5.952, 7.198-9.198 and whistling at 1.399-9.784", "frequencyCaption": "gunshot three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4690.wav", "onoffCaption": "tapping clicking clanking at 1.909-5.349 and dog barking at 7.675-9.675", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4868.wav", "onoffCaption": "tapping clicking clanking at 0.529-3.969", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4926.wav", "onoffCaption": "gunshot at 1.876-3.876, 4.842-6.842, 7.718-9.718", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4973.wav", "onoffCaption": "woman laughing at 2.777-5.546", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4996.wav", "onoffCaption": "gunshot at 2.571-4.701, 5.405-7.535", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2046.wav", "onoffCaption": "thump thud at 0.536-3.307 and gunshot at 3.348-5.348", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2235.wav", "onoffCaption": "explosion at 2.514-5.514, 7.383-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2260.wav", "onoffCaption": "train horn at 2.617-7.694", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2285.wav", "onoffCaption": "tapping clicking clanking at 0.136-3.576 and dog barking at 0.345-2.745", "frequencyCaption": "tapping clicking clanking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2334.wav", "onoffCaption": "thump thud at 0.05-3.097, 4.756-7.218 and dog barking at 0.212-2.212", "frequencyCaption": "thump thud two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2410.wav", "onoffCaption": "gunshot at 1.449-3.449 and cat meowing at 2.183-4.143 and sneeze at 3.218-4.392", "frequencyCaption": "gunshot one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2445.wav", "onoffCaption": "cat meowing at 0.275-1.811, 3.469-5.005, 6.461-7.997 and dog barking at 2.825-4.825 and sneeze at 5.34-8.0", "frequencyCaption": "cat meowing three times and dog barking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2636.wav", "onoffCaption": "train horn at 2.96-6.36, 7.243-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2663.wav", "onoffCaption": "whistling at 0.052-2.281, 4.45-7.014 and train horn at 4.814-8.214", "frequencyCaption": "whistling two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2679.wav", "onoffCaption": "spraying at 2.8-3.3, 5.687-7.42", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2686.wav", "onoffCaption": "train horn at 2.685-7.762", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2762.wav", "onoffCaption": "door slamming at 0.1-2.32, 4.242-6.462 and dog barking at 3.512-5.512, 6.099-8.099", "frequencyCaption": "door slamming two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2778.wav", "onoffCaption": "car horn honking at 0.462-3.681 and sheep goat bleating at 5.35-7.35", "frequencyCaption": "car horn honking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2787.wav", "onoffCaption": "tapping clicking clanking at 1.579-5.019 and burping belching at 2.926-5.926, 7.082-10.0", "frequencyCaption": "tapping clicking clanking one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2914.wav", "onoffCaption": "cow mooing at 2.723-7.152", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2941.wav", "onoffCaption": "duck quacking at 0.171-2.171 and spraying at 3.39-3.994 and sneeze at 5.409-6.795", "frequencyCaption": "duck quacking one times and spraying one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4005.wav", "onoffCaption": "sneeze at 1.605-4.066, 6.518-8.437", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4050.wav", "onoffCaption": "cat meowing at 0.481-1.525, 3.829-4.873 and tapping clicking clanking at 1.071-4.511", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4104.wav", "onoffCaption": "explosion at 0.572-2.665, 5.145-7.238 and door slamming at 3.172-4.289, 5.792-6.909", "frequencyCaption": "explosion two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4223.wav", "onoffCaption": "duck quacking at 0.184-2.184, 3.018-5.018, 6.317-8.317 and sheep goat bleating at 6.077-8.077", "frequencyCaption": "duck quacking three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4276.wav", "onoffCaption": "gunshot at 3.244-5.718, 6.806-9.28", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4293.wav", "onoffCaption": "sneeze at 2.122-3.579, 5.94-7.397 and cow mooing at 3.03-6.012", "frequencyCaption": "sneeze two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4338.wav", "onoffCaption": "burping belching at 1.041-3.406 and sneeze at 2.307-4.392", "frequencyCaption": "burping belching one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4388.wav", "onoffCaption": "sneeze at 1.411-2.657 and explosion at 2.597-5.591", "frequencyCaption": "sneeze one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4406.wav", "onoffCaption": "door knocking at 0.49-3.61, 4.378-7.498 and train horn at 5.486-7.966", "frequencyCaption": "door knocking two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4453.wav", "onoffCaption": "explosion at 0.123-3.25, 5.341-8.468", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4548.wav", "onoffCaption": "whistling at 2.538-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4620.wav", "onoffCaption": "woman laughing at 1.411-4.465, 4.987-8.041 and gunshot at 1.892-3.892", "frequencyCaption": "woman laughing two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4675.wav", "onoffCaption": "whistling at 0.21-8.595 and sneeze at 0.27-2.229", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4819.wav", "onoffCaption": "cat meowing at 1.092-4.333", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4856.wav", "onoffCaption": "thump thud at 0.115-2.577, 4.284-6.746, 7.956-10.0", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4902.wav", "onoffCaption": "spraying at 0.111-0.611, 1.41-1.979, 2.877-4.052 and gunshot at 6.942-8.942", "frequencyCaption": "spraying three times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4957.wav", "onoffCaption": "whistling at 1.592-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2009.wav", "onoffCaption": "thump thud at 0.776-5.151", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2077.wav", "onoffCaption": "whistling at 1.112-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2112.wav", "onoffCaption": "explosion at 2.23-5.23 and door slamming at 8.138-9.157", "frequencyCaption": "explosion one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2147.wav", "onoffCaption": "woman laughing at 1.159-4.547", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2189.wav", "onoffCaption": "train horn at 0.649-3.323, 4.516-7.037", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2361.wav", "onoffCaption": "train horn at 0.466-2.621, 3.841-5.996, 7.71-9.865", "frequencyCaption": "train horn three times"}
+{"filepath": "data/multi_event_train/syn_2384.wav", "onoffCaption": "door slamming at 0.36-2.488, 4.172-6.3", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2421.wav", "onoffCaption": "spraying at 0.679-1.926, 3.703-4.95 and door knocking at 3.2-5.93", "frequencyCaption": "spraying two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2511.wav", "onoffCaption": "sheep goat bleating at 0.209-2.209", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2544.wav", "onoffCaption": "whistling at 2.084-4.313 and door slamming at 4.281-5.534", "frequencyCaption": "whistling one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2737.wav", "onoffCaption": "sneeze at 0.645-2.558, 3.721-5.634, 6.689-8.602", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2815.wav", "onoffCaption": "cat meowing at 2.48-4.23", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2840.wav", "onoffCaption": "burping belching at 0.204-3.484, 4.239-7.156 and whistling at 0.478-8.863", "frequencyCaption": "burping belching two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2925.wav", "onoffCaption": "tapping clicking clanking at 0.44-3.88", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4151.wav", "onoffCaption": "gunshot at 0.002-2.002, 3.068-5.068, 6.124-8.124", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4239.wav", "onoffCaption": "cow mooing at 1.84-5.138, 6.741-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4247.wav", "onoffCaption": "gunshot at 3.188-5.318, 7.34-9.34", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4289.wav", "onoffCaption": "spraying at 2.817-3.992, 5.823-6.823", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4322.wav", "onoffCaption": "cow mooing at 0.833-3.802, 5.116-8.098", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4377.wav", "onoffCaption": "whistling at 0.24-5.74 and spraying at 0.76-2.007, 2.668-3.915 and explosion at 1.028-6.028", "frequencyCaption": "whistling one times and spraying two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4392.wav", "onoffCaption": "spraying at 0.221-0.79, 1.312-1.881, 4.348-4.917", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4449.wav", "onoffCaption": "sheep goat bleating at 3.247-5.247", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4507.wav", "onoffCaption": "cat meowing at 0.37-2.364 and woman laughing at 2.598-5.184 and spraying at 4.817-7.836", "frequencyCaption": "cat meowing one times and woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4552.wav", "onoffCaption": "whistling at 1.266-6.441, 7.836-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4721.wav", "onoffCaption": "car horn honking at 1.973-6.295", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4774.wav", "onoffCaption": "woman laughing at 0.137-2.331, 4.071-6.657 and cat meowing at 1.489-2.799", "frequencyCaption": "woman laughing two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4791.wav", "onoffCaption": "thump thud at 1.876-4.338, 5.469-7.931 and door knocking at 5.537-8.0", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4803.wav", "onoffCaption": "cat meowing at 1.28-2.546 and door slamming at 3.938-4.789, 5.958-8.184", "frequencyCaption": "cat meowing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4918.wav", "onoffCaption": "car horn honking at 0.143-3.056, 5.143-8.069", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_5.wav", "onoffCaption": "train horn at 3.048-7.116 and cat meowing at 6.564-7.75", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2022.wav", "onoffCaption": "sneeze at 0.421-2.738", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2092.wav", "onoffCaption": "train horn at 0.127-4.457, 5.738-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2139.wav", "onoffCaption": "tapping clicking clanking at 3.088-6.528, 7.884-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2204.wav", "onoffCaption": "cat meowing at 3.308-5.058, 7.31-9.06", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2251.wav", "onoffCaption": "thump thud at 0.879-5.329", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2350.wav", "onoffCaption": "train horn at 0.102-4.302, 5.661-8.516", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2474.wav", "onoffCaption": "gunshot at 0.141-2.141, 4.521-6.521, 7.761-9.891", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2491.wav", "onoffCaption": "spraying at 0.317-0.921 and sheep goat bleating at 2.923-4.923 and burping belching at 3.97-6.97", "frequencyCaption": "spraying one times and sheep goat bleating one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2607.wav", "onoffCaption": "door knocking at 0.375-3.105 and duck quacking at 2.675-4.675", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2652.wav", "onoffCaption": "thump thud at 3.045-7.495", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2706.wav", "onoffCaption": "duck quacking at 1.028-3.028, 4.755-6.755, 7.606-9.606", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2749.wav", "onoffCaption": "spraying at 1.223-2.304", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2970.wav", "onoffCaption": "sneeze at 0.91-3.524", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2995.wav", "onoffCaption": "train horn at 3.448-9.917", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4034.wav", "onoffCaption": "dog barking at 1.213-4.533, 6.858-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4061.wav", "onoffCaption": "door slamming at 3.947-4.798 and dog barking at 7.487-9.487", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4084.wav", "onoffCaption": "sheep goat bleating at 0.916-2.916, 4.112-6.918, 7.548-9.548", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4160.wav", "onoffCaption": "gunshot at 1.982-4.483, 5.194-7.194", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4185.wav", "onoffCaption": "cow mooing at 1.666-4.964, 6.339-9.308", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4212.wav", "onoffCaption": "gunshot at 2.485-4.485, 5.365-7.365", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4309.wav", "onoffCaption": "car horn honking at 0.258-4.507, 5.729-9.978", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4437.wav", "onoffCaption": "cow mooing at 0.883-3.852, 4.639-7.608", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4462.wav", "onoffCaption": "spraying at 1.642-4.661 and train horn at 2.0-5.32", "frequencyCaption": "spraying one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4487.wav", "onoffCaption": "woman laughing at 0.162-3.257, 4.068-6.552 and dog barking at 0.198-2.198", "frequencyCaption": "woman laughing two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4536.wav", "onoffCaption": "woman laughing at 2.541-5.636, 7.344-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4579.wav", "onoffCaption": "burping belching at 0.802-4.242, 5.248-8.688 and gunshot at 2.58-4.58, 6.553-8.553", "frequencyCaption": "burping belching two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4611.wav", "onoffCaption": "dog barking at 2.046-4.046, 5.252-7.252", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4644.wav", "onoffCaption": "gunshot at 1.209-3.209, 5.051-7.324 and train horn at 2.986-7.106, 7.822-10.0", "frequencyCaption": "gunshot two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4828.wav", "onoffCaption": "car horn honking at 2.865-7.114 and thump thud at 4.049-6.277", "frequencyCaption": "car horn honking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4832.wav", "onoffCaption": "whistling at 1.547-9.202", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4898.wav", "onoffCaption": "spraying at 0.281-1.538, 2.407-3.664, 4.717-5.974", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4929.wav", "onoffCaption": "gunshot at 1.601-3.601", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4933.wav", "onoffCaption": "sheep goat bleating at 2.998-4.998, 6.444-8.444", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4966.wav", "onoffCaption": "cow mooing at 3.137-6.106", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4983.wav", "onoffCaption": "whistling at 1.786-4.015, 5.199-7.52", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2038.wav", "onoffCaption": "duck quacking at 0.312-2.312, 3.475-5.475, 6.684-8.684 and thump thud at 1.912-4.412", "frequencyCaption": "duck quacking three times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2088.wav", "onoffCaption": "cat meowing at 0.012-1.902 and burping belching at 5.241-8.531", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2123.wav", "onoffCaption": "tapping clicking clanking at 0.644-4.084, 5.681-8.568", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2176.wav", "onoffCaption": "cow mooing at 2.931-7.911", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2193.wav", "onoffCaption": "whistling at 0.783-9.168", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2305.wav", "onoffCaption": "sneeze at 0.788-2.747, 3.395-4.683 and gunshot at 1.228-3.228", "frequencyCaption": "sneeze two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2374.wav", "onoffCaption": "sneeze at 2.903-4.067, 4.883-6.047 and train horn at 3.325-5.965, 7.216-9.683", "frequencyCaption": "sneeze two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2520.wav", "onoffCaption": "sheep goat bleating at 0.244-4.164", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2575.wav", "onoffCaption": "door knocking at 0.167-2.519, 3.115-5.467, 7.305-9.657", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_2590.wav", "onoffCaption": "door slamming at 0.593-1.533, 3.478-4.418", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2639.wav", "onoffCaption": "train horn at 3.428-10.0", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2648.wav", "onoffCaption": "gunshot at 0.06-2.06", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2722.wav", "onoffCaption": "duck quacking at 1.492-3.492, 5.29-7.29", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2753.wav", "onoffCaption": "thump thud at 0.078-2.306, 4.188-6.416 and train horn at 0.566-3.206 and dog barking at 2.021-5.341, 6.271-9.591", "frequencyCaption": "thump thud two times and train horn one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2824.wav", "onoffCaption": "woman laughing at 0.649-3.004, 5.141-7.589 and door knocking at 3.792-6.859, 7.738-10.0", "frequencyCaption": "woman laughing two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2871.wav", "onoffCaption": "cow mooing at 1.316-4.614, 5.217-8.144", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2894.wav", "onoffCaption": "gunshot at 1.267-3.267, 4.861-6.861", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4135.wav", "onoffCaption": "duck quacking at 0.516-2.516", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4144.wav", "onoffCaption": "sneeze at 0.05-2.163, 4.515-6.628, 7.634-9.747", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4208.wav", "onoffCaption": "dog barking at 3.387-5.387", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4313.wav", "onoffCaption": "burping belching at 1.703-3.826, 4.405-7.405 and dog barking at 5.109-7.109, 7.775-9.775", "frequencyCaption": "burping belching two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4346.wav", "onoffCaption": "tapping clicking clanking at 0.824-4.264 and cow mooing at 6.055-10.0", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4478.wav", "onoffCaption": "duck quacking at 0.198-2.198, 2.933-4.933 and cow mooing at 1.378-4.347", "frequencyCaption": "duck quacking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4563.wav", "onoffCaption": "explosion at 0.394-3.394", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4586.wav", "onoffCaption": "door knocking at 3.024-6.133 and cat meowing at 3.165-4.525", "frequencyCaption": "door knocking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4710.wav", "onoffCaption": "burping belching at 1.228-4.407 and sneeze at 2.434-3.728", "frequencyCaption": "burping belching one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4745.wav", "onoffCaption": "burping belching at 0.033-3.212, 4.738-7.917 and sheep goat bleating at 3.435-5.435, 6.487-8.487", "frequencyCaption": "burping belching two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4816.wav", "onoffCaption": "woman laughing at 3.804-6.029 and dog barking at 5.915-7.915", "frequencyCaption": "woman laughing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4867.wav", "onoffCaption": "duck quacking at 0.123-2.123, 4.217-6.217 and gunshot at 4.27-6.27", "frequencyCaption": "duck quacking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4882.wav", "onoffCaption": "spraying at 0.017-0.868, 2.396-3.247", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4999.wav", "onoffCaption": "car horn honking at 0.007-3.226, 5.675-8.894 and cat meowing at 0.879-1.981", "frequencyCaption": "car horn honking two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2049.wav", "onoffCaption": "sheep goat bleating at 2.376-4.376, 5.004-7.004", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2053.wav", "onoffCaption": "cow mooing at 3.22-6.23", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2107.wav", "onoffCaption": "door knocking at 0.202-2.579 and thump thud at 7.565-10.0", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2148.wav", "onoffCaption": "spraying at 1.992-3.754 and door slamming at 2.991-4.382, 5.138-6.529, 7.82-9.211", "frequencyCaption": "spraying one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2152.wav", "onoffCaption": "tapping clicking clanking at 3.493-6.933 and cow mooing at 3.754-7.052", "frequencyCaption": "tapping clicking clanking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2321.wav", "onoffCaption": "whistling at 2.723-7.207", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2391.wav", "onoffCaption": "cat meowing at 2.812-4.772, 6.893-8.853", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2405.wav", "onoffCaption": "whistling at 0.352-3.227", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2504.wav", "onoffCaption": "sneeze at 3.377-5.336", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2551.wav", "onoffCaption": "door slamming at 2.538-3.677, 4.833-5.972", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2689.wav", "onoffCaption": "woman laughing at 2.249-8.983", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2777.wav", "onoffCaption": "burping belching at 0.147-3.691", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2792.wav", "onoffCaption": "burping belching at 1.622-5.491", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2800.wav", "onoffCaption": "door slamming at 0.311-1.544, 3.521-4.754", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2855.wav", "onoffCaption": "car horn honking at 0.434-5.341", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2901.wav", "onoffCaption": "whistling at 0.354-5.854", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4111.wav", "onoffCaption": "woman laughing at 2.429-4.534, 5.334-7.689 and dog barking at 3.371-5.371, 6.548-8.548", "frequencyCaption": "woman laughing two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4279.wav", "onoffCaption": "car horn honking at 2.345-7.252", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4286.wav", "onoffCaption": "cow mooing at 0.317-5.297", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4337.wav", "onoffCaption": "spraying at 2.863-3.73 and cat meowing at 6.571-9.601", "frequencyCaption": "spraying one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4362.wav", "onoffCaption": "dog barking at 2.986-6.306", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4387.wav", "onoffCaption": "cow mooing at 1.656-4.666, 6.806-9.081", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4409.wav", "onoffCaption": "cat meowing at 0.497-2.491, 4.841-6.835 and car horn honking at 1.246-5.568", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4512.wav", "onoffCaption": "car horn honking at 1.26-5.101, 5.957-8.883", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4547.wav", "onoffCaption": "train horn at 1.784-4.944", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4734.wav", "onoffCaption": "gunshot at 3.456-5.475, 7.633-9.633", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4761.wav", "onoffCaption": "car horn honking at 1.518-5.172, 6.256-9.91", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4784.wav", "onoffCaption": "dog barking at 2.425-4.425, 5.507-7.507 and spraying at 5.914-7.171, 8.215-9.39", "frequencyCaption": "dog barking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4843.wav", "onoffCaption": "door slamming at 1.734-3.862, 5.016-7.144 and woman laughing at 4.551-6.632, 7.972-10.0", "frequencyCaption": "door slamming two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4958.wav", "onoffCaption": "cat meowing at 0.111-1.213, 2.139-3.241 and tapping clicking clanking at 1.733-5.173 and gunshot at 4.216-6.216", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2006.wav", "onoffCaption": "explosion at 0.8-3.794, 5.316-8.31", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2220.wav", "onoffCaption": "cow mooing at 0.297-3.307, 4.166-7.148", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2227.wav", "onoffCaption": "train horn at 0.365-2.765, 3.678-6.078, 6.632-9.032 and cat meowing at 3.73-5.301", "frequencyCaption": "train horn three times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2275.wav", "onoffCaption": "dog barking at 0.194-2.194, 4.569-6.569 and tapping clicking clanking at 0.203-3.643", "frequencyCaption": "dog barking two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2290.wav", "onoffCaption": "door knocking at 0.345-4.617, 6.113-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2450.wav", "onoffCaption": "explosion at 3.699-8.293 and sheep goat bleating at 4.38-6.38, 7.404-9.404", "frequencyCaption": "explosion one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2623.wav", "onoffCaption": "sheep goat bleating at 1.467-6.187", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2671.wav", "onoffCaption": "sneeze at 0.735-2.012, 4.199-5.476", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2676.wav", "onoffCaption": "gunshot at 2.71-4.71", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2693.wav", "onoffCaption": "sheep goat bleating at 0.134-2.134, 2.98-4.98, 6.05-8.05", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2694.wav", "onoffCaption": "cat meowing at 2.385-4.333, 5.505-7.453", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2738.wav", "onoffCaption": "whistling at 0.685-5.169, 7.608-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2788.wav", "onoffCaption": "burping belching at 0.0-4.023, 5.919-8.245 and whistling at 0.417-2.426", "frequencyCaption": "burping belching two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2954.wav", "onoffCaption": "sneeze at 0.438-2.397, 3.235-5.194 and spraying at 3.26-3.835, 5.502-6.077", "frequencyCaption": "sneeze two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4010.wav", "onoffCaption": "door slamming at 1.845-2.962, 5.286-6.403, 7.731-8.848", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4017.wav", "onoffCaption": "explosion at 2.384-5.112 and woman laughing at 7.997-10.0", "frequencyCaption": "explosion one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4045.wav", "onoffCaption": "spraying at 1.773-2.273, 3.359-3.859, 5.793-6.293", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4236.wav", "onoffCaption": "door slamming at 3.015-5.241, 6.472-8.698 and spraying at 3.798-4.373, 5.108-7.544, 8.162-8.894", "frequencyCaption": "door slamming two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4263.wav", "onoffCaption": "train horn at 3.019-6.789 and sneeze at 3.37-5.483", "frequencyCaption": "train horn one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4378.wav", "onoffCaption": "tapping clicking clanking at 0.679-4.119, 5.177-7.992 and cow mooing at 3.921-6.89", "frequencyCaption": "tapping clicking clanking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4413.wav", "onoffCaption": "whistling at 0.02-7.77 and car horn honking at 3.683-6.858", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4441.wav", "onoffCaption": "car horn honking at 0.078-4.478, 5.779-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4446.wav", "onoffCaption": "thump thud at 0.432-2.932, 4.167-6.395", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4508.wav", "onoffCaption": "door knocking at 0.277-4.027, 5.925-8.174 and dog barking at 1.887-3.887", "frequencyCaption": "door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4635.wav", "onoffCaption": "sneeze at 0.402-2.715, 5.087-7.4", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4660.wav", "onoffCaption": "thump thud at 0.05-2.278, 3.47-5.565", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4685.wav", "onoffCaption": "thump thud at 1.015-4.682, 6.109-8.296", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4859.wav", "onoffCaption": "door knocking at 0.375-2.75, 4.289-6.664 and thump thud at 4.902-7.364", "frequencyCaption": "door knocking two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4917.wav", "onoffCaption": "door knocking at 0.45-3.559 and cow mooing at 1.189-4.171, 5.069-7.646", "frequencyCaption": "door knocking one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4942.wav", "onoffCaption": "duck quacking at 2.841-4.841, 6.439-8.439", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4945.wav", "onoffCaption": "sneeze at 3.167-4.421, 5.355-6.609", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2001.wav", "onoffCaption": "cat meowing at 3.737-6.767, 7.269-10.0", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2054.wav", "onoffCaption": "explosion at 3.827-8.421", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2100.wav", "onoffCaption": "spraying at 2.764-3.368, 4.537-5.141", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2124.wav", "onoffCaption": "gunshot at 3.266-5.436, 7.546-9.546", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2155.wav", "onoffCaption": "cow mooing at 1.989-4.999, 7.306-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2268.wav", "onoffCaption": "whistling at 2.877-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2272.wav", "onoffCaption": "door slamming at 2.858-3.997, 4.816-5.621, 6.487-8.405", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2297.wav", "onoffCaption": "car horn honking at 3.459-5.924, 7.102-9.567", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2326.wav", "onoffCaption": "train horn at 1.47-5.67 and gunshot at 1.569-3.569, 4.393-6.393, 7.694-9.694 and whistling at 2.957-5.832", "frequencyCaption": "train horn one times and gunshot three times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2369.wav", "onoffCaption": "woman laughing at 1.147-3.739, 4.978-7.57", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2373.wav", "onoffCaption": "cow mooing at 0.035-4.464 and cat meowing at 7.286-8.426", "frequencyCaption": "cow mooing one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2396.wav", "onoffCaption": "door slamming at 0.228-2.448, 3.938-6.158", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2402.wav", "onoffCaption": "door knocking at 2.793-5.849", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2418.wav", "onoffCaption": "dog barking at 0.291-2.291 and explosion at 0.294-5.294, 6.146-8.21", "frequencyCaption": "dog barking one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_2457.wav", "onoffCaption": "dog barking at 3.02-5.02, 6.881-8.881", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2503.wav", "onoffCaption": "sheep goat bleating at 0.386-3.706", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2519.wav", "onoffCaption": "spraying at 3.627-4.228, 5.239-5.971, 6.818-8.976", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2556.wav", "onoffCaption": "train horn at 0.173-5.917", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2572.wav", "onoffCaption": "spraying at 0.005-0.589, 1.147-1.669, 2.456-3.703 and sheep goat bleating at 5.9-8.98", "frequencyCaption": "spraying three times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2624.wav", "onoffCaption": "gunshot at 1.164-3.638", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2725.wav", "onoffCaption": "cat meowing at 0.226-1.439, 2.668-3.881 and duck quacking at 7.397-9.397", "frequencyCaption": "cat meowing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2770.wav", "onoffCaption": "train horn at 0.143-5.858, 6.746-9.611", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2795.wav", "onoffCaption": "woman laughing at 3.182-5.298, 5.997-8.113", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2807.wav", "onoffCaption": "woman laughing at 2.828-9.562 and train horn at 3.108-7.966", "frequencyCaption": "woman laughing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2848.wav", "onoffCaption": "whistling at 2.81-9.727", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2852.wav", "onoffCaption": "sneeze at 2.011-3.718 and duck quacking at 2.159-4.159", "frequencyCaption": "sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2906.wav", "onoffCaption": "cow mooing at 2.08-5.049", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2949.wav", "onoffCaption": "woman laughing at 3.906-6.261", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2953.wav", "onoffCaption": "explosion at 0.864-3.864, 5.072-8.009", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4042.wav", "onoffCaption": "dog barking at 2.567-5.488, 7.079-10.0", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4058.wav", "onoffCaption": "sneeze at 0.092-2.495, 3.276-4.863 and duck quacking at 0.416-2.416, 4.8-6.8 and gunshot at 3.904-5.904, 6.856-8.856", "frequencyCaption": "sneeze two times and duck quacking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4116.wav", "onoffCaption": "sheep goat bleating at 0.718-2.718, 4.892-6.892 and door knocking at 1.376-5.909, 6.721-8.865 and dog barking at 5.228-7.228", "frequencyCaption": "sheep goat bleating two times and door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4143.wav", "onoffCaption": "whistling at 0.576-8.961 and dog barking at 0.748-6.385 and woman laughing at 2.533-5.138", "frequencyCaption": "whistling one times and dog barking one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4159.wav", "onoffCaption": "thump thud at 0.331-4.249 and duck quacking at 2.631-4.631, 5.476-7.476", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4231.wav", "onoffCaption": "car horn honking at 1.169-3.634 and door slamming at 7.65-9.563", "frequencyCaption": "car horn honking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4264.wav", "onoffCaption": "burping belching at 1.205-4.711, 5.317-8.823 and woman laughing at 2.257-5.311", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4281.wav", "onoffCaption": "woman laughing at 0.493-2.718", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4330.wav", "onoffCaption": "gunshot at 1.206-3.206, 4.669-6.669 and sheep goat bleating at 2.87-4.87, 6.162-8.162", "frequencyCaption": "gunshot two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4365.wav", "onoffCaption": "tapping clicking clanking at 0.181-3.621, 4.89-7.477 and door slamming at 0.518-2.059, 2.809-4.35", "frequencyCaption": "tapping clicking clanking two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4380.wav", "onoffCaption": "cow mooing at 1.129-4.427 and tapping clicking clanking at 2.306-5.746, 7.032-9.339", "frequencyCaption": "cow mooing one times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4414.wav", "onoffCaption": "door slamming at 1.825-2.663, 4.903-5.741", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4515.wav", "onoffCaption": "sneeze at 0.751-2.664, 3.184-5.097", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4540.wav", "onoffCaption": "train horn at 0.83-6.545, 7.829-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4628.wav", "onoffCaption": "duck quacking at 0.235-2.235, 3.981-5.981, 7.014-9.014 and door slamming at 0.25-1.15", "frequencyCaption": "duck quacking three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4632.wav", "onoffCaption": "spraying at 0.339-1.071, 2.214-2.798, 3.436-4.5 and sneeze at 3.126-5.211, 6.084-8.169", "frequencyCaption": "spraying three times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4667.wav", "onoffCaption": "door knocking at 2.552-4.642, 6.518-8.608", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4682.wav", "onoffCaption": "door slamming at 0.297-1.776, 2.503-3.354, 4.599-5.577", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4698.wav", "onoffCaption": "cow mooing at 0.428-5.408 and whistling at 0.539-8.924", "frequencyCaption": "cow mooing one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4729.wav", "onoffCaption": "car horn honking at 1.494-4.42, 5.18-7.375", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4733.wav", "onoffCaption": "woman laughing at 1.817-4.011 and door knocking at 5.746-8.593", "frequencyCaption": "woman laughing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4766.wav", "onoffCaption": "sheep goat bleating at 0.695-2.695, 3.642-5.642, 6.254-8.254", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4783.wav", "onoffCaption": "cow mooing at 2.223-5.233, 7.477-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4799.wav", "onoffCaption": "sneeze at 0.383-4.912, 6.313-7.59", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4811.wav", "onoffCaption": "door knocking at 3.503-5.815, 7.402-9.714", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4844.wav", "onoffCaption": "car horn honking at 0.575-3.04 and whistling at 1.136-6.311, 7.842-10.0 and door knocking at 5.702-8.47", "frequencyCaption": "car horn honking one times and whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4910.wav", "onoffCaption": "dog barking at 3.332-5.332, 6.55-8.55", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2171.wav", "onoffCaption": "gunshot at 3.247-5.247", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2194.wav", "onoffCaption": "door knocking at 2.709-6.262, 7.986-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2219.wav", "onoffCaption": "cow mooing at 0.867-5.296, 6.169-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2302.wav", "onoffCaption": "burping belching at 2.824-5.824, 6.756-8.973", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2318.wav", "onoffCaption": "whistling at 0.004-2.233, 3.194-6.069, 7.167-9.818", "frequencyCaption": "whistling three times"}
+{"filepath": "data/multi_event_train/syn_2357.wav", "onoffCaption": "sneeze at 3.056-4.148, 6.17-7.262, 8.722-9.814", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2469.wav", "onoffCaption": "burping belching at 0.092-3.092, 3.702-5.932, 7.024-9.674", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2527.wav", "onoffCaption": "spraying at 1.357-3.941, 4.523-7.107", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2597.wav", "onoffCaption": "cat meowing at 3.19-5.13, 6.832-8.772", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2701.wav", "onoffCaption": "explosion at 1.956-4.685, 5.357-8.086", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2754.wav", "onoffCaption": "sneeze at 0.324-1.824, 2.589-3.843 and tapping clicking clanking at 6.289-9.729", "frequencyCaption": "sneeze two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2823.wav", "onoffCaption": "sheep goat bleating at 0.14-3.14", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2876.wav", "onoffCaption": "duck quacking at 2.428-4.428, 6.174-8.174", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2893.wav", "onoffCaption": "door slamming at 1.258-3.741", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2938.wav", "onoffCaption": "duck quacking at 0.341-2.341, 3.83-5.83", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2988.wav", "onoffCaption": "cow mooing at 1.038-4.02", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4029.wav", "onoffCaption": "tapping clicking clanking at 3.41-6.85", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4099.wav", "onoffCaption": "thump thud at 0.883-3.222, 5.031-7.37 and sneeze at 1.897-4.557, 5.208-7.167", "frequencyCaption": "thump thud two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4132.wav", "onoffCaption": "whistling at 0.88-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4167.wav", "onoffCaption": "burping belching at 0.189-7.357 and explosion at 2.581-7.581 and cat meowing at 7.678-8.689", "frequencyCaption": "burping belching one times and explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4182.wav", "onoffCaption": "gunshot at 1.168-3.168, 5.131-7.131", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4314.wav", "onoffCaption": "burping belching at 2.924-5.924, 6.913-8.948", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4341.wav", "onoffCaption": "burping belching at 1.82-4.82", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4531.wav", "onoffCaption": "whistling at 2.216-9.966", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4564.wav", "onoffCaption": "burping belching at 1.236-4.236, 5.53-8.53 and sneeze at 5.994-7.288", "frequencyCaption": "burping belching two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4581.wav", "onoffCaption": "door knocking at 0.688-4.241 and door slamming at 7.512-8.412", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4659.wav", "onoffCaption": "whistling at 0.504-8.08 and explosion at 2.227-6.227, 6.929-9.249 and spraying at 6.965-7.473, 7.993-8.501, 9.439-9.947", "frequencyCaption": "whistling one times and explosion two times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4717.wav", "onoffCaption": "sheep goat bleating at 0.537-2.537, 3.04-5.78 and duck quacking at 1.164-3.164", "frequencyCaption": "sheep goat bleating two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4742.wav", "onoffCaption": "woman laughing at 3.527-6.165 and duck quacking at 5.318-7.318", "frequencyCaption": "woman laughing one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4835.wav", "onoffCaption": "cow mooing at 2.109-5.119 and gunshot at 7.477-9.477", "frequencyCaption": "cow mooing one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4860.wav", "onoffCaption": "thump thud at 0.222-2.561, 4.496-7.267", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4885.wav", "onoffCaption": "train horn at 0.303-3.663 and sheep goat bleating at 0.483-2.483 and cat meowing at 4.115-5.262, 7.386-8.533", "frequencyCaption": "train horn one times and sheep goat bleating one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2.wav", "onoffCaption": "gunshot at 3.979-5.979, 7.863-9.863", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2070.wav", "onoffCaption": "tapping clicking clanking at 0.312-3.752, 6.233-8.612", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2095.wav", "onoffCaption": "explosion at 0.659-3.531, 5.683-8.555", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2203.wav", "onoffCaption": "train horn at 0.751-3.991, 4.924-7.607 and burping belching at 1.081-3.842", "frequencyCaption": "train horn two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2256.wav", "onoffCaption": "whistling at 0.039-7.789", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2426.wav", "onoffCaption": "burping belching at 0.083-3.952, 4.847-8.716", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2496.wav", "onoffCaption": "burping belching at 0.734-2.841, 4.296-7.291 and duck quacking at 5.046-7.046, 7.787-9.787", "frequencyCaption": "burping belching two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2568.wav", "onoffCaption": "door knocking at 1.848-4.088, 5.701-7.941", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2600.wav", "onoffCaption": "spraying at 1.873-3.606, 5.628-7.361", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2655.wav", "onoffCaption": "door slamming at 3.365-5.728, 6.77-9.133 and woman laughing at 5.218-7.784", "frequencyCaption": "door slamming two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2839.wav", "onoffCaption": "train horn at 2.721-7.623", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2889.wav", "onoffCaption": "woman laughing at 0.784-3.389, 5.389-7.994", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2922.wav", "onoffCaption": "sheep goat bleating at 2.514-4.514, 6.364-9.354", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2977.wav", "onoffCaption": "door knocking at 3.285-6.838, 7.652-9.812", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2992.wav", "onoffCaption": "sneeze at 2.421-4.66, 5.402-7.641", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4033.wav", "onoffCaption": "door knocking at 1.626-5.922, 6.87-9.119", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4066.wav", "onoffCaption": "cat meowing at 0.594-2.73, 3.248-5.384, 7.175-9.311", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4083.wav", "onoffCaption": "door slamming at 2.678-3.911 and tapping clicking clanking at 6.349-9.789", "frequencyCaption": "door slamming one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4128.wav", "onoffCaption": "car horn honking at 0.637-5.149, 6.439-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4198.wav", "onoffCaption": "door knocking at 2.713-5.025, 6.532-9.144 and burping belching at 2.951-4.986, 6.772-8.807", "frequencyCaption": "door knocking two times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4215.wav", "onoffCaption": "explosion at 0.955-3.949 and cat meowing at 2.343-3.759", "frequencyCaption": "explosion one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4240.wav", "onoffCaption": "train horn at 3.919-8.249", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4430.wav", "onoffCaption": "burping belching at 0.108-3.367, 4.557-7.03 and explosion at 3.841-8.841", "frequencyCaption": "burping belching two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4465.wav", "onoffCaption": "spraying at 1.705-2.213, 4.527-5.102", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4480.wav", "onoffCaption": "door slamming at 0.287-2.052, 4.508-7.482", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4616.wav", "onoffCaption": "woman laughing at 2.164-4.769 and spraying at 6.976-7.976", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4643.wav", "onoffCaption": "burping belching at 0.436-2.559 and explosion at 0.7-5.621 and spraying at 7.99-9.723", "frequencyCaption": "burping belching one times and explosion one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4758.wav", "onoffCaption": "explosion at 0.898-5.898 and train horn at 2.17-8.639 and duck quacking at 2.778-4.778, 5.802-7.802", "frequencyCaption": "explosion one times and train horn one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4934.wav", "onoffCaption": "duck quacking at 2.328-4.328, 5.244-7.244 and sneeze at 2.934-4.997, 5.606-7.669", "frequencyCaption": "duck quacking two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4961.wav", "onoffCaption": "train horn at 0.419-3.219, 5.239-7.706", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4984.wav", "onoffCaption": "door knocking at 0.082-2.434, 4.087-6.099 and thump thud at 1.577-5.244 and dog barking at 5.732-7.732", "frequencyCaption": "door knocking two times and thump thud one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2025.wav", "onoffCaption": "cat meowing at 0.213-2.349, 3.579-5.163", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2031.wav", "onoffCaption": "tapping clicking clanking at 0.15-3.59, 4.176-7.053, 7.619-9.914", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_2064.wav", "onoffCaption": "gunshot at 1.132-3.132", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2081.wav", "onoffCaption": "cow mooing at 0.092-3.074, 5.366-8.348", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2217.wav", "onoffCaption": "door knocking at 0.221-3.983, 4.815-8.577 and whistling at 0.271-8.282 and car horn honking at 1.033-5.94", "frequencyCaption": "door knocking two times and whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2242.wav", "onoffCaption": "burping belching at 3.447-6.727, 7.825-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2359.wav", "onoffCaption": "spraying at 0.454-2.149 and whistling at 5.572-10.0", "frequencyCaption": "spraying one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2432.wav", "onoffCaption": "whistling at 0.178-9.045 and sneeze at 2.702-5.91", "frequencyCaption": "whistling one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2467.wav", "onoffCaption": "door slamming at 2.516-4.879, 5.771-8.745", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2473.wav", "onoffCaption": "explosion at 0.536-3.536 and door knocking at 3.283-5.595, 6.181-8.493", "frequencyCaption": "explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2482.wav", "onoffCaption": "thump thud at 2.321-5.988, 6.902-9.402", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2529.wav", "onoffCaption": "cat meowing at 1.85-6.85, 7.75-10.0 and thump thud at 3.577-6.039", "frequencyCaption": "cat meowing two times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2599.wav", "onoffCaption": "sneeze at 0.377-1.611, 3.551-4.785", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2614.wav", "onoffCaption": "train horn at 0.234-2.701, 3.897-6.569", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2641.wav", "onoffCaption": "woman laughing at 0.106-3.206 and sheep goat bleating at 3.387-5.387, 6.381-8.381 and cat meowing at 5.944-6.971, 8.04-9.067", "frequencyCaption": "woman laughing one times and sheep goat bleating two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2878.wav", "onoffCaption": "sneeze at 2.708-4.621, 6.857-9.318", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2936.wav", "onoffCaption": "car horn honking at 2.339-6.588", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2963.wav", "onoffCaption": "duck quacking at 0.591-2.591, 3.535-5.535", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2986.wav", "onoffCaption": "thump thud at 0.431-2.931, 3.694-6.194, 6.919-9.854", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4027.wav", "onoffCaption": "burping belching at 2.781-5.304, 7.447-9.97", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4072.wav", "onoffCaption": "gunshot at 0.659-2.659, 4.787-6.787", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4097.wav", "onoffCaption": "duck quacking at 2.279-4.279", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4169.wav", "onoffCaption": "thump thud at 1.232-5.682 and sheep goat bleating at 1.31-3.31", "frequencyCaption": "thump thud one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4201.wav", "onoffCaption": "thump thud at 0.928-4.595 and duck quacking at 1.99-3.99, 4.976-6.976", "frequencyCaption": "thump thud one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4254.wav", "onoffCaption": "burping belching at 0.617-4.117, 5.91-9.41", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4424.wav", "onoffCaption": "explosion at 4.143-9.064", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4471.wav", "onoffCaption": "door knocking at 0.0-6.06 and gunshot at 2.217-4.258", "frequencyCaption": "door knocking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4494.wav", "onoffCaption": "spraying at 0.609-1.673 and duck quacking at 0.711-2.711, 3.85-5.85", "frequencyCaption": "spraying one times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4602.wav", "onoffCaption": "tapping clicking clanking at 2.77-6.21, 7.235-9.279 and door knocking at 3.111-7.621", "frequencyCaption": "tapping clicking clanking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4657.wav", "onoffCaption": "car horn honking at 0.411-4.66, 5.261-7.726", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4719.wav", "onoffCaption": "sneeze at 3.096-5.041", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4920.wav", "onoffCaption": "sneeze at 0.108-2.722, 4.24-5.528, 6.997-9.077", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4975.wav", "onoffCaption": "tapping clicking clanking at 2.428-5.868, 7.423-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4990.wav", "onoffCaption": "cow mooing at 0.034-3.016 and train horn at 0.166-5.881", "frequencyCaption": "cow mooing one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2130.wav", "onoffCaption": "door slamming at 0.343-2.826, 4.639-7.122 and sheep goat bleating at 0.499-2.499, 4.127-6.127", "frequencyCaption": "door slamming two times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2165.wav", "onoffCaption": "door slamming at 1.723-3.264, 4.189-5.73", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2180.wav", "onoffCaption": "gunshot at 0.152-2.152, 3.02-5.02, 5.995-8.496", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2258.wav", "onoffCaption": "door knocking at 0.871-5.573", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2316.wav", "onoffCaption": "dog barking at 0.47-3.79 and car horn honking at 1.013-5.413 and door knocking at 3.196-5.384", "frequencyCaption": "dog barking one times and car horn honking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2343.wav", "onoffCaption": "dog barking at 0.514-2.514 and train horn at 2.989-8.704 and woman laughing at 5.063-7.701", "frequencyCaption": "dog barking one times and train horn one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2382.wav", "onoffCaption": "duck quacking at 1.839-3.839", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2428.wav", "onoffCaption": "sneeze at 2.249-4.863, 5.588-6.819, 7.797-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2498.wav", "onoffCaption": "sheep goat bleating at 3.531-5.531, 6.48-8.48", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2533.wav", "onoffCaption": "sheep goat bleating at 3.267-5.267", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2566.wav", "onoffCaption": "spraying at 0.111-0.611, 2.052-2.656, 4.625-7.015 and door knocking at 3.291-5.479, 7.437-9.625", "frequencyCaption": "spraying three times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2583.wav", "onoffCaption": "woman laughing at 1.036-3.404", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2715.wav", "onoffCaption": "sheep goat bleating at 0.66-5.54, 7.267-9.267", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2740.wav", "onoffCaption": "train horn at 0.131-3.901, 6.116-9.886", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2837.wav", "onoffCaption": "whistling at 0.681-3.656, 5.848-8.823", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2862.wav", "onoffCaption": "dog barking at 0.087-2.525, 4.272-6.272", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2887.wav", "onoffCaption": "sheep goat bleating at 1.009-3.009, 4.409-6.409 and spraying at 2.342-4.778", "frequencyCaption": "sheep goat bleating two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2979.wav", "onoffCaption": "dog barking at 1.195-3.195, 4.515-6.515, 7.789-9.789 and woman laughing at 3.19-5.776 and door slamming at 3.266-5.486", "frequencyCaption": "dog barking three times and woman laughing one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4068.wav", "onoffCaption": "sneeze at 0.318-4.818, 5.939-7.265, 8.03-9.284", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4126.wav", "onoffCaption": "whistling at 2.345-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4173.wav", "onoffCaption": "sheep goat bleating at 0.475-2.475 and cat meowing at 0.576-1.661 and sneeze at 2.632-3.735", "frequencyCaption": "sheep goat bleating one times and cat meowing one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4196.wav", "onoffCaption": "gunshot at 3.737-5.737", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4300.wav", "onoffCaption": "car horn honking at 1.553-5.14", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4355.wav", "onoffCaption": "cow mooing at 0.892-4.19, 5.52-7.717", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4501.wav", "onoffCaption": "whistling at 0.143-7.798", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4525.wav", "onoffCaption": "woman laughing at 0.406-2.631, 3.797-6.022", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4570.wav", "onoffCaption": "door slamming at 0.189-1.129, 3.598-4.538, 5.299-6.239", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4595.wav", "onoffCaption": "tapping clicking clanking at 0.081-3.521", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4618.wav", "onoffCaption": "explosion at 2.497-4.504 and door knocking at 7.28-10.0", "frequencyCaption": "explosion one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4703.wav", "onoffCaption": "woman laughing at 1.525-3.725, 5.878-8.086", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4756.wav", "onoffCaption": "cow mooing at 4.047-9.027", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4821.wav", "onoffCaption": "cat meowing at 2.463-4.423 and burping belching at 7.505-10.0", "frequencyCaption": "cat meowing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4874.wav", "onoffCaption": "burping belching at 1.013-4.013, 4.734-6.857", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4891.wav", "onoffCaption": "explosion at 0.084-3.924 and sneeze at 3.631-5.314, 7.57-9.529", "frequencyCaption": "explosion one times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2040.wav", "onoffCaption": "sneeze at 0.191-2.136, 2.684-4.629", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2114.wav", "onoffCaption": "gunshot at 0.672-2.672, 3.76-5.76 and tapping clicking clanking at 7.833-10.0", "frequencyCaption": "gunshot two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2141.wav", "onoffCaption": "whistling at 0.057-8.442", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2229.wav", "onoffCaption": "explosion at 1.961-4.961, 6.699-9.699 and sheep goat bleating at 2.522-4.522", "frequencyCaption": "explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2299.wav", "onoffCaption": "door slamming at 0.233-1.748, 2.833-4.348", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2332.wav", "onoffCaption": "train horn at 1.612-4.932", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2367.wav", "onoffCaption": "burping belching at 2.641-5.006 and door slamming at 3.208-5.925 and dog barking at 4.614-6.614", "frequencyCaption": "burping belching one times and door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2416.wav", "onoffCaption": "whistling at 0.288-8.901", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2459.wav", "onoffCaption": "whistling at 1.055-6.636", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2517.wav", "onoffCaption": "dog barking at 2.923-4.923, 6.67-8.67", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2542.wav", "onoffCaption": "cow mooing at 0.253-3.263 and cat meowing at 0.397-1.707 and spraying at 1.526-2.393", "frequencyCaption": "cow mooing one times and cat meowing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2731.wav", "onoffCaption": "burping belching at 2.229-8.316", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2764.wav", "onoffCaption": "tapping clicking clanking at 1.128-4.568, 5.185-7.486", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2781.wav", "onoffCaption": "spraying at 2.741-3.345 and dog barking at 2.749-4.749", "frequencyCaption": "spraying one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2809.wav", "onoffCaption": "dog barking at 1.625-3.625, 4.457-6.457", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2813.wav", "onoffCaption": "whistling at 0.073-8.084 and duck quacking at 0.917-2.917, 4.215-6.215 and sheep goat bleating at 2.786-4.786", "frequencyCaption": "whistling one times and duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2846.wav", "onoffCaption": "thump thud at 3.629-6.129", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2908.wav", "onoffCaption": "sheep goat bleating at 0.053-2.053, 4.129-6.129", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2912.wav", "onoffCaption": "thump thud at 0.997-3.497, 4.826-7.597", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4003.wav", "onoffCaption": "burping belching at 2.477-8.564", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4019.wav", "onoffCaption": "explosion at 1.534-5.534, 6.985-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4102.wav", "onoffCaption": "door knocking at 2.936-7.232 and door slamming at 5.741-8.458", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4157.wav", "onoffCaption": "tapping clicking clanking at 1.987-5.427, 6.586-8.643 and car horn honking at 3.357-6.576", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4270.wav", "onoffCaption": "gunshot at 0.201-2.201, 4.088-6.088", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4295.wav", "onoffCaption": "dog barking at 2.052-4.052", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4324.wav", "onoffCaption": "sneeze at 1.859-4.934, 6.814-9.889", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4371.wav", "onoffCaption": "spraying at 0.137-0.869, 2.847-3.911, 5.493-8.077", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4394.wav", "onoffCaption": "cat meowing at 0.311-1.586, 2.6-3.875 and tapping clicking clanking at 2.157-5.597", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4455.wav", "onoffCaption": "duck quacking at 0.261-2.261, 3.583-5.583, 6.193-8.193", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4554.wav", "onoffCaption": "car horn honking at 2.582-6.423, 7.928-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4626.wav", "onoffCaption": "burping belching at 2.098-5.642, 6.868-9.323", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4669.wav", "onoffCaption": "thump thud at 2.425-4.887, 5.471-8.074", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4727.wav", "onoffCaption": "tapping clicking clanking at 2.643-6.083, 7.748-9.849", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4772.wav", "onoffCaption": "sneeze at 0.269-1.5, 3.185-4.416", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4797.wav", "onoffCaption": "thump thud at 1.096-5.014", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4805.wav", "onoffCaption": "cat meowing at 0.861-3.765", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4850.wav", "onoffCaption": "whistling at 3.246-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2015.wav", "onoffCaption": "tapping clicking clanking at 0.308-3.748, 4.664-8.104 and spraying at 1.951-2.732, 4.676-5.457", "frequencyCaption": "tapping clicking clanking two times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2047.wav", "onoffCaption": "tapping clicking clanking at 3.433-6.873", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2233.wav", "onoffCaption": "train horn at 3.961-6.428", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2234.wav", "onoffCaption": "duck quacking at 0.296-2.296, 2.961-4.961", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2266.wav", "onoffCaption": "car horn honking at 0.862-4.449, 5.271-8.858", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2283.wav", "onoffCaption": "explosion at 0.487-2.58, 3.857-6.725", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2328.wav", "onoffCaption": "whistling at 0.999-6.047 and sheep goat bleating at 1.176-3.176, 3.966-5.966, 6.751-8.751", "frequencyCaption": "whistling one times and sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2398.wav", "onoffCaption": "door knocking at 1.309-4.418, 6.562-9.671 and woman laughing at 1.446-4.146, 4.759-6.771 and door slamming at 2.147-3.662, 5.397-6.397", "frequencyCaption": "door knocking two times and woman laughing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2411.wav", "onoffCaption": "tapping clicking clanking at 0.115-3.555, 5.71-8.203", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2443.wav", "onoffCaption": "cat meowing at 2.254-4.004, 4.915-6.19 and train horn at 2.628-5.988, 6.935-9.02", "frequencyCaption": "cat meowing two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2558.wav", "onoffCaption": "whistling at 1.279-3.288", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2630.wav", "onoffCaption": "whistling at 0.199-9.864 and thump thud at 0.926-4.593", "frequencyCaption": "whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2662.wav", "onoffCaption": "car horn honking at 2.538-6.938", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2665.wav", "onoffCaption": "door knocking at 0.383-5.085 and door slamming at 7.078-9.561", "frequencyCaption": "door knocking one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2680.wav", "onoffCaption": "cow mooing at 0.205-5.185, 7.154-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2687.wav", "onoffCaption": "woman laughing at 2.787-6.074, 6.661-8.766", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2779.wav", "onoffCaption": "explosion at 2.806-5.067, 6.494-8.755", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2915.wav", "onoffCaption": "sheep goat bleating at 0.649-2.649, 3.203-5.203, 5.869-7.869 and dog barking at 3.465-5.465", "frequencyCaption": "sheep goat bleating three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2947.wav", "onoffCaption": "dog barking at 0.971-2.971 and thump thud at 5.698-10.0", "frequencyCaption": "dog barking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4004.wav", "onoffCaption": "tapping clicking clanking at 1.792-5.232, 7.002-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4056.wav", "onoffCaption": "duck quacking at 3.429-5.429, 6.15-8.15", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4118.wav", "onoffCaption": "sheep goat bleating at 3.47-5.47, 6.715-8.715", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4225.wav", "onoffCaption": "gunshot at 0.572-2.572 and cat meowing at 7.208-8.22", "frequencyCaption": "gunshot one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4277.wav", "onoffCaption": "thump thud at 2.987-6.034", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4292.wav", "onoffCaption": "explosion at 0.097-2.185, 4.513-6.926", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4389.wav", "onoffCaption": "burping belching at 0.173-6.117, 6.715-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4400.wav", "onoffCaption": "sheep goat bleating at 2.724-7.604 and thump thud at 3.747-6.247", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4452.wav", "onoffCaption": "cat meowing at 3.312-4.856, 6.196-8.904", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4549.wav", "onoffCaption": "sneeze at 1.038-4.113, 5.958-7.665 and door slamming at 5.131-8.092", "frequencyCaption": "sneeze two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4621.wav", "onoffCaption": "whistling at 0.065-5.24", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4673.wav", "onoffCaption": "tapping clicking clanking at 0.128-3.568", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4696.wav", "onoffCaption": "dog barking at 3.747-9.384", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4768.wav", "onoffCaption": "tapping clicking clanking at 1.367-4.807", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4904.wav", "onoffCaption": "train horn at 3.17-5.637, 7.827-9.933", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4951.wav", "onoffCaption": "sneeze at 2.27-3.434", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4956.wav", "onoffCaption": "cat meowing at 2.574-4.145, 4.744-6.315, 7.461-9.032", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2008.wav", "onoffCaption": "gunshot at 0.098-2.098 and burping belching at 0.614-3.816, 5.977-8.798", "frequencyCaption": "gunshot one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2012.wav", "onoffCaption": "thump thud at 0.055-2.394, 3.578-6.349", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2109.wav", "onoffCaption": "sneeze at 1.144-4.792, 6.589-7.835", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2113.wav", "onoffCaption": "thump thud at 0.539-3.039, 4.952-7.18", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2146.wav", "onoffCaption": "car horn honking at 0.552-3.338, 5.746-8.532", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2261.wav", "onoffCaption": "car horn honking at 0.166-3.341 and dog barking at 2.611-4.611 and sheep goat bleating at 5.761-7.761", "frequencyCaption": "car horn honking one times and dog barking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2284.wav", "onoffCaption": "whistling at 1.837-4.812, 5.973-8.948 and sheep goat bleating at 5.456-8.536", "frequencyCaption": "whistling two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2360.wav", "onoffCaption": "tapping clicking clanking at 2.241-5.681, 7.788-9.791", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2385.wav", "onoffCaption": "cat meowing at 2.894-3.921", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2444.wav", "onoffCaption": "gunshot at 0.397-2.397", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2510.wav", "onoffCaption": "gunshot at 0.453-2.693, 4.329-6.459", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2545.wav", "onoffCaption": "woman laughing at 0.559-3.145, 5.591-8.177", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2637.wav", "onoffCaption": "woman laughing at 0.151-3.438, 5.006-8.293", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2736.wav", "onoffCaption": "woman laughing at 0.176-3.752, 4.722-8.298 and explosion at 0.858-2.951 and door knocking at 4.024-6.376, 7.281-9.633", "frequencyCaption": "woman laughing two times and explosion one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2814.wav", "onoffCaption": "gunshot at 1.732-3.732, 6.222-8.222", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2841.wav", "onoffCaption": "burping belching at 0.477-3.477 and cat meowing at 2.608-4.358, 6.377-8.127", "frequencyCaption": "burping belching one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2940.wav", "onoffCaption": "door slamming at 0.759-3.476, 5.126-6.45, 7.827-9.13", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4051.wav", "onoffCaption": "dog barking at 0.092-2.092", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4150.wav", "onoffCaption": "whistling at 0.595-5.77, 7.744-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4222.wav", "onoffCaption": "spraying at 0.814-1.755, 3.619-4.369 and cow mooing at 2.19-5.2 and gunshot at 3.053-5.053, 6.329-8.329", "frequencyCaption": "spraying two times and cow mooing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4238.wav", "onoffCaption": "whistling at 0.013-7.763 and thump thud at 0.796-3.296, 4.227-6.566 and sneeze at 4.06-6.005, 7.006-8.951", "frequencyCaption": "whistling one times and thump thud two times and sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4288.wav", "onoffCaption": "door knocking at 0.772-3.152 and gunshot at 0.952-2.952, 3.707-5.707, 6.781-8.781", "frequencyCaption": "door knocking one times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4323.wav", "onoffCaption": "door knocking at 2.129-6.831", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4339.wav", "onoffCaption": "door knocking at 2.612-4.702", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4393.wav", "onoffCaption": "explosion at 0.73-3.73, 5.196-8.196", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4407.wav", "onoffCaption": "thump thud at 0.715-3.215 and spraying at 2.174-3.869, 5.959-6.534, 7.932-8.713", "frequencyCaption": "thump thud one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4506.wav", "onoffCaption": "duck quacking at 2.427-4.427, 5.446-7.446", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4674.wav", "onoffCaption": "woman laughing at 0.359-3.459 and gunshot at 0.473-2.473, 4.596-6.596 and cat meowing at 1.726-3.914", "frequencyCaption": "woman laughing one times and gunshot two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4691.wav", "onoffCaption": "dog barking at 3.46-5.86", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4775.wav", "onoffCaption": "dog barking at 1.351-3.351, 5.849-7.849", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4790.wav", "onoffCaption": "spraying at 2.35-3.434, 3.962-5.046, 5.902-6.986 and car horn honking at 3.116-7.438", "frequencyCaption": "spraying three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4802.wav", "onoffCaption": "cow mooing at 0.621-3.59, 5.64-8.609 and woman laughing at 1.406-6.445, 7.25-9.348", "frequencyCaption": "cow mooing two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4818.wav", "onoffCaption": "burping belching at 1.606-8.774 and gunshot at 5.63-7.63", "frequencyCaption": "burping belching one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4903.wav", "onoffCaption": "train horn at 3.661-8.101", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4919.wav", "onoffCaption": "door knocking at 2.36-5.207, 6.01-8.74", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2079.wav", "onoffCaption": "gunshot at 2.234-4.234", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2137.wav", "onoffCaption": "cow mooing at 2.569-5.538", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2162.wav", "onoffCaption": "thump thud at 0.332-4.707, 5.963-8.463 and car horn honking at 2.236-4.701", "frequencyCaption": "thump thud two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2335.wav", "onoffCaption": "tapping clicking clanking at 0.144-3.584, 5.766-8.185", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2344.wav", "onoffCaption": "whistling at 3.642-8.126", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2534.wav", "onoffCaption": "burping belching at 2.018-4.049, 4.623-6.654, 7.545-9.576", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2561.wav", "onoffCaption": "thump thud at 2.0-5.667, 6.541-8.88", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2584.wav", "onoffCaption": "explosion at 1.234-5.828, 6.719-10.0 and burping belching at 2.001-4.036", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2609.wav", "onoffCaption": "door knocking at 0.956-5.106", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2678.wav", "onoffCaption": "whistling at 2.467-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2712.wav", "onoffCaption": "car horn honking at 2.13-6.53, 7.444-9.957", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2763.wav", "onoffCaption": "sneeze at 2.235-3.466, 4.736-7.053, 7.596-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2786.wav", "onoffCaption": "thump thud at 0.526-3.297, 5.681-7.846 and whistling at 3.429-6.404", "frequencyCaption": "thump thud two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2830.wav", "onoffCaption": "burping belching at 1.964-4.167", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2865.wav", "onoffCaption": "sheep goat bleating at 0.352-2.352 and cow mooing at 1.577-4.559, 5.826-8.808", "frequencyCaption": "sheep goat bleating one times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2880.wav", "onoffCaption": "thump thud at 0.527-3.027 and spraying at 5.201-6.896, 9.043-9.543", "frequencyCaption": "thump thud one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_4105.wav", "onoffCaption": "thump thud at 2.466-6.916", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4174.wav", "onoffCaption": "train horn at 3.731-5.886 and cat meowing at 6.148-7.719", "frequencyCaption": "train horn one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4191.wav", "onoffCaption": "tapping clicking clanking at 1.128-4.568, 5.787-8.419", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4249.wav", "onoffCaption": "woman laughing at 3.737-6.809", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4307.wav", "onoffCaption": "woman laughing at 2.521-9.255", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4352.wav", "onoffCaption": "duck quacking at 2.31-4.31, 5.752-7.752 and door knocking at 6.221-8.684", "frequencyCaption": "duck quacking two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4376.wav", "onoffCaption": "door slamming at 3.066-4.39, 4.96-6.284, 7.039-8.363", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4439.wav", "onoffCaption": "spraying at 1.833-2.7, 3.222-4.089, 5.309-6.176", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4448.wav", "onoffCaption": "train horn at 2.19-8.25", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4522.wav", "onoffCaption": "door knocking at 0.198-2.575, 4.561-6.938 and woman laughing at 1.012-8.024", "frequencyCaption": "door knocking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4553.wav", "onoffCaption": "cat meowing at 0.582-2.576", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4720.wav", "onoffCaption": "sneeze at 0.726-4.374 and spraying at 1.105-2.352, 3.415-4.662, 5.838-7.085", "frequencyCaption": "sneeze one times and spraying three times"}
+{"filepath": "data/multi_event_train/syn_4751.wav", "onoffCaption": "cow mooing at 0.27-3.28 and burping belching at 6.449-9.449", "frequencyCaption": "cow mooing one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4826.wav", "onoffCaption": "dog barking at 0.773-2.773 and cow mooing at 1.422-6.402, 7.024-10.0 and tapping clicking clanking at 3.237-6.677", "frequencyCaption": "dog barking one times and cow mooing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4857.wav", "onoffCaption": "burping belching at 0.561-2.596, 3.115-5.15, 6.224-8.259", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2036.wav", "onoffCaption": "whistling at 0.608-3.483, 4.251-6.383", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2063.wav", "onoffCaption": "cow mooing at 0.154-3.123 and car horn honking at 5.831-10.0", "frequencyCaption": "cow mooing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2086.wav", "onoffCaption": "thump thud at 2.044-4.815, 6.182-8.692", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2178.wav", "onoffCaption": "cow mooing at 2.45-7.43", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2187.wav", "onoffCaption": "thump thud at 2.111-4.339, 5.423-8.194", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2210.wav", "onoffCaption": "dog barking at 0.214-2.214, 3.126-5.126, 6.017-8.017 and burping belching at 1.459-5.003, 6.152-8.659", "frequencyCaption": "dog barking three times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2245.wav", "onoffCaption": "gunshot at 0.323-2.323", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2311.wav", "onoffCaption": "duck quacking at 0.64-2.64, 4.525-6.525, 7.803-9.803", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2435.wav", "onoffCaption": "gunshot at 2.92-4.92, 6.165-8.165", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2613.wav", "onoffCaption": "sneeze at 3.626-4.79", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2646.wav", "onoffCaption": "sneeze at 1.137-2.431, 3.879-6.36", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2708.wav", "onoffCaption": "woman laughing at 1.941-6.98", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2747.wav", "onoffCaption": "train horn at 2.336-6.536, 7.35-9.66", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2931.wav", "onoffCaption": "woman laughing at 2.301-4.495", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4020.wav", "onoffCaption": "dog barking at 2.093-4.093, 6.532-8.532 and gunshot at 5.252-7.252 and sneeze at 5.483-7.19", "frequencyCaption": "dog barking two times and gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4075.wav", "onoffCaption": "cow mooing at 0.295-3.277", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4090.wav", "onoffCaption": "sheep goat bleating at 3.411-5.411, 6.681-8.681", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4121.wav", "onoffCaption": "thump thud at 0.085-4.46, 5.692-8.463 and door slamming at 0.091-2.091", "frequencyCaption": "thump thud two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4253.wav", "onoffCaption": "woman laughing at 3.8-7.087", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4348.wav", "onoffCaption": "sneeze at 2.276-3.564, 4.675-6.358", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4423.wav", "onoffCaption": "door slamming at 2.555-3.858 and spraying at 7.213-9.341", "frequencyCaption": "door slamming one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4476.wav", "onoffCaption": "burping belching at 0.592-4.928, 6.205-8.838", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4489.wav", "onoffCaption": "duck quacking at 1.562-3.562, 4.618-6.618, 7.146-9.146", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4493.wav", "onoffCaption": "burping belching at 1.073-4.332, 5.408-8.408 and cat meowing at 4.695-6.585, 7.258-9.148", "frequencyCaption": "burping belching two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4538.wav", "onoffCaption": "thump thud at 0.955-3.294, 4.777-7.332 and cat meowing at 1.094-2.989, 5.376-7.316", "frequencyCaption": "thump thud two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4577.wav", "onoffCaption": "train horn at 0.185-4.366 and gunshot at 7.36-9.36", "frequencyCaption": "train horn one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4588.wav", "onoffCaption": "sneeze at 0.785-3.024, 5.081-7.194, 7.912-9.975", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4592.wav", "onoffCaption": "duck quacking at 2.429-4.429, 6.607-8.607", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4605.wav", "onoffCaption": "cat meowing at 0.056-1.056, 2.31-4.677, 5.816-7.352 and door slamming at 1.937-2.437, 3.086-3.586", "frequencyCaption": "cat meowing three times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4704.wav", "onoffCaption": "spraying at 2.106-2.681, 3.676-6.112 and cat meowing at 2.174-3.44, 4.07-5.336, 6.08-7.346", "frequencyCaption": "spraying two times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4869.wav", "onoffCaption": "gunshot at 2.43-4.43 and sneeze at 6.425-10.0", "frequencyCaption": "gunshot one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4873.wav", "onoffCaption": "woman laughing at 2.935-6.02, 6.817-9.1", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4896.wav", "onoffCaption": "cow mooing at 1.844-5.142, 6.268-9.566", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4927.wav", "onoffCaption": "cow mooing at 2.408-5.418, 7.326-9.414", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4968.wav", "onoffCaption": "duck quacking at 0.788-2.788, 4.242-6.242 and cow mooing at 2.553-5.522, 7.683-9.783", "frequencyCaption": "duck quacking two times and cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4972.wav", "onoffCaption": "gunshot at 0.697-2.697, 4.279-6.279", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4997.wav", "onoffCaption": "sneeze at 0.037-1.283 and dog barking at 4.35-6.35, 7.332-9.332", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2051.wav", "onoffCaption": "burping belching at 0.475-3.296, 5.659-8.092", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2222.wav", "onoffCaption": "explosion at 3.458-5.551 and cat meowing at 5.386-6.386, 7.208-8.208", "frequencyCaption": "explosion one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2277.wav", "onoffCaption": "explosion at 1.115-3.208, 4.127-6.189", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2292.wav", "onoffCaption": "gunshot at 1.525-3.525", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2339.wav", "onoffCaption": "tapping clicking clanking at 3.039-6.479", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2389.wav", "onoffCaption": "explosion at 1.037-6.037, 6.705-9.214", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2407.wav", "onoffCaption": "duck quacking at 0.106-2.106, 3.694-5.694", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2452.wav", "onoffCaption": "gunshot at 2.289-4.289, 5.735-7.735", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2460.wav", "onoffCaption": "duck quacking at 0.55-2.55 and tapping clicking clanking at 4.483-7.923", "frequencyCaption": "duck quacking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2485.wav", "onoffCaption": "tapping clicking clanking at 2.678-6.118", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2549.wav", "onoffCaption": "thump thud at 1.047-4.094, 4.803-7.303", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2621.wav", "onoffCaption": "car horn honking at 1.926-4.839, 6.023-8.936", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2674.wav", "onoffCaption": "door knocking at 0.096-3.846, 5.923-9.673", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2691.wav", "onoffCaption": "door knocking at 0.614-4.91, 5.98-8.229 and cow mooing at 4.553-7.535", "frequencyCaption": "door knocking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2818.wav", "onoffCaption": "spraying at 2.945-3.945, 5.674-6.674, 7.659-8.659", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2903.wav", "onoffCaption": "cow mooing at 2.138-5.436, 6.232-9.201", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2964.wav", "onoffCaption": "duck quacking at 3.185-5.185, 6.724-8.724", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2981.wav", "onoffCaption": "woman laughing at 0.149-2.349, 3.471-6.171", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4012.wav", "onoffCaption": "sheep goat bleating at 2.964-6.044", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4047.wav", "onoffCaption": "whistling at 1.46-9.065", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4109.wav", "onoffCaption": "gunshot at 2.838-4.838, 6.567-8.567", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4206.wav", "onoffCaption": "car horn honking at 1.392-6.299, 7.213-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4261.wav", "onoffCaption": "gunshot at 0.303-2.809, 5.036-7.036", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4284.wav", "onoffCaption": "spraying at 0.458-1.062 and cat meowing at 3.769-5.044, 5.973-7.248, 7.752-9.027", "frequencyCaption": "spraying one times and cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4411.wav", "onoffCaption": "train horn at 0.37-2.507", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4444.wav", "onoffCaption": "duck quacking at 0.591-2.591", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4637.wav", "onoffCaption": "cow mooing at 3.246-6.228", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4650.wav", "onoffCaption": "dog barking at 3.055-5.055", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4687.wav", "onoffCaption": "car horn honking at 2.758-4.758, 7.034-9.034", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4915.wav", "onoffCaption": "gunshot at 0.044-2.044, 2.611-4.611, 5.506-7.506", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4940.wav", "onoffCaption": "explosion at 0.203-3.203, 5.192-8.192", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2004.wav", "onoffCaption": "burping belching at 2.243-4.337, 5.194-8.075", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2105.wav", "onoffCaption": "gunshot at 0.096-2.336, 3.644-5.774, 6.28-8.28 and door slamming at 0.203-0.703", "frequencyCaption": "gunshot three times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2121.wav", "onoffCaption": "door knocking at 3.089-5.713, 7.505-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2150.wav", "onoffCaption": "tapping clicking clanking at 3.558-6.998 and duck quacking at 6.009-8.009", "frequencyCaption": "tapping clicking clanking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_2174.wav", "onoffCaption": "dog barking at 0.653-2.653, 4.141-6.141", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2238.wav", "onoffCaption": "sheep goat bleating at 2.407-6.047, 6.802-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2249.wav", "onoffCaption": "spraying at 0.058-0.925", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2288.wav", "onoffCaption": "thump thud at 3.546-5.885 and spraying at 7.512-8.687", "frequencyCaption": "thump thud one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2323.wav", "onoffCaption": "door slamming at 0.137-2.137, 2.976-4.976, 6.469-8.469", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2352.wav", "onoffCaption": "whistling at 1.105-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2376.wav", "onoffCaption": "woman laughing at 2.081-5.368, 6.66-9.298 and gunshot at 3.24-5.24, 5.749-7.749", "frequencyCaption": "woman laughing two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2393.wav", "onoffCaption": "sneeze at 0.13-4.63, 5.693-10.0", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2439.wav", "onoffCaption": "thump thud at 1.519-5.186 and woman laughing at 3.888-5.97, 7.447-9.529", "frequencyCaption": "thump thud one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2448.wav", "onoffCaption": "whistling at 2.912-8.412 and tapping clicking clanking at 5.008-8.448 and gunshot at 7.107-9.107", "frequencyCaption": "whistling one times and tapping clicking clanking one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2489.wav", "onoffCaption": "tapping clicking clanking at 1.137-4.577, 6.143-8.158", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2506.wav", "onoffCaption": "tapping clicking clanking at 0.235-3.675, 5.904-9.344", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2522.wav", "onoffCaption": "thump thud at 0.022-2.793 and car horn honking at 1.65-6.557", "frequencyCaption": "thump thud one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2553.wav", "onoffCaption": "spraying at 0.053-0.628, 1.756-2.82, 3.538-4.795 and train horn at 0.586-5.026, 7.084-9.849", "frequencyCaption": "spraying three times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2577.wav", "onoffCaption": "whistling at 0.231-5.731, 6.91-8.919", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2592.wav", "onoffCaption": "explosion at 2.034-4.562, 5.841-8.369", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2704.wav", "onoffCaption": "spraying at 1.223-1.792, 2.643-3.393 and dog barking at 4.984-6.984", "frequencyCaption": "spraying two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2720.wav", "onoffCaption": "whistling at 2.643-4.872, 5.859-8.467", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_2775.wav", "onoffCaption": "sheep goat bleating at 2.154-4.154, 6.08-8.08", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2790.wav", "onoffCaption": "sneeze at 2.831-4.894, 6.753-8.816", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2802.wav", "onoffCaption": "whistling at 3.011-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2857.wav", "onoffCaption": "sneeze at 3.411-6.025, 6.998-9.45", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2873.wav", "onoffCaption": "cow mooing at 0.596-3.578, 6.07-9.052", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2896.wav", "onoffCaption": "explosion at 3.203-8.203", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2919.wav", "onoffCaption": "thump thud at 0.206-2.977, 3.705-6.476", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2956.wav", "onoffCaption": "woman laughing at 1.045-3.637, 4.725-7.363", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2968.wav", "onoffCaption": "sneeze at 0.034-1.36 and tapping clicking clanking at 5.197-8.637", "frequencyCaption": "sneeze one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4008.wav", "onoffCaption": "woman laughing at 1.326-3.52, 5.953-8.147", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4079.wav", "onoffCaption": "spraying at 0.063-1.549 and car horn honking at 3.496-6.422", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4113.wav", "onoffCaption": "car horn honking at 0.329-3.983, 4.793-7.611 and cow mooing at 5.071-8.04", "frequencyCaption": "car horn honking two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4146.wav", "onoffCaption": "sheep goat bleating at 3.618-5.618, 6.355-8.355", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4162.wav", "onoffCaption": "tapping clicking clanking at 1.893-5.333, 7.249-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4187.wav", "onoffCaption": "cow mooing at 0.198-3.496, 5.511-7.81", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4234.wav", "onoffCaption": "door knocking at 1.232-3.695 and duck quacking at 7.38-9.38", "frequencyCaption": "door knocking one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4311.wav", "onoffCaption": "gunshot at 0.178-2.178, 3.114-5.114, 5.762-8.002 and explosion at 2.45-7.45", "frequencyCaption": "gunshot three times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4335.wav", "onoffCaption": "sheep goat bleating at 0.499-2.499, 4.063-6.063, 7.651-9.651", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4360.wav", "onoffCaption": "train horn at 2.318-6.648", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4385.wav", "onoffCaption": "burping belching at 0.088-5.088, 5.724-7.818", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4510.wav", "onoffCaption": "sneeze at 0.848-2.807 and car horn honking at 7.218-10.0", "frequencyCaption": "sneeze one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4534.wav", "onoffCaption": "duck quacking at 0.48-2.48, 3.415-5.415, 7.851-9.851", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4545.wav", "onoffCaption": "train horn at 1.683-4.883, 6.583-9.433", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4662.wav", "onoffCaption": "thump thud at 2.995-5.334, 5.838-8.209 and door knocking at 4.48-8.033", "frequencyCaption": "thump thud two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4678.wav", "onoffCaption": "cow mooing at 1.341-4.351, 5.937-8.017", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4736.wav", "onoffCaption": "door knocking at 2.538-7.24 and sheep goat bleating at 5.123-7.123", "frequencyCaption": "door knocking one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4747.wav", "onoffCaption": "train horn at 2.593-4.748, 7.117-9.772", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4763.wav", "onoffCaption": "door knocking at 0.231-3.981", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4779.wav", "onoffCaption": "cow mooing at 3.146-6.156 and door knocking at 3.765-6.265, 7.273-9.33", "frequencyCaption": "cow mooing one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4786.wav", "onoffCaption": "train horn at 1.798-6.875 and dog barking at 5.284-7.284", "frequencyCaption": "train horn one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4814.wav", "onoffCaption": "woman laughing at 2.443-5.238, 7.538-10.0", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4830.wav", "onoffCaption": "spraying at 1.24-2.107, 4.396-5.263, 6.077-6.944", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4841.wav", "onoffCaption": "explosion at 0.673-5.673", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_7.wav", "onoffCaption": "sheep goat bleating at 1.687-3.687, 4.963-6.963", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2075.wav", "onoffCaption": "sneeze at 2.897-7.397", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2090.wav", "onoffCaption": "cat meowing at 3.18-4.393, 6.703-7.916", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2191.wav", "onoffCaption": "door knocking at 1.462-3.839 and thump thud at 5.248-9.623", "frequencyCaption": "door knocking one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2206.wav", "onoffCaption": "sneeze at 2.72-4.008, 4.819-6.502, 7.747-9.073", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2253.wav", "onoffCaption": "sneeze at 0.069-1.161, 1.844-4.595, 5.283-7.491", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2307.wav", "onoffCaption": "train horn at 0.621-3.061, 5.358-8.033", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2348.wav", "onoffCaption": "door slamming at 2.104-3.869, 5.812-8.786 and dog barking at 2.325-4.325", "frequencyCaption": "door slamming two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2423.wav", "onoffCaption": "gunshot at 2.171-4.672, 7.152-9.152", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2538.wav", "onoffCaption": "door knocking at 2.954-6.329", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2605.wav", "onoffCaption": "sheep goat bleating at 2.344-4.344, 6.559-8.559", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2650.wav", "onoffCaption": "burping belching at 1.323-4.829, 5.63-7.737", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2751.wav", "onoffCaption": "cow mooing at 0.385-3.354, 5.307-7.518", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2826.wav", "onoffCaption": "thump thud at 2.171-6.546, 7.936-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2927.wav", "onoffCaption": "sheep goat bleating at 2.616-4.616, 5.942-7.942", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4036.wav", "onoffCaption": "door slamming at 1.098-3.294 and explosion at 2.32-5.32, 6.136-9.136 and sneeze at 8.145-9.309", "frequencyCaption": "door slamming one times and explosion two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4086.wav", "onoffCaption": "thump thud at 0.396-2.735, 3.807-6.146, 7.197-9.536", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_4137.wav", "onoffCaption": "explosion at 0.362-3.362, 5.024-8.024 and tapping clicking clanking at 3.969-7.409", "frequencyCaption": "explosion two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4245.wav", "onoffCaption": "car horn honking at 0.888-4.729, 6.844-9.24", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4344.wav", "onoffCaption": "gunshot at 2.648-4.648, 5.271-7.312", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4460.wav", "onoffCaption": "gunshot at 2.04-4.04, 4.742-6.742", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4485.wav", "onoffCaption": "cat meowing at 0.363-2.303, 3.082-5.022", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4561.wav", "onoffCaption": "cat meowing at 0.256-5.256", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4584.wav", "onoffCaption": "thump thud at 0.001-4.451, 5.57-10.0 and sneeze at 0.165-1.665", "frequencyCaption": "thump thud two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4609.wav", "onoffCaption": "duck quacking at 0.204-2.204 and sneeze at 4.938-7.341", "frequencyCaption": "duck quacking one times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4613.wav", "onoffCaption": "cat meowing at 0.077-1.104, 1.625-2.652 and woman laughing at 5.845-8.94", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4708.wav", "onoffCaption": "burping belching at 2.626-4.751, 5.854-7.979", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4712.wav", "onoffCaption": "sneeze at 0.454-2.066, 4.387-5.915 and sheep goat bleating at 4.63-6.63", "frequencyCaption": "sneeze two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4865.wav", "onoffCaption": "thump thud at 2.49-4.718, 6.791-9.291", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4880.wav", "onoffCaption": "sheep goat bleating at 0.069-2.069 and spraying at 1.372-2.122", "frequencyCaption": "sheep goat bleating one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_4931.wav", "onoffCaption": "door slamming at 2.992-4.516, 5.7-7.224", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4964.wav", "onoffCaption": "door knocking at 0.461-4.9, 5.822-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4981.wav", "onoffCaption": "train horn at 0.592-8.792", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2020.wav", "onoffCaption": "burping belching at 0.805-4.805, 7.041-9.152", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2072.wav", "onoffCaption": "gunshot at 0.343-2.343, 4.093-6.093 and whistling at 0.77-6.27", "frequencyCaption": "gunshot two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2097.wav", "onoffCaption": "cat meowing at 0.499-2.07 and woman laughing at 4.3-6.382", "frequencyCaption": "cat meowing one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2169.wav", "onoffCaption": "thump thud at 1.556-4.327 and gunshot at 3.176-5.176", "frequencyCaption": "thump thud one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2201.wav", "onoffCaption": "door slamming at 3.56-4.365, 6.282-7.087, 8.662-9.467", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2424.wav", "onoffCaption": "cow mooing at 1.181-4.191 and explosion at 1.485-3.573 and gunshot at 2.08-4.08", "frequencyCaption": "cow mooing one times and explosion one times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2471.wav", "onoffCaption": "tapping clicking clanking at 0.354-3.794 and car horn honking at 0.545-5.452", "frequencyCaption": "tapping clicking clanking one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2476.wav", "onoffCaption": "woman laughing at 2.307-4.59", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2493.wav", "onoffCaption": "gunshot at 2.205-4.205 and thump thud at 6.122-9.789", "frequencyCaption": "gunshot one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2588.wav", "onoffCaption": "sheep goat bleating at 0.077-4.957, 5.948-10.0", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2657.wav", "onoffCaption": "door slamming at 0.527-2.747, 3.423-4.947, 6.797-8.71", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2869.wav", "onoffCaption": "burping belching at 0.032-3.472, 4.014-6.414 and explosion at 0.35-3.35, 4.571-7.435 and whistling at 3.319-8.494", "frequencyCaption": "burping belching two times and explosion two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_2920.wav", "onoffCaption": "burping belching at 1.183-5.183, 6.898-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2972.wav", "onoffCaption": "thump thud at 0.306-2.806, 3.519-6.279, 6.88-9.06", "frequencyCaption": "thump thud three times"}
+{"filepath": "data/multi_event_train/syn_2997.wav", "onoffCaption": "burping belching at 1.791-5.791, 7.164-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4031.wav", "onoffCaption": "sheep goat bleating at 0.181-2.181 and door knocking at 0.401-3.457", "frequencyCaption": "sheep goat bleating one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4063.wav", "onoffCaption": "gunshot at 0.06-2.101, 3.276-5.317, 6.463-8.504", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4178.wav", "onoffCaption": "gunshot at 1.149-3.149, 3.925-5.925, 6.522-8.522 and tapping clicking clanking at 5.417-8.857", "frequencyCaption": "gunshot three times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4210.wav", "onoffCaption": "cow mooing at 2.521-5.49", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4242.wav", "onoffCaption": "duck quacking at 2.613-4.613", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4359.wav", "onoffCaption": "door slamming at 2.052-3.169, 3.806-4.923, 5.493-6.61 and woman laughing at 3.52-6.106", "frequencyCaption": "door slamming three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4435.wav", "onoffCaption": "spraying at 4.083-4.934", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4467.wav", "onoffCaption": "sheep goat bleating at 1.74-3.74 and thump thud at 6.828-9.328", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4482.wav", "onoffCaption": "explosion at 2.5-7.5 and duck quacking at 4.221-6.221", "frequencyCaption": "explosion one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4599.wav", "onoffCaption": "sneeze at 0.066-1.16 and door knocking at 5.548-9.101", "frequencyCaption": "sneeze one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4614.wav", "onoffCaption": "duck quacking at 2.031-4.031 and train horn at 2.112-4.267, 5.729-8.564", "frequencyCaption": "duck quacking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4646.wav", "onoffCaption": "sneeze at 0.795-4.443 and sheep goat bleating at 6.423-8.423", "frequencyCaption": "sneeze one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4878.wav", "onoffCaption": "dog barking at 0.779-2.779", "frequencyCaption": "dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4963.wav", "onoffCaption": "door knocking at 1.64-4.14, 5.135-7.775", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4986.wav", "onoffCaption": "explosion at 0.519-2.612, 3.807-5.9, 7.053-9.146", "frequencyCaption": "explosion three times"}
+{"filepath": "data/multi_event_train/syn_2027.wav", "onoffCaption": "cat meowing at 2.271-5.512, 7.359-8.499", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2126.wav", "onoffCaption": "spraying at 1.5-3.628, 4.53-6.658, 7.824-9.952", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2254.wav", "onoffCaption": "spraying at 3.935-4.51, 5.563-6.344, 7.392-9.784", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2300.wav", "onoffCaption": "sneeze at 3.324-5.8, 6.929-9.405", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2355.wav", "onoffCaption": "duck quacking at 0.3-2.3, 3.841-5.841, 6.667-8.667", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_2494.wav", "onoffCaption": "car horn honking at 1.413-3.926, 4.97-7.349 and tapping clicking clanking at 2.467-5.907, 7.887-10.0", "frequencyCaption": "car horn honking two times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2570.wav", "onoffCaption": "door slamming at 0.573-1.876, 2.84-4.143", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2595.wav", "onoffCaption": "woman laughing at 0.777-3.145, 5.247-7.615 and sheep goat bleating at 1.517-3.517", "frequencyCaption": "woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2602.wav", "onoffCaption": "burping belching at 3.235-6.235, 7.539-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2618.wav", "onoffCaption": "thump thud at 1.002-3.773", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2703.wav", "onoffCaption": "cow mooing at 0.447-5.427, 7.486-9.883", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2719.wav", "onoffCaption": "burping belching at 0.966-4.246, 5.414-7.779 and cat meowing at 3.195-4.812", "frequencyCaption": "burping belching two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2756.wav", "onoffCaption": "thump thud at 1.042-3.27, 4.464-6.692", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2874.wav", "onoffCaption": "cow mooing at 2.648-5.63, 6.95-9.932", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2891.wav", "onoffCaption": "explosion at 0.143-3.143", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2975.wav", "onoffCaption": "burping belching at 2.339-5.619, 6.718-9.701", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2990.wav", "onoffCaption": "burping belching at 0.11-2.141, 2.953-4.984 and woman laughing at 7.255-9.538", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4064.wav", "onoffCaption": "whistling at 1.653-6.137, 7.449-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4081.wav", "onoffCaption": "door knocking at 0.283-4.722 and woman laughing at 1.01-3.71, 4.329-6.947", "frequencyCaption": "door knocking one times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4165.wav", "onoffCaption": "tapping clicking clanking at 1.022-4.462, 5.978-9.418 and gunshot at 1.274-3.274, 5.12-7.12", "frequencyCaption": "tapping clicking clanking two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4180.wav", "onoffCaption": "cow mooing at 1.464-6.444", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4217.wav", "onoffCaption": "dog barking at 1.46-3.46, 4.833-6.833 and cat meowing at 1.471-3.411, 5.693-7.237", "frequencyCaption": "dog barking two times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4316.wav", "onoffCaption": "car horn honking at 0.347-3.165 and cow mooing at 5.694-10.0", "frequencyCaption": "car horn honking one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4428.wav", "onoffCaption": "door knocking at 1.117-3.338, 4.015-6.392, 7.322-9.807", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4432.wav", "onoffCaption": "door knocking at 0.13-2.977 and train horn at 1.697-4.577, 5.557-8.437", "frequencyCaption": "door knocking one times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_4529.wav", "onoffCaption": "sheep goat bleating at 1.449-3.449, 4.761-6.761 and door knocking at 4.19-6.69, 7.357-9.857", "frequencyCaption": "sheep goat bleating two times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4533.wav", "onoffCaption": "door slamming at 0.536-1.555, 2.537-3.537, 5.811-6.616", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4641.wav", "onoffCaption": "explosion at 0.631-5.631 and sheep goat bleating at 2.075-5.995", "frequencyCaption": "explosion one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4740.wav", "onoffCaption": "tapping clicking clanking at 0.306-3.746, 5.611-8.049", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4837.wav", "onoffCaption": "thump thud at 1.079-5.454", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4936.wav", "onoffCaption": "spraying at 0.214-2.798, 3.672-6.256, 7.074-9.658", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2019.wav", "onoffCaption": "dog barking at 0.089-2.089, 2.72-4.72, 5.569-7.569 and train horn at 4.343-7.877 and door slamming at 5.704-7.924", "frequencyCaption": "dog barking three times and train horn one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2068.wav", "onoffCaption": "duck quacking at 0.047-2.047, 2.698-4.698, 5.474-7.474 and cat meowing at 2.237-3.377 and dog barking at 6.219-8.219", "frequencyCaption": "duck quacking three times and cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2102.wav", "onoffCaption": "spraying at 1.298-3.06, 3.89-5.065, 6.352-8.788", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2173.wav", "onoffCaption": "car horn honking at 0.126-3.78 and cat meowing at 0.557-4.917, 6.481-7.628", "frequencyCaption": "car horn honking one times and cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2196.wav", "onoffCaption": "sneeze at 1.355-4.43, 5.374-8.449", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2371.wav", "onoffCaption": "cat meowing at 1.912-3.178, 4.339-5.605, 6.882-8.148", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2394.wav", "onoffCaption": "tapping clicking clanking at 2.112-5.552, 6.278-9.718", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2525.wav", "onoffCaption": "train horn at 0.234-3.768, 4.385-7.919 and sneeze at 5.081-8.729", "frequencyCaption": "train horn two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_2554.wav", "onoffCaption": "thump thud at 2.155-4.655", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2669.wav", "onoffCaption": "door knocking at 0.303-3.856 and train horn at 5.632-9.7", "frequencyCaption": "door knocking one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2727.wav", "onoffCaption": "train horn at 0.066-4.066, 4.98-8.98 and cat meowing at 1.727-2.727", "frequencyCaption": "train horn two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2772.wav", "onoffCaption": "burping belching at 1.808-4.808, 5.553-8.183 and car horn honking at 3.667-8.067", "frequencyCaption": "burping belching two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2821.wav", "onoffCaption": "door knocking at 0.226-3.073, 4.879-7.422", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2850.wav", "onoffCaption": "dog barking at 0.215-2.215, 3.162-5.162, 6.429-8.429", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4130.wav", "onoffCaption": "dog barking at 0.005-2.005 and explosion at 3.331-6.387, 7.394-10.0", "frequencyCaption": "dog barking one times and explosion two times"}
+{"filepath": "data/multi_event_train/syn_4141.wav", "onoffCaption": "burping belching at 1.866-6.202, 7.316-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4229.wav", "onoffCaption": "cow mooing at 2.388-5.398, 7.519-9.548", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4258.wav", "onoffCaption": "explosion at 2.276-5.276, 6.95-9.95", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4332.wav", "onoffCaption": "cat meowing at 1.898-3.482, 5.203-6.935", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_4343.wav", "onoffCaption": "sheep goat bleating at 0.153-2.153, 2.766-4.766", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4498.wav", "onoffCaption": "door slamming at 1.793-2.966, 4.605-6.129, 8.222-9.162", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4517.wav", "onoffCaption": "cat meowing at 0.438-1.578, 3.851-6.376 and door slamming at 2.388-4.388, 4.942-6.942", "frequencyCaption": "cat meowing two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4566.wav", "onoffCaption": "whistling at 2.552-7.727 and car horn honking at 3.898-8.41", "frequencyCaption": "whistling one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4583.wav", "onoffCaption": "spraying at 3.279-5.671, 6.189-7.056, 9.134-9.866", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4715.wav", "onoffCaption": "duck quacking at 1.168-3.168, 3.913-5.913, 7.163-9.163 and dog barking at 1.171-3.171", "frequencyCaption": "duck quacking three times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_4764.wav", "onoffCaption": "woman laughing at 1.085-3.368 and door slamming at 5.864-6.702, 7.848-8.987", "frequencyCaption": "woman laughing one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4781.wav", "onoffCaption": "woman laughing at 0.104-2.709, 3.749-5.831, 6.999-9.199", "frequencyCaption": "woman laughing three times"}
+{"filepath": "data/multi_event_train/syn_4813.wav", "onoffCaption": "burping belching at 0.089-3.089, 4.311-7.192", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4862.wav", "onoffCaption": "cow mooing at 1.786-4.768, 5.558-8.527", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4887.wav", "onoffCaption": "woman laughing at 0.005-3.292", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4908.wav", "onoffCaption": "gunshot at 0.567-2.608, 3.838-5.879, 6.969-9.01", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4979.wav", "onoffCaption": "cow mooing at 0.323-4.752, 5.985-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2003.wav", "onoffCaption": "door slamming at 2.017-4.475 and dog barking at 6.941-8.941", "frequencyCaption": "door slamming one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2056.wav", "onoffCaption": "whistling at 2.868-4.877 and burping belching at 2.97-6.149, 7.093-10.0", "frequencyCaption": "whistling one times and burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2118.wav", "onoffCaption": "door slamming at 1.955-3.104, 4.415-5.564, 6.925-8.074 and car horn honking at 4.279-8.679", "frequencyCaption": "door slamming three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2157.wav", "onoffCaption": "door slamming at 3.384-4.617, 5.78-7.013, 7.976-9.209", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2225.wav", "onoffCaption": "duck quacking at 2.673-4.673, 6.37-8.37 and spraying at 3.012-5.596", "frequencyCaption": "duck quacking two times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2324.wav", "onoffCaption": "cow mooing at 2.87-6.168, 7.884-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2400.wav", "onoffCaption": "cat meowing at 0.016-1.229", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2455.wav", "onoffCaption": "cat meowing at 1.766-2.778, 3.833-4.973, 5.712-7.444", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2501.wav", "onoffCaption": "gunshot at 0.124-2.124, 2.96-4.96 and door slamming at 0.289-1.438", "frequencyCaption": "gunshot two times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2673.wav", "onoffCaption": "woman laughing at 0.241-2.466", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2696.wav", "onoffCaption": "cow mooing at 2.987-6.285", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2768.wav", "onoffCaption": "sneeze at 0.124-4.653, 5.641-6.935", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2797.wav", "onoffCaption": "sheep goat bleating at 0.044-2.044, 3.147-5.147, 6.598-8.598", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_2805.wav", "onoffCaption": "sneeze at 0.029-1.123, 3.204-5.521", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2904.wav", "onoffCaption": "whistling at 0.037-9.702 and dog barking at 0.082-2.082 and burping belching at 6.199-9.199", "frequencyCaption": "whistling one times and dog barking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2951.wav", "onoffCaption": "whistling at 2.043-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4015.wav", "onoffCaption": "car horn honking at 2.146-6.468, 7.709-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4114.wav", "onoffCaption": "sneeze at 0.7-3.103, 5.122-7.525", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_4233.wav", "onoffCaption": "whistling at 1.754-4.729 and woman laughing at 2.721-5.076", "frequencyCaption": "whistling one times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4266.wav", "onoffCaption": "burping belching at 1.694-5.563, 6.453-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4283.wav", "onoffCaption": "cow mooing at 0.564-3.862", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4299.wav", "onoffCaption": "car horn honking at 0.783-5.69, 7.139-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4367.wav", "onoffCaption": "train horn at 0.133-3.373, 5.09-7.557 and dog barking at 2.429-4.429, 6.478-8.878", "frequencyCaption": "train horn two times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4382.wav", "onoffCaption": "gunshot at 0.052-2.553, 3.667-5.667, 6.354-8.354", "frequencyCaption": "gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4398.wav", "onoffCaption": "cat meowing at 0.031-1.18 and gunshot at 0.572-2.742, 5.068-7.238", "frequencyCaption": "cat meowing one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4443.wav", "onoffCaption": "whistling at 0.006-7.756", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4459.wav", "onoffCaption": "cat meowing at 0.808-2.363, 3.023-4.035 and tapping clicking clanking at 6.989-10.0", "frequencyCaption": "cat meowing two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4542.wav", "onoffCaption": "explosion at 0.903-4.03, 6.338-9.465", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4558.wav", "onoffCaption": "whistling at 1.603-9.258", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4630.wav", "onoffCaption": "cow mooing at 3.094-6.076, 7.671-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4665.wav", "onoffCaption": "burping belching at 4.194-7.453", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4680.wav", "onoffCaption": "cow mooing at 0.292-3.302, 4.366-7.335 and duck quacking at 6.865-8.865", "frequencyCaption": "cow mooing two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4731.wav", "onoffCaption": "burping belching at 0.174-3.174, 4.471-7.471 and woman laughing at 1.095-3.733", "frequencyCaption": "burping belching two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4846.wav", "onoffCaption": "sheep goat bleating at 3.216-5.216, 7.445-9.445", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4947.wav", "onoffCaption": "cow mooing at 0.181-3.15, 4.35-7.332 and gunshot at 2.445-4.946, 5.779-7.779 and whistling at 4.017-6.892, 7.993-10.0", "frequencyCaption": "cow mooing two times and gunshot two times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_2042.wav", "onoffCaption": "sneeze at 2.179-4.175 and thump thud at 2.763-5.81, 7.173-9.414", "frequencyCaption": "sneeze one times and thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2159.wav", "onoffCaption": "door knocking at 2.29-5.137, 6.138-8.906", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2231.wav", "onoffCaption": "cow mooing at 3.414-6.396, 7.031-9.087", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2264.wav", "onoffCaption": "car horn honking at 3.121-5.468", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2270.wav", "onoffCaption": "train horn at 1.497-5.678, 6.589-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2281.wav", "onoffCaption": "spraying at 0.544-2.306, 3.813-5.575 and train horn at 2.976-6.976", "frequencyCaption": "spraying two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_2295.wav", "onoffCaption": "spraying at 3.423-4.05, 5.308-5.912, 7.048-7.632", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_2414.wav", "onoffCaption": "sheep goat bleating at 1.213-3.213, 5.402-7.402", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2626.wav", "onoffCaption": "cow mooing at 0.604-3.902, 5.015-7.984", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2632.wav", "onoffCaption": "duck quacking at 2.26-4.26, 6.028-8.028", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2667.wav", "onoffCaption": "sheep goat bleating at 0.259-2.259, 3.658-5.658 and cat meowing at 2.779-4.511", "frequencyCaption": "sheep goat bleating two times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2682.wav", "onoffCaption": "cat meowing at 3.69-5.106, 7.025-8.965", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2729.wav", "onoffCaption": "woman laughing at 0.702-3.774, 4.528-7.6 and burping belching at 3.235-5.329", "frequencyCaption": "woman laughing two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2799.wav", "onoffCaption": "sheep goat bleating at 0.527-2.527", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2910.wav", "onoffCaption": "sneeze at 1.626-3.309, 3.833-5.516, 6.281-7.964", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2945.wav", "onoffCaption": "thump thud at 0.075-2.846, 3.421-6.192 and burping belching at 7.915-10.0", "frequencyCaption": "thump thud two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4001.wav", "onoffCaption": "train horn at 0.358-6.073, 7.942-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4040.wav", "onoffCaption": "door slamming at 1.741-4.522", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4054.wav", "onoffCaption": "explosion at 0.122-5.122", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4227.wav", "onoffCaption": "tapping clicking clanking at 0.449-3.889, 4.56-6.915, 7.848-9.866", "frequencyCaption": "tapping clicking clanking three times"}
+{"filepath": "data/multi_event_train/syn_4272.wav", "onoffCaption": "sheep goat bleating at 0.103-2.103 and thump thud at 5.105-9.48", "frequencyCaption": "sheep goat bleating one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4297.wav", "onoffCaption": "door knocking at 1.869-3.891, 4.557-6.579, 7.312-9.334", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4328.wav", "onoffCaption": "door slamming at 0.071-2.5, 3.7-5.618, 6.407-8.32", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_4369.wav", "onoffCaption": "thump thud at 4.339-6.839", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4402.wav", "onoffCaption": "train horn at 3.277-6.597", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4416.wav", "onoffCaption": "sneeze at 0.054-1.641, 2.766-4.353, 5.195-6.782", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4457.wav", "onoffCaption": "tapping clicking clanking at 1.373-4.813, 5.85-8.035", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4519.wav", "onoffCaption": "burping belching at 2.319-8.999 and door slamming at 3.957-5.957, 7.089-10.0", "frequencyCaption": "burping belching one times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4624.wav", "onoffCaption": "train horn at 0.444-4.884, 5.959-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4671.wav", "onoffCaption": "door knocking at 3.288-6.125", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4694.wav", "onoffCaption": "thump thud at 2.224-4.724, 6.344-8.844", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4809.wav", "onoffCaption": "door slamming at 0.796-3.513, 5.446-6.424", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4848.wav", "onoffCaption": "spraying at 1.562-2.643, 3.178-4.259, 5.809-6.89", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4906.wav", "onoffCaption": "thump thud at 1.126-3.588 and cow mooing at 2.735-6.033", "frequencyCaption": "thump thud one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4912.wav", "onoffCaption": "thump thud at 1.033-5.408 and explosion at 6.642-9.514", "frequencyCaption": "thump thud one times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_4953.wav", "onoffCaption": "sneeze at 2.441-3.767 and dog barking at 2.683-4.683, 6.445-8.445", "frequencyCaption": "sneeze one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2017.wav", "onoffCaption": "duck quacking at 2.591-4.591, 6.473-8.473", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2029.wav", "onoffCaption": "thump thud at 0.12-4.57 and dog barking at 0.898-2.898, 4.555-6.555", "frequencyCaption": "thump thud one times and dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2058.wav", "onoffCaption": "door slamming at 0.54-1.713, 2.823-3.996, 5.038-6.211", "frequencyCaption": "door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2099.wav", "onoffCaption": "spraying at 1.117-1.625", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2116.wav", "onoffCaption": "sneeze at 3.4-4.634, 5.74-6.974", "frequencyCaption": "sneeze two times"}
+{"filepath": "data/multi_event_train/syn_2132.wav", "onoffCaption": "tapping clicking clanking at 2.156-5.596", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2143.wav", "onoffCaption": "door slamming at 0.016-1.319, 2.87-4.173, 5.945-7.248 and spraying at 4.635-5.135, 7.255-7.882", "frequencyCaption": "door slamming three times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2167.wav", "onoffCaption": "tapping clicking clanking at 0.555-3.995, 4.982-7.959 and car horn honking at 2.45-4.915 and dog barking at 3.612-5.612", "frequencyCaption": "tapping clicking clanking two times and car horn honking one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2182.wav", "onoffCaption": "tapping clicking clanking at 2.81-6.25", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2330.wav", "onoffCaption": "explosion at 0.7-3.572 and thump thud at 5.734-9.652", "frequencyCaption": "explosion one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2341.wav", "onoffCaption": "duck quacking at 0.59-2.59 and burping belching at 4.988-7.611", "frequencyCaption": "duck quacking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2365.wav", "onoffCaption": "tapping clicking clanking at 1.395-4.835 and door knocking at 3.354-5.731", "frequencyCaption": "tapping clicking clanking one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2380.wav", "onoffCaption": "explosion at 3.921-6.915", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_2441.wav", "onoffCaption": "sneeze at 0.431-1.523, 2.574-3.666 and gunshot at 0.854-2.854, 4.67-6.67", "frequencyCaption": "sneeze two times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2515.wav", "onoffCaption": "dog barking at 3.882-5.882, 6.574-8.574", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2540.wav", "onoffCaption": "sheep goat bleating at 0.606-2.606, 5.052-7.052 and door slamming at 5.485-6.624, 8.528-9.781", "frequencyCaption": "sheep goat bleating two times and door slamming two times"}
+{"filepath": "data/multi_event_train/syn_2564.wav", "onoffCaption": "burping belching at 1.28-3.387, 4.428-6.535, 7.767-9.874", "frequencyCaption": "burping belching three times"}
+{"filepath": "data/multi_event_train/syn_2581.wav", "onoffCaption": "cat meowing at 0.248-1.388, 3.452-4.727, 6.932-8.926", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2628.wav", "onoffCaption": "thump thud at 0.371-4.038", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2698.wav", "onoffCaption": "explosion at 1.193-3.911, 5.087-7.427", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2717.wav", "onoffCaption": "door slamming at 0.151-2.151", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2733.wav", "onoffCaption": "sheep goat bleating at 2.91-4.91", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2766.wav", "onoffCaption": "woman laughing at 1.206-3.288, 3.961-6.868 and car horn honking at 2.938-6.113", "frequencyCaption": "woman laughing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2783.wav", "onoffCaption": "door knocking at 2.496-4.736, 6.515-8.755", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2811.wav", "onoffCaption": "whistling at 3.028-8.203", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2835.wav", "onoffCaption": "whistling at 0.039-9.704 and spraying at 0.105-1.362, 3.426-4.33", "frequencyCaption": "whistling one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2844.wav", "onoffCaption": "thump thud at 3.178-6.225, 7.297-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2860.wav", "onoffCaption": "whistling at 0.213-3.088, 3.926-6.389 and door knocking at 3.404-7.554", "frequencyCaption": "whistling two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2885.wav", "onoffCaption": "duck quacking at 0.313-2.313, 4.547-6.547", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4100.wav", "onoffCaption": "tapping clicking clanking at 3.081-6.521", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4155.wav", "onoffCaption": "door slamming at 0.152-1.052, 1.604-2.584, 3.551-4.668 and woman laughing at 7.554-10.0", "frequencyCaption": "door slamming three times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4171.wav", "onoffCaption": "car horn honking at 0.875-3.661", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4194.wav", "onoffCaption": "duck quacking at 0.045-2.045, 4.189-6.189 and woman laughing at 0.11-7.122", "frequencyCaption": "duck quacking two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4219.wav", "onoffCaption": "whistling at 3.867-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4268.wav", "onoffCaption": "train horn at 0.779-3.246, 4.852-6.993, 7.924-10.0 and car horn honking at 4.157-7.376", "frequencyCaption": "train horn three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4302.wav", "onoffCaption": "whistling at 1.87-9.525", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4326.wav", "onoffCaption": "sheep goat bleating at 0.284-4.204, 5.202-9.122 and tapping clicking clanking at 2.879-6.319", "frequencyCaption": "sheep goat bleating two times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4357.wav", "onoffCaption": "car horn honking at 0.415-2.762 and spraying at 1.964-2.591, 3.444-3.944, 5.15-6.407 and duck quacking at 2.399-4.399", "frequencyCaption": "car horn honking one times and spraying three times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4373.wav", "onoffCaption": "woman laughing at 2.689-9.423", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4396.wav", "onoffCaption": "woman laughing at 2.564-5.202, 6.278-8.561", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4418.wav", "onoffCaption": "sneeze at 0.199-1.363, 2.216-3.38, 3.88-5.044", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4503.wav", "onoffCaption": "cat meowing at 2.088-3.115 and whistling at 2.403-5.278, 6.471-9.346", "frequencyCaption": "cat meowing one times and whistling two times"}
+{"filepath": "data/multi_event_train/syn_4527.wav", "onoffCaption": "whistling at 1.299-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4556.wav", "onoffCaption": "sneeze at 0.693-2.193, 3.282-5.395, 6.632-9.449", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4701.wav", "onoffCaption": "cat meowing at 1.878-3.153", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4725.wav", "onoffCaption": "door slamming at 0.723-2.026 and whistling at 1.016-8.671 and thump thud at 4.221-8.671", "frequencyCaption": "door slamming one times and whistling one times and thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4754.wav", "onoffCaption": "car horn honking at 0.071-2.536, 4.312-6.777", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4770.wav", "onoffCaption": "cow mooing at 0.271-5.251, 6.816-9.785", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4795.wav", "onoffCaption": "sheep goat bleating at 1.536-3.536 and train horn at 7.424-10.0", "frequencyCaption": "sheep goat bleating one times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4807.wav", "onoffCaption": "whistling at 0.378-7.496 and duck quacking at 0.978-2.978", "frequencyCaption": "whistling one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4823.wav", "onoffCaption": "door knocking at 0.464-2.776, 3.809-6.112, 6.749-8.876", "frequencyCaption": "door knocking three times"}
+{"filepath": "data/multi_event_train/syn_4852.wav", "onoffCaption": "cat meowing at 0.345-2.235 and door knocking at 5.757-7.978", "frequencyCaption": "cat meowing one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4938.wav", "onoffCaption": "woman laughing at 3.72-6.774", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4949.wav", "onoffCaption": "cat meowing at 2.909-3.994 and dog barking at 6.602-8.602", "frequencyCaption": "cat meowing one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2033.wav", "onoffCaption": "cow mooing at 3.185-7.614", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2061.wav", "onoffCaption": "door slamming at 1.789-4.668 and door knocking at 2.337-4.586, 6.731-9.083 and dog barking at 3.416-5.416", "frequencyCaption": "door slamming one times and door knocking two times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2066.wav", "onoffCaption": "woman laughing at 0.892-3.247 and spraying at 8.202-9.283", "frequencyCaption": "woman laughing one times and spraying one times"}
+{"filepath": "data/multi_event_train/syn_2083.wav", "onoffCaption": "spraying at 0.188-1.092", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_2084.wav", "onoffCaption": "dog barking at 0.685-2.685 and tapping clicking clanking at 6.937-10.0", "frequencyCaption": "dog barking one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2128.wav", "onoffCaption": "tapping clicking clanking at 1.229-4.669 and spraying at 6.867-7.471, 8.495-9.099", "frequencyCaption": "tapping clicking clanking one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2198.wav", "onoffCaption": "dog barking at 0.7-2.7, 4.34-7.261 and gunshot at 2.054-4.073, 4.78-6.78, 7.354-9.354", "frequencyCaption": "dog barking two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_2215.wav", "onoffCaption": "whistling at 3.911-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2240.wav", "onoffCaption": "cat meowing at 0.114-3.144, 4.645-6.585 and sheep goat bleating at 0.204-2.204", "frequencyCaption": "cat meowing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2309.wav", "onoffCaption": "cow mooing at 1.54-5.969, 6.587-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2314.wav", "onoffCaption": "whistling at 0.043-2.052, 3.186-5.195, 6.106-8.115 and tapping clicking clanking at 1.55-4.99, 5.612-9.052", "frequencyCaption": "whistling three times and tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2430.wav", "onoffCaption": "spraying at 3.021-3.872, 4.802-5.706 and cow mooing at 7.834-10.0", "frequencyCaption": "spraying two times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2437.wav", "onoffCaption": "gunshot at 2.885-4.885 and door slamming at 7.153-8.093", "frequencyCaption": "gunshot one times and door slamming one times"}
+{"filepath": "data/multi_event_train/syn_2465.wav", "onoffCaption": "cow mooing at 3.343-6.353", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2480.wav", "onoffCaption": "spraying at 1.693-2.544, 3.905-4.756", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2531.wav", "onoffCaption": "thump thud at 0.081-4.531, 6.844-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2616.wav", "onoffCaption": "whistling at 0.165-8.792 and door slamming at 0.317-1.168, 2.4-3.251, 4.02-4.871", "frequencyCaption": "whistling one times and door slamming three times"}
+{"filepath": "data/multi_event_train/syn_2643.wav", "onoffCaption": "gunshot at 0.789-2.789, 3.662-5.662 and train horn at 2.218-6.399, 7.69-10.0", "frequencyCaption": "gunshot two times and train horn two times"}
+{"filepath": "data/multi_event_train/syn_2644.wav", "onoffCaption": "cat meowing at 0.168-1.195, 1.774-3.751, 4.349-6.416", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_2659.wav", "onoffCaption": "whistling at 0.1-5.275", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2742.wav", "onoffCaption": "train horn at 2.253-5.653, 7.221-10.0", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2758.wav", "onoffCaption": "car horn honking at 0.059-3.234, 5.075-8.25", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2828.wav", "onoffCaption": "whistling at 0.57-8.32", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2933.wav", "onoffCaption": "explosion at 0.276-3.148, 3.917-6.789", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_2934.wav", "onoffCaption": "burping belching at 0.338-7.315", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2961.wav", "onoffCaption": "spraying at 4.013-6.473, 8.516-9.266", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_2984.wav", "onoffCaption": "sheep goat bleating at 0.618-2.618 and gunshot at 5.151-7.151, 7.731-9.731", "frequencyCaption": "sheep goat bleating one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4025.wav", "onoffCaption": "train horn at 3.629-6.096", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_4070.wav", "onoffCaption": "door slamming at 0.029-1.32 and cow mooing at 2.727-6.025, 7.22-9.384 and sneeze at 4.803-7.279", "frequencyCaption": "door slamming one times and cow mooing two times and sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4095.wav", "onoffCaption": "thump thud at 3.766-6.537", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4124.wav", "onoffCaption": "duck quacking at 0.025-2.025", "frequencyCaption": "duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4203.wav", "onoffCaption": "spraying at 0.356-2.816, 4.552-5.202", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4251.wav", "onoffCaption": "woman laughing at 0.47-3.062, 5.365-7.586", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4256.wav", "onoffCaption": "sheep goat bleating at 0.135-2.135, 3.296-5.296, 6.567-8.567", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4318.wav", "onoffCaption": "gunshot at 3.162-5.162, 5.706-7.706 and duck quacking at 3.962-5.962, 7.3-9.3", "frequencyCaption": "gunshot two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4426.wav", "onoffCaption": "door knocking at 2.595-4.783, 5.891-8.728", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4469.wav", "onoffCaption": "gunshot at 0.545-2.545 and woman laughing at 0.563-3.358, 4.196-6.991 and sheep goat bleating at 1.829-3.829", "frequencyCaption": "gunshot one times and woman laughing two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4473.wav", "onoffCaption": "sneeze at 3.586-8.086", "frequencyCaption": "sneeze one times"}
+{"filepath": "data/multi_event_train/syn_4496.wav", "onoffCaption": "explosion at 2.265-5.265, 6.072-8.944", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4568.wav", "onoffCaption": "cat meowing at 0.568-1.653, 2.506-3.591 and car horn honking at 7.41-10.0", "frequencyCaption": "cat meowing two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4572.wav", "onoffCaption": "duck quacking at 0.41-2.41, 4.509-6.509 and gunshot at 1.916-3.916", "frequencyCaption": "duck quacking two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4597.wav", "onoffCaption": "cow mooing at 1.045-4.027, 4.606-7.588", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4600.wav", "onoffCaption": "explosion at 0.464-4.464, 6.669-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4607.wav", "onoffCaption": "cat meowing at 2.122-3.739, 4.759-6.069, 7.198-9.192", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4655.wav", "onoffCaption": "train horn at 0.649-3.889, 6.249-9.489", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_4839.wav", "onoffCaption": "door knocking at 3.565-6.94, 7.981-10.0", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_4876.wav", "onoffCaption": "sneeze at 0.389-3.499 and whistling at 0.703-5.878", "frequencyCaption": "sneeze one times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4889.wav", "onoffCaption": "spraying at 2.837-3.412 and car horn honking at 6.908-10.0", "frequencyCaption": "spraying one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4893.wav", "onoffCaption": "door slamming at 3.845-4.864, 5.397-6.416 and duck quacking at 4.624-6.624", "frequencyCaption": "door slamming two times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4922.wav", "onoffCaption": "door slamming at 0.069-0.75", "frequencyCaption": "door slamming one times"}
+{"filepath": "data/multi_event_train/syn_4977.wav", "onoffCaption": "duck quacking at 1.623-3.623, 4.63-6.63, 7.915-9.915", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4988.wav", "onoffCaption": "spraying at 2.075-2.807, 4.719-5.57", "frequencyCaption": "spraying two times"}
+{"filepath": "data/multi_event_train/syn_4992.wav", "onoffCaption": "door knocking at 0.681-5.056 and gunshot at 3.1-5.1, 7.191-9.191", "frequencyCaption": "door knocking one times and gunshot two times"}
+{"filepath": "data/multi_event_train/syn_2034.wav", "onoffCaption": "duck quacking at 0.504-2.504, 3.307-5.307", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2212.wav", "onoffCaption": "train horn at 2.927-6.697", "frequencyCaption": "train horn one times"}
+{"filepath": "data/multi_event_train/syn_2247.wav", "onoffCaption": "dog barking at 0.081-2.081 and car horn honking at 0.117-4.366 and cat meowing at 6.32-7.904", "frequencyCaption": "dog barking one times and car horn honking one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2346.wav", "onoffCaption": "door knocking at 0.207-3.582", "frequencyCaption": "door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2462.wav", "onoffCaption": "tapping clicking clanking at 0.119-3.559, 4.558-6.763", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2487.wav", "onoffCaption": "burping belching at 0.648-7.328 and door knocking at 2.358-6.126, 7.457-9.846", "frequencyCaption": "burping belching one times and door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2563.wav", "onoffCaption": "door slamming at 0.181-1.472 and explosion at 3.64-6.168 and dog barking at 4.954-6.954", "frequencyCaption": "door slamming one times and explosion one times and dog barking one times"}
+{"filepath": "data/multi_event_train/syn_2579.wav", "onoffCaption": "cat meowing at 0.093-2.087", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2611.wav", "onoffCaption": "tapping clicking clanking at 0.362-3.802, 6.013-9.453", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2710.wav", "onoffCaption": "duck quacking at 3.339-5.339, 7.228-9.228", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_2898.wav", "onoffCaption": "car horn honking at 2.151-6.473, 6.993-10.0", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_2966.wav", "onoffCaption": "train horn at 0.045-3.365, 4.328-7.648", "frequencyCaption": "train horn two times"}
+{"filepath": "data/multi_event_train/syn_2983.wav", "onoffCaption": "gunshot at 0.554-2.554, 3.272-5.272 and train horn at 3.312-7.312", "frequencyCaption": "gunshot two times and train horn one times"}
+{"filepath": "data/multi_event_train/syn_4022.wav", "onoffCaption": "car horn honking at 0.689-3.908", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4077.wav", "onoffCaption": "sheep goat bleating at 0.949-2.949, 3.937-5.937, 6.447-8.447", "frequencyCaption": "sheep goat bleating three times"}
+{"filepath": "data/multi_event_train/syn_4088.wav", "onoffCaption": "spraying at 2.992-5.452", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4092.wav", "onoffCaption": "explosion at 2.473-5.073", "frequencyCaption": "explosion one times"}
+{"filepath": "data/multi_event_train/syn_4139.wav", "onoffCaption": "whistling at 0.518-8.173", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4176.wav", "onoffCaption": "car horn honking at 0.455-4.042", "frequencyCaption": "car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4189.wav", "onoffCaption": "cat meowing at 0.138-1.498, 2.544-3.904, 5.026-6.386", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4193.wav", "onoffCaption": "woman laughing at 0.153-5.192 and cow mooing at 3.326-7.755", "frequencyCaption": "woman laughing one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4204.wav", "onoffCaption": "train horn at 3.537-7.977 and duck quacking at 6.259-8.259", "frequencyCaption": "train horn one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4421.wav", "onoffCaption": "cat meowing at 0.141-1.288 and sheep goat bleating at 0.715-4.035 and door knocking at 6.083-8.583", "frequencyCaption": "cat meowing one times and sheep goat bleating one times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_4474.wav", "onoffCaption": "burping belching at 2.963-5.378, 6.525-8.94", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4491.wav", "onoffCaption": "tapping clicking clanking at 2.783-6.223, 7.925-10.0", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_4520.wav", "onoffCaption": "dog barking at 1.345-4.665, 6.247-9.567", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_4652.wav", "onoffCaption": "thump thud at 0.178-2.678, 4.247-6.747 and gunshot at 0.859-2.859, 3.774-5.774, 6.922-8.922", "frequencyCaption": "thump thud two times and gunshot three times"}
+{"filepath": "data/multi_event_train/syn_4749.wav", "onoffCaption": "duck quacking at 1.65-3.65, 5.955-7.955", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4824.wav", "onoffCaption": "woman laughing at 0.292-2.408 and tapping clicking clanking at 4.885-8.325", "frequencyCaption": "woman laughing one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4925.wav", "onoffCaption": "spraying at 0.314-1.314 and cow mooing at 5.093-8.075", "frequencyCaption": "spraying one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4970.wav", "onoffCaption": "gunshot at 1.78-3.78, 5.766-7.766", "frequencyCaption": "gunshot two times"}
+{"filepath": "data/multi_event_train/syn_4995.wav", "onoffCaption": "duck quacking at 1.467-3.467, 5.668-7.668", "frequencyCaption": "duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_9.wav", "onoffCaption": "gunshot at 0.635-2.635", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2135.wav", "onoffCaption": "cat meowing at 4.338-5.423, 7.709-8.794", "frequencyCaption": "cat meowing two times"}
+{"filepath": "data/multi_event_train/syn_2160.wav", "onoffCaption": "car horn honking at 0.225-5.132, 6.96-9.007 and burping belching at 2.253-4.776", "frequencyCaption": "car horn honking two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2185.wav", "onoffCaption": "thump thud at 2.942-7.392", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_2208.wav", "onoffCaption": "woman laughing at 1.114-8.559", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2279.wav", "onoffCaption": "thump thud at 1.323-5.773, 7.153-10.0", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2313.wav", "onoffCaption": "duck quacking at 1.898-3.898, 4.892-6.892 and sheep goat bleating at 4.573-6.573", "frequencyCaption": "duck quacking two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2478.wav", "onoffCaption": "dog barking at 0.087-2.087, 3.299-5.299, 6.614-8.614", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2536.wav", "onoffCaption": "cat meowing at 0.427-3.668, 5.917-9.158 and woman laughing at 0.858-3.627", "frequencyCaption": "cat meowing two times and woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2586.wav", "onoffCaption": "dog barking at 1.358-3.358, 5.489-7.489", "frequencyCaption": "dog barking two times"}
+{"filepath": "data/multi_event_train/syn_2734.wav", "onoffCaption": "burping belching at 1.247-3.354 and spraying at 2.437-3.064, 4.131-5.215", "frequencyCaption": "burping belching one times and spraying two times"}
+{"filepath": "data/multi_event_train/syn_2745.wav", "onoffCaption": "cow mooing at 0.875-3.857, 5.932-8.914", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2832.wav", "onoffCaption": "explosion at 0.339-2.341, 4.836-7.714 and car horn honking at 1.971-4.757", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2867.wav", "onoffCaption": "tapping clicking clanking at 0.598-4.038, 5.634-9.074", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2882.wav", "onoffCaption": "cow mooing at 2.804-7.784", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2929.wav", "onoffCaption": "tapping clicking clanking at 1.97-5.41 and sheep goat bleating at 2.461-4.461, 4.975-6.975", "frequencyCaption": "tapping clicking clanking one times and sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2999.wav", "onoffCaption": "thump thud at 0.135-2.597, 3.618-6.08", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4038.wav", "onoffCaption": "cow mooing at 2.757-7.186", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4123.wav", "onoffCaption": "thump thud at 0.25-4.7, 5.564-8.064", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_4305.wav", "onoffCaption": "whistling at 2.362-7.862", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_4350.wav", "onoffCaption": "burping belching at 3.881-6.207, 6.742-9.742", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4575.wav", "onoffCaption": "explosion at 2.14-7.14, 7.84-10.0", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4590.wav", "onoffCaption": "thump thud at 1.677-4.724", "frequencyCaption": "thump thud one times"}
+{"filepath": "data/multi_event_train/syn_4648.wav", "onoffCaption": "gunshot at 3.838-6.078", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4706.wav", "onoffCaption": "explosion at 0.24-2.993, 4.501-7.501 and duck quacking at 2.746-4.746, 6.768-8.768", "frequencyCaption": "explosion two times and duck quacking two times"}
+{"filepath": "data/multi_event_train/syn_4753.wav", "onoffCaption": "gunshot at 2.893-4.893", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4871.wav", "onoffCaption": "car horn honking at 1.632-3.979, 5.537-8.355", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4894.wav", "onoffCaption": "gunshot at 1.001-3.001", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2111.wav", "onoffCaption": "duck quacking at 0.295-2.295, 2.851-4.851 and woman laughing at 1.563-5.615, 7.148-10.0", "frequencyCaption": "duck quacking two times and woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_2144.wav", "onoffCaption": "dog barking at 0.587-2.587, 3.375-5.375, 6.652-8.652", "frequencyCaption": "dog barking three times"}
+{"filepath": "data/multi_event_train/syn_2337.wav", "onoffCaption": "cow mooing at 0.134-5.114, 6.062-8.242", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2362.wav", "onoffCaption": "sheep goat bleating at 3.271-5.271, 7.111-9.111", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2387.wav", "onoffCaption": "sheep goat bleating at 2.984-4.984", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2409.wav", "onoffCaption": "cow mooing at 1.935-5.233, 6.077-9.059", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_2413.wav", "onoffCaption": "gunshot at 0.733-2.733", "frequencyCaption": "gunshot one times"}
+{"filepath": "data/multi_event_train/syn_2512.wav", "onoffCaption": "whistling at 3.548-10.0", "frequencyCaption": "whistling one times"}
+{"filepath": "data/multi_event_train/syn_2547.wav", "onoffCaption": "explosion at 0.219-2.221, 3.47-5.534 and sheep goat bleating at 6.765-8.765", "frequencyCaption": "explosion two times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2761.wav", "onoffCaption": "duck quacking at 2.498-4.498, 5.686-7.686 and explosion at 5.745-7.809", "frequencyCaption": "duck quacking two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2784.wav", "onoffCaption": "thump thud at 1.867-4.206, 5.36-7.699", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2816.wav", "onoffCaption": "door slamming at 2.818-5.599 and cow mooing at 2.905-7.885", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2843.wav", "onoffCaption": "burping belching at 2.029-7.973", "frequencyCaption": "burping belching one times"}
+{"filepath": "data/multi_event_train/syn_2958.wav", "onoffCaption": "spraying at 0.929-3.321", "frequencyCaption": "spraying one times"}
+{"filepath": "data/multi_event_train/syn_4049.wav", "onoffCaption": "woman laughing at 0.476-7.921 and car horn honking at 3.279-5.779", "frequencyCaption": "woman laughing one times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4107.wav", "onoffCaption": "cow mooing at 0.839-5.268, 6.236-10.0", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4152.wav", "onoffCaption": "burping belching at 1.002-3.625, 6.031-8.654", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_4321.wav", "onoffCaption": "sneeze at 0.133-1.379, 2.353-3.965 and whistling at 1.027-7.317", "frequencyCaption": "sneeze two times and whistling one times"}
+{"filepath": "data/multi_event_train/syn_4374.wav", "onoffCaption": "explosion at 0.919-3.672, 5.793-8.522 and burping belching at 5.774-8.976", "frequencyCaption": "explosion two times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4391.wav", "onoffCaption": "sheep goat bleating at 2.629-4.629", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4504.wav", "onoffCaption": "cat meowing at 0.188-1.374 and dog barking at 0.587-2.587, 3.86-5.86, 6.47-8.47", "frequencyCaption": "cat meowing one times and dog barking three times"}
+{"filepath": "data/multi_event_train/syn_4551.wav", "onoffCaption": "woman laughing at 3.557-6.642", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_4639.wav", "onoffCaption": "car horn honking at 1.638-4.813 and burping belching at 7.602-9.633", "frequencyCaption": "car horn honking one times and burping belching one times"}
+{"filepath": "data/multi_event_train/syn_4689.wav", "onoffCaption": "sheep goat bleating at 1.268-3.268, 5.556-7.556", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_4722.wav", "onoffCaption": "whistling at 1.885-7.385, 7.979-10.0", "frequencyCaption": "whistling two times"}
+{"filepath": "data/multi_event_train/syn_4777.wav", "onoffCaption": "train horn at 1.37-5.81, 7.867-10.0 and gunshot at 2.589-4.589", "frequencyCaption": "train horn two times and gunshot one times"}
+{"filepath": "data/multi_event_train/syn_4792.wav", "onoffCaption": "woman laughing at 1.201-3.395, 5.62-7.814", "frequencyCaption": "woman laughing two times"}
+{"filepath": "data/multi_event_train/syn_4800.wav", "onoffCaption": "train horn at 0.444-3.924 and sneeze at 2.515-4.198 and duck quacking at 6.469-8.469", "frequencyCaption": "train horn one times and sneeze one times and duck quacking one times"}
+{"filepath": "data/multi_event_train/syn_4855.wav", "onoffCaption": "door slamming at 0.924-3.353 and tapping clicking clanking at 6.856-10.0", "frequencyCaption": "door slamming one times and tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_2010.wav", "onoffCaption": "thump thud at 0.321-2.783 and cat meowing at 4.756-9.756", "frequencyCaption": "thump thud one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_2045.wav", "onoffCaption": "door slamming at 0.243-2.469 and cow mooing at 4.409-7.707", "frequencyCaption": "door slamming one times and cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_2236.wav", "onoffCaption": "thump thud at 1.714-3.942, 6.215-8.443", "frequencyCaption": "thump thud two times"}
+{"filepath": "data/multi_event_train/syn_2263.wav", "onoffCaption": "car horn honking at 2.479-6.801 and door slamming at 2.505-3.524, 5.637-6.656 and door knocking at 3.955-6.455", "frequencyCaption": "car horn honking one times and door slamming two times and door knocking one times"}
+{"filepath": "data/multi_event_train/syn_2286.wav", "onoffCaption": "sneeze at 0.115-1.349, 1.861-3.155, 3.884-6.727", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_2378.wav", "onoffCaption": "duck quacking at 1.898-3.898, 4.706-6.706, 7.974-9.974 and car horn honking at 3.594-6.38", "frequencyCaption": "duck quacking three times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_2446.wav", "onoffCaption": "door knocking at 3.375-5.999, 7.052-9.129", "frequencyCaption": "door knocking two times"}
+{"filepath": "data/multi_event_train/syn_2508.wav", "onoffCaption": "woman laughing at 1.307-3.532, 4.108-6.316 and explosion at 3.427-6.427", "frequencyCaption": "woman laughing two times and explosion one times"}
+{"filepath": "data/multi_event_train/syn_2635.wav", "onoffCaption": "sheep goat bleating at 0.781-2.781", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_2660.wav", "onoffCaption": "woman laughing at 3.713-7.765", "frequencyCaption": "woman laughing one times"}
+{"filepath": "data/multi_event_train/syn_2685.wav", "onoffCaption": "sheep goat bleating at 0.99-2.99, 5.032-7.088", "frequencyCaption": "sheep goat bleating two times"}
+{"filepath": "data/multi_event_train/syn_2859.wav", "onoffCaption": "burping belching at 2.044-5.55, 7.897-10.0", "frequencyCaption": "burping belching two times"}
+{"filepath": "data/multi_event_train/syn_2917.wav", "onoffCaption": "tapping clicking clanking at 2.146-5.586, 6.162-9.602", "frequencyCaption": "tapping clicking clanking two times"}
+{"filepath": "data/multi_event_train/syn_2942.wav", "onoffCaption": "duck quacking at 0.672-2.672, 5.106-7.106, 7.837-9.837", "frequencyCaption": "duck quacking three times"}
+{"filepath": "data/multi_event_train/syn_4006.wav", "onoffCaption": "tapping clicking clanking at 2.901-6.341", "frequencyCaption": "tapping clicking clanking one times"}
+{"filepath": "data/multi_event_train/syn_4053.wav", "onoffCaption": "cat meowing at 0.589-1.599, 2.313-3.34, 4.373-5.52", "frequencyCaption": "cat meowing three times"}
+{"filepath": "data/multi_event_train/syn_4148.wav", "onoffCaption": "spraying at 1.484-3.217, 3.795-5.052, 5.923-7.007", "frequencyCaption": "spraying three times"}
+{"filepath": "data/multi_event_train/syn_4220.wav", "onoffCaption": "cow mooing at 0.893-4.191", "frequencyCaption": "cow mooing one times"}
+{"filepath": "data/multi_event_train/syn_4275.wav", "onoffCaption": "train horn at 0.043-3.443 and sheep goat bleating at 5.656-7.656", "frequencyCaption": "train horn one times and sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4290.wav", "onoffCaption": "dog barking at 0.053-2.053 and sneeze at 1.421-3.008 and cat meowing at 1.901-3.087", "frequencyCaption": "dog barking one times and sneeze one times and cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4405.wav", "onoffCaption": "explosion at 0.051-3.051, 3.851-6.574", "frequencyCaption": "explosion two times"}
+{"filepath": "data/multi_event_train/syn_4450.wav", "onoffCaption": "sneeze at 2.386-3.62, 4.447-6.154, 7.403-10.0", "frequencyCaption": "sneeze three times"}
+{"filepath": "data/multi_event_train/syn_4623.wav", "onoffCaption": "cat meowing at 2.217-7.217", "frequencyCaption": "cat meowing one times"}
+{"filepath": "data/multi_event_train/syn_4676.wav", "onoffCaption": "sheep goat bleating at 0.572-2.572", "frequencyCaption": "sheep goat bleating one times"}
+{"filepath": "data/multi_event_train/syn_4693.wav", "onoffCaption": "cow mooing at 2.486-5.455, 7.47-9.667", "frequencyCaption": "cow mooing two times"}
+{"filepath": "data/multi_event_train/syn_4738.wav", "onoffCaption": "door slamming at 3.65-4.628, 6.613-8.613", "frequencyCaption": "door slamming two times"}
+{"filepath": "data/multi_event_train/syn_4788.wav", "onoffCaption": "explosion at 0.578-4.447, 5.213-9.082 and car horn honking at 0.746-3.564", "frequencyCaption": "explosion two times and car horn honking one times"}
+{"filepath": "data/multi_event_train/syn_4901.wav", "onoffCaption": "car horn honking at 2.729-5.642, 6.454-8.801", "frequencyCaption": "car horn honking two times"}
+{"filepath": "data/multi_event_train/syn_4954.wav", "onoffCaption": "cat meowing at 0.74-3.77", "frequencyCaption": "cat meowing one times"}
diff --git a/picoaudio/models/controllable_dataset.py b/picoaudio/models/controllable_dataset.py
new file mode 100644
index 0000000000000000000000000000000000000000..cb9468bcc3fb9f5e6c667c0dbf83ed010ddbe87e
--- /dev/null
+++ b/picoaudio/models/controllable_dataset.py
@@ -0,0 +1,126 @@
+import numpy as np
+import torch 
+import pandas as pd
+from data.filter_data import get_event_list
+
+
+
+class Text_Onset_2_Audio_Dataset(torch.utils.data.Dataset):
+    def __init__(self, dataset, args):
+
+        self.captions = list(dataset[args.text_column])
+        self.audios = list(dataset[args.audio_column])
+        self.onsets = list(dataset[args.onset_column])
+        self.indices = list(range(len(self.captions)))
+
+        self.mapper = {}
+        for index, audio, caption, onset in zip(self.indices, self.audios, self.captions, self.onsets):
+            self.mapper[index] = [audio, caption, onset]
+
+        num_examples = args.num_examples
+        if num_examples != -1:
+            self.captions, self.audios, self.onsets = self.captions[:num_examples], self.audios[:num_examples], self.onsets[:num_examples]
+            self.indices = self.indices[:num_examples]
+        self.class2id = {event: idx for idx, event in enumerate(args.event_list)}
+
+    def decode_data(self, line_onset_str):
+        # data    { "location":     audio_path,
+        #           "captions" :    "event1 n times and event2 n times",
+        #           "onset_str":        "event1__onset1-offset1_onset2-offset2--event2__onset1-offset1"}
+        
+        line_onset_index = np.zeros((32, 256))
+        line_event = []
+        for event_onset in line_onset_str.split('--'):
+            # event_onset : event1__onset1-offset1_onset2-offset2
+            (event, instance) = event_onset.split('__')
+            line_event.append(event)
+            # instance : onset1-offset1_onset2-offset2
+            for start_end in instance.split('_'):
+                (start, end) = start_end.split('-')         
+                start, end = int(float(start)*250/10), int(float(end)*250/10)
+                if end > 255: break
+                line_onset_index[self.class2id[event], start: end] = 1
+        line_event_str = " and ".join(line_event)
+        return line_onset_index, line_event_str
+      
+    def __len__(self):
+        return len(self.captions)
+
+    def get_num_instances(self):
+        return len(self.captions)
+
+    def __getitem__(self, index):
+        onset_str, filename, idx, caption = self.onsets[index], self.audios[index], self.indices[index], self.captions[index]
+        onset, _ = self.decode_data(onset_str)
+        #"onset_str":        "event1__onset1-offset1_onset2-offset2--event2__onset1-offset1"
+        #assert len(onset_str.split("--")) == 1
+        first_class_id = self.class2id[onset_str.split("__")[0]]
+        return idx, onset, first_class_id, filename, caption, onset_str
+
+    def collate_fn(self, data):
+        dat = pd.DataFrame(data)
+        batch = []
+        for i in dat:     
+            if i==1:
+                batch.append(torch.tensor(np.array(dat[i].tolist()), dtype=torch.float32))
+            elif i==2:
+                batch.append(torch.tensor(dat[i]))
+            else:
+                batch.append(dat[i].tolist())
+        return batch
+    
+class Clap_Onset_2_Audio_Dataset(Text_Onset_2_Audio_Dataset):
+    def __init__(self, dataset, args):
+        super().__init__(dataset, args)
+        import laion_clap
+        from laion_clap.clap_module.factory import load_state_dict as clap_load_state_dict
+
+        self.clap_scorer = laion_clap.CLAP_Module(enable_fusion=False)
+        ckpt_path = 'miniconda3/envs/py3.10.11/lib/python3.10/site-packages/laion_clap/630k-audioset-best.pt'
+        ckpt = clap_load_state_dict(ckpt_path, skip_params=True)
+        del_parameter_key = ["text_branch.embeddings.position_ids"]
+        ckpt = {"model."+k:v for k, v in ckpt.items() if k not in del_parameter_key}
+        self.clap_scorer.load_state_dict(ckpt)
+
+    def __getitem__(self, index):
+        onset_str, filename, idx, caption = self.onsets[index], self.audios[index], self.indices[index], self.captions[index]
+        onset, event = self.decode_data(onset_str)
+        with torch.no_grad():
+            clap_embed = self.clap_scorer.get_text_embedding([event, ""], use_tensor=False)[0]
+        return idx, onset, clap_embed, filename, caption, onset_str
+
+    def collate_fn(self, data):
+        dat = pd.DataFrame(data)
+        batch = []
+        for i in dat:     
+            if i==1 or i==2:
+                batch.append(torch.tensor(np.array(dat[i].tolist()), dtype=torch.float32))
+            else:
+                batch.append(dat[i].tolist())
+        return batch
+
+
+if __name__ == "__main__":
+    import torch
+    from torch.utils.data import Dataset, DataLoader
+    import datasets
+    import argparse
+    import sys
+    
+    import models.controllable_dataset as ConDataset
+    from data_utils.filter_data import get_event_list
+
+    
+    parser = argparse.ArgumentParser(description=".")
+    args = parser.parse_args()
+    args.event_list = get_event_list()
+    args.train_file = ""
+    
+
+    extension = args.train_file.split(".")[-1]
+    raw_datasets = load_dataset(extension, data_files={"train": args.train_file})
+    train_dataset = Clap_Onset_2_Audio_Dataset(raw_datasets["train"], args)
+    train_dataloader = DataLoader(train_dataset, shuffle=True, batch_size=args.batch_size, collate_fn=train_dataset.collate_fn)
+    for batch in train_dataloader:
+        import pdb; pdb.set_trace()
+        idx, onset, event_info, audios, caption, onset_str = batch
diff --git a/picoaudio/models/controllable_diffusion.py b/picoaudio/models/controllable_diffusion.py
new file mode 100644
index 0000000000000000000000000000000000000000..fd12a014a013eaed7657ded3be07e7c198b74dec
--- /dev/null
+++ b/picoaudio/models/controllable_diffusion.py
@@ -0,0 +1,291 @@
+import random
+import numpy as np
+from tqdm import tqdm
+from einops import repeat
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+
+from transformers import CLIPTokenizer, AutoTokenizer
+from transformers import CLIPTextModel, T5EncoderModel, AutoModel
+import diffusers
+from diffusers.utils.torch_utils import randn_tensor
+from diffusers import DDPMScheduler, UNet2DConditionModel
+from diffusers import AutoencoderKL as DiffuserAutoencoderKL
+
+from utils.torch_tools import wav_to_fbank
+from audioldm.audio.stft import TacotronSTFT
+from audioldm.variational_autoencoder.autoencoder import AutoencoderKL
+from audioldm.utils import default_audioldm_config, get_metadata
+
+def build_pretrained_models(name):
+    checkpoint = torch.load(get_metadata()[name]["path"], map_location="cpu")
+    scale_factor = checkpoint["state_dict"]["scale_factor"].item()
+
+    vae_state_dict = {k[18:]: v for k, v in checkpoint["state_dict"].items() if "first_stage_model." in k}
+
+    config = default_audioldm_config(name)
+    vae_config = config["model"]["params"]["first_stage_config"]["params"]
+    vae_config["scale_factor"] = scale_factor
+
+    vae = AutoencoderKL(**vae_config)
+    vae.load_state_dict(vae_state_dict)
+
+    fn_STFT = TacotronSTFT(
+        config["preprocessing"]["stft"]["filter_length"],
+        config["preprocessing"]["stft"]["hop_length"],
+        config["preprocessing"]["stft"]["win_length"],
+        config["preprocessing"]["mel"]["n_mel_channels"],
+        config["preprocessing"]["audio"]["sampling_rate"],
+        config["preprocessing"]["mel"]["mel_fmin"],
+        config["preprocessing"]["mel"]["mel_fmax"],
+    )
+
+    vae.eval()
+    fn_STFT.eval()
+
+    return vae, fn_STFT
+
+def _init_layer(layer):
+    """Initialize a Linear or Convolutional layer. """
+    nn.init.xavier_uniform_(layer.weight)
+ 
+    if hasattr(layer, 'bias'):
+        if layer.bias is not None:
+            layer.bias.data.fill_(0.)
+
+class BaseDiffusion(nn.Module):
+    def __init__(
+        self,
+        scheduler_name,
+        unet_model_config_path=None,
+        snr_gamma=None,
+        uncondition=False,
+    ):
+        super().__init__()
+
+        assert unet_model_config_path is not None, "Either UNet pretrain model name or a config file path is required"
+        self.scheduler_name = scheduler_name
+        self.unet_model_config_path = unet_model_config_path
+        self.snr_gamma = snr_gamma
+        self.uncondition = uncondition
+        self.device = "cuda" if torch.cuda.is_available() else "cpu"
+        # https://huggingface.co/docs/diffusers/v0.14.0/en/api/schedulers/overview
+        self.noise_scheduler = DDPMScheduler.from_pretrained(self.scheduler_name, subfolder="scheduler")
+        self.inference_scheduler = DDPMScheduler.from_pretrained(self.scheduler_name, subfolder="scheduler")
+        unet_config = UNet2DConditionModel.load_config(unet_model_config_path)
+        self.unet = UNet2DConditionModel.from_config(unet_config, subfolder="unet")
+        print("UNet initialized randomly.")
+        """
+        self.text_encoder_name = "./checkpoint/models--google--flan-t5-large/" + \
+           "snapshots/0613663d0d48ea86ba8cb3d7a44f0f65dc596a2a/"
+        self.tokenizer = AutoTokenizer.from_pretrained(self.text_encoder_name)
+        self.text_encoder = T5EncoderModel.from_pretrained(self.text_encoder_name)
+        """
+
+    def compute_snr(self, timesteps):
+        """
+        Computes SNR as per https://github.com/TiankaiHang/Min-SNR-Diffusion-Training/blob/521b624bd70c67cee4bdf49225915f5945a872e3/guided_diffusion/gaussian_diffusion.py#L847-L849
+        """
+        alphas_cumprod = self.noise_scheduler.alphas_cumprod
+        sqrt_alphas_cumprod = alphas_cumprod**0.5
+        sqrt_one_minus_alphas_cumprod = (1.0 - alphas_cumprod) ** 0.5
+
+        # Expand the tensors.
+        # Adapted from https://github.com/TiankaiHang/Min-SNR-Diffusion-Training/blob/521b624bd70c67cee4bdf49225915f5945a872e3/guided_diffusion/gaussian_diffusion.py#L1026
+        sqrt_alphas_cumprod = sqrt_alphas_cumprod.to(device=timesteps.device)[timesteps].float()
+        while len(sqrt_alphas_cumprod.shape) < len(timesteps.shape):
+            sqrt_alphas_cumprod = sqrt_alphas_cumprod[..., None]
+        alpha = sqrt_alphas_cumprod.expand(timesteps.shape)
+
+        sqrt_one_minus_alphas_cumprod = sqrt_one_minus_alphas_cumprod.to(device=timesteps.device)[timesteps].float()
+        while len(sqrt_one_minus_alphas_cumprod.shape) < len(timesteps.shape):
+            sqrt_one_minus_alphas_cumprod = sqrt_one_minus_alphas_cumprod[..., None]
+        sigma = sqrt_one_minus_alphas_cumprod.expand(timesteps.shape)
+
+        # Compute SNR.
+        snr = (alpha / sigma) ** 2
+        return snr
+
+    def encode_text(self, input_dict):    
+        raise NotImplementedError  
+
+    def forward(self, input_dict):
+        raise NotImplementedError
+
+    @torch.no_grad()
+    def inference(self, input_dict):
+        raise NotImplementedError
+
+class Text_Onset_2_Audio_Diffusion(BaseDiffusion):
+    def __init__(self,
+        scheduler_name,
+        unet_model_config_path=None,
+        snr_gamma=None,
+        freeze_text_encoder=True,
+        uncondition=False,
+    ):
+        super().__init__(scheduler_name, unet_model_config_path, snr_gamma, uncondition)
+        self.freeze_text_encoder = freeze_text_encoder
+        self.class_emb = nn.Embedding(24, 1024)
+        # self.channel_emb = nn.Linear(24, 16)
+        # _init_layer(self.channel_emb)
+
+    def encode_channel(self, input):
+        # input [batch, 32, 256] -> [batch, 2, 256, 16]
+        return input.reshape(input.shape[0], 2, 16, 256).transpose(2, 3)
+        # return self.channel_emb(input).unsqueeze(1)
+
+    def encode_text(self, input_dict):      
+        device = self.device   
+
+        encoder_hidden_states = self.class_emb(input_dict["event_info"].unsqueeze(-1))
+        boolean_encoder_mask = (torch.ones(len(encoder_hidden_states), 1) == 1).to(device)
+
+        return encoder_hidden_states, boolean_encoder_mask
+
+    def forward(self, input_dict, validation_mode=False):
+        device = self.device
+        latents = input_dict["latent"]
+        num_train_timesteps = self.noise_scheduler.num_train_timesteps
+        self.noise_scheduler.set_timesteps(num_train_timesteps, device=device)
+
+       
+        # [batch, 1, 1024], [batch, 1]
+
+        encoder_hidden_states, boolean_encoder_mask = self.encode_text(input_dict)
+        if self.uncondition:
+            mask_indices = [k for k in range(len(latents)) if random.random() < 0.1]
+            if len(mask_indices) > 0:
+                encoder_hidden_states[mask_indices] = 0
+
+        bsz = latents.shape[0]
+        if validation_mode:
+            timesteps = (self.noise_scheduler.num_train_timesteps//2) * torch.ones((bsz,), dtype=torch.int64, device=device)
+        else:
+            # Sample a random timestep for each instance
+            timesteps = torch.randint(0, self.noise_scheduler.num_train_timesteps, (bsz,), device=device)
+        timesteps = timesteps.long()
+
+        noise = torch.randn_like(latents)
+        noisy_latents = self.noise_scheduler.add_noise(latents, noise, timesteps)
+        
+        onset_emb = self.encode_channel(input_dict["onset"])
+        # [batch, channel:8, 256, 16] + [batch, onset:2, 256, 16]
+        onset_noisy_latents = torch.cat((onset_emb, noisy_latents), dim=1)
+
+        # Get the target for loss depending on the prediction type
+        if self.noise_scheduler.config.prediction_type == "epsilon":
+            target = noise
+        elif self.noise_scheduler.config.prediction_type == "v_prediction":
+            target = self.noise_scheduler.get_velocity(latents, noise, timesteps)
+        else:
+            raise ValueError(f"Unknown prediction type {self.noise_scheduler.config.prediction_type}")
+        
+        model_pred = self.unet(
+            onset_noisy_latents, timesteps, encoder_hidden_states, 
+            #encoder_attention_mask=boolean_encoder_mask
+        ).sample
+
+        if self.snr_gamma is None:
+            loss = F.mse_loss(model_pred.float(), target.float(), reduction="mean")
+        else:
+            # Compute loss-weights as per Section 3.4 of https://arxiv.org/abs/2303.09556.
+            # Adaptef from huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image.py
+            snr = self.compute_snr(timesteps)
+            mse_loss_weights = (
+                torch.stack([snr, self.snr_gamma * torch.ones_like(timesteps)], dim=1).min(dim=1)[0] / snr
+            )
+            loss = F.mse_loss(model_pred.float(), target.float(), reduction="none")
+            loss = loss.mean(dim=list(range(1, len(loss.shape)))) * mse_loss_weights
+            loss = loss.mean()
+
+        return loss
+
+    def prepare_latents(self, batch_size, inference_scheduler, num_channels_latents, dtype, device):
+        shape = (batch_size, num_channels_latents, 256, 16)
+        latents = randn_tensor(shape, generator=None, device=device, dtype=dtype)
+        # scale the initial noise by the standard deviation required by the scheduler
+        latents = latents * inference_scheduler.init_noise_sigma
+        return latents
+
+    def encode_text_classifier_free(self, input_dict, num_samples_per_prompt):
+        device = self.device
+        prompt_embeds, boolean_prompt_mask = self.encode_text(input_dict)
+        prompt_embeds = prompt_embeds.repeat_interleave(num_samples_per_prompt, 0)
+        attention_mask = boolean_prompt_mask.repeat_interleave(num_samples_per_prompt, 0)
+        # get unconditional embeddings for classifier free guidance
+        negative_prompt_embeds = torch.zeros(prompt_embeds.shape).to(device)
+        uncond_attention_mask = (torch.ones(attention_mask.shape) == 1).to(device)            
+        # negative_prompt_embeds = negative_prompt_embeds.repeat_interleave(num_samples_per_prompt, 0)
+        # uncond_attention_mask = uncond_attention_mask.repeat_interleave(num_samples_per_prompt, 0)
+
+        # For classifier free guidance, we need to do two forward passes.
+        # We concatenate the unconditional and text embeddings into a single batch to avoid doing two forward passes
+        prompt_embeds = torch.cat([negative_prompt_embeds, prompt_embeds])
+        prompt_mask = torch.cat([uncond_attention_mask, attention_mask])
+        boolean_prompt_mask = (prompt_mask == 1).to(device)
+
+        return prompt_embeds, boolean_prompt_mask
+
+    @torch.no_grad()
+    def inference(self, input_dict, inference_scheduler, num_steps=20, guidance_scale=3, num_samples_per_prompt=1, 
+                  disable_progress=True):
+        prompt = input_dict["onset"]
+        device = self.device
+        classifier_free_guidance = guidance_scale > 1.0
+        batch_size = len(prompt) * num_samples_per_prompt
+
+        if classifier_free_guidance:
+            prompt_embeds, boolean_prompt_mask = self.encode_text_classifier_free(input_dict, num_samples_per_prompt)
+        else:
+            prompt_embeds, boolean_prompt_mask = self.encode_text(input_dict)
+            prompt_embeds = prompt_embeds.repeat_interleave(num_samples_per_prompt, 0)
+            boolean_prompt_mask = boolean_prompt_mask.repeat_interleave(num_samples_per_prompt, 0)
+
+        inference_scheduler.set_timesteps(num_steps, device=device)
+        timesteps = inference_scheduler.timesteps
+        
+        num_channels_latents = self.unet.config.in_channels - 2
+        latents = self.prepare_latents(batch_size, inference_scheduler, num_channels_latents, prompt_embeds.dtype, device)
+        onset_emb = self.encode_channel(input_dict["onset"]).repeat_interleave(num_samples_per_prompt, 0)
+        onset_latents = torch.cat((onset_emb, latents), dim=1)
+
+        num_warmup_steps = len(timesteps) - num_steps * inference_scheduler.order
+        progress_bar = tqdm(range(num_steps), disable=disable_progress)
+        
+        for i, t in tqdm(enumerate(timesteps)):
+            # expand the latents if we are doing classifier free guidance
+            latent_model_input = torch.cat([onset_latents] * 2) if classifier_free_guidance else onset_latents
+            latent_model_input = inference_scheduler.scale_model_input(latent_model_input, t)
+            noise_pred = self.unet(
+                latent_model_input, t, encoder_hidden_states=prompt_embeds,
+                encoder_attention_mask=boolean_prompt_mask
+            ).sample
+
+            # perform guidance
+            if classifier_free_guidance:
+                noise_pred_uncond, noise_pred_text = noise_pred.chunk(2)
+                noise_pred = noise_pred_uncond + guidance_scale * (noise_pred_text - noise_pred_uncond)
+
+            # compute the previous noisy sample x_t -> x_t-1
+            latents = inference_scheduler.step(noise_pred, t, latents).prev_sample
+            onset_latents = torch.cat((onset_emb, latents), dim=1)
+
+            # call the callback, if provided
+            if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % inference_scheduler.order == 0):
+                progress_bar.update(1)
+
+
+        return latents
+
+class ClapText_Onset_2_Audio_Diffusion(Text_Onset_2_Audio_Diffusion):
+    def encode_text(self, input_dict):      
+        device = self.device   
+        
+        encoder_hidden_states = input_dict["event_info"].repeat_interleave(2, -1).unsqueeze(1)
+        boolean_encoder_mask = (torch.ones(len(encoder_hidden_states), 1) == 1).to(device)
+
+        return encoder_hidden_states, boolean_encoder_mask
+
diff --git a/picoaudio/requirements.txt b/picoaudio/requirements.txt
new file mode 100644
index 0000000000000000000000000000000000000000..9eaf4207459bfa08ecd90d0350d954ce1a089f20
--- /dev/null
+++ b/picoaudio/requirements.txt
@@ -0,0 +1,30 @@
+torch==2.0.1
+torchaudio==2.0.2
+torchvision==0.15.2
+transformers==4.37.2
+accelerate==0.26.1
+datasets==2.16.1
+diffusers==0.18.2
+einops==0.7.0
+h5py==3.10.0
+huggingface_hub==0.20.3
+importlib_metadata==7.0.1
+librosa==0.10.1
+matplotlib==3.8.2
+numpy==1.23.5
+omegaconf==2.0.6
+packaging==23.2
+pandas==2.2.0
+progressbar33==2.4
+protobuf==3.20.*
+resampy==0.4.2
+scikit_image==0.22.0
+scikit_learn==1.4.0
+scipy==1.12.0
+soundfile==0.12.1
+ssr_eval==0.0.7
+torchlibrosa==0.1.0
+tqdm==4.63.1
+laion-clap==1.1.4
+gradio
+google-generativeai
\ No newline at end of file
diff --git a/picoaudio/runner/controllable_inference.py b/picoaudio/runner/controllable_inference.py
new file mode 100644
index 0000000000000000000000000000000000000000..9270119ee2668521972e113ff9a1eb219b9bef7e
--- /dev/null
+++ b/picoaudio/runner/controllable_inference.py
@@ -0,0 +1,165 @@
+
+import os
+import sys
+import copy
+import json
+import time
+import random
+import argparse
+import soundfile as sf
+import numpy as np
+import librosa
+import torchaudio
+from tqdm import tqdm
+import laion_clap
+from laion_clap.clap_module.factory import load_state_dict as clap_load_state_dict
+from sklearn.metrics.pairwise import cosine_similarity
+
+import torch
+from datetime import datetime
+from diffusers import DDPMScheduler
+from datasets import load_dataset
+from torch.utils.data import Dataset, DataLoader
+
+import models.controllable_diffusion as ConDiffusion
+import models.controllable_dataset as ConDataset
+from data.filter_data import get_event_list
+
+class dotdict(dict):
+    """dot.notation access to dictionary attributes"""
+    __getattr__ = dict.get
+    __setattr__ = dict.__setitem__
+    __delattr__ = dict.__delitem__
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="Inference for text to audio generation task.")
+    parser.add_argument(
+        "--exp_path", '-exp', type=str, default=None,
+        help="Path for experiment."
+    )
+    parser.add_argument(
+        "--test_file", type=str, default="data/meta_data/test-onoff-control_multi-event.json",
+        help="Path for test_file."
+    )
+    parser.add_argument(
+        "--original_args", type=str, default="summary.jsonl",
+        help="Path for summary jsonl file saved during training."
+    )
+    parser.add_argument(
+        "--model_pt", type=str, default="best.pt",
+        help="Path for saved model bin file."
+    )
+    parser.add_argument(
+        "--num_steps", type=int, default=200,
+        help="How many denoising steps for generation.",
+    )
+    parser.add_argument(
+        "--guidance", '-g', type=float, 
+        #default=3,
+        default=1,
+        help="Guidance scale for classifier free guidance."
+    )
+    parser.add_argument(
+        "--batch_size", '-b', type=int, default=32,
+        help="Batch size for generation.",
+    )
+    parser.add_argument(
+        "--num_samples", type=int, default=1,
+        help="How many samples per prompt.",
+    )
+    parser.add_argument(
+        "--seed", type=int, default=0,
+        help="seed.",
+    )
+    
+    args = parser.parse_args()
+    
+    args.original_args = os.path.join(args.exp_path, args.original_args)    
+    args.model_pt = os.path.join(args.exp_path, args.model_pt)
+    return args
+
+def main():
+    args = parse_args() 
+    train_args = dotdict(json.loads(open(args.original_args).readlines()[0]))
+  
+    seed = args.seed
+    random.seed(seed)
+    np.random.seed(seed)
+    torch.manual_seed(seed)    
+    if torch.cuda.is_available():
+        torch.backends.cudnn.deterministic = True
+        torch.backends.cudnn.benchmark = False
+
+    # Prepare Data #
+    extension = args.test_file.split(".")[-1]
+    raw_datasets = load_dataset(extension, data_files={"test": args.test_file})
+    test_dataset = getattr(ConDataset, train_args.dataset_class)(raw_datasets["test"], train_args)
+    test_dataloader = DataLoader(test_dataset, shuffle=False, batch_size=args.batch_size, collate_fn=test_dataset.collate_fn)
+
+    
+    # Load Models #  
+    print("\n------Load model")
+    name = "audioldm-s-full"
+    vae, stft = ConDiffusion.build_pretrained_models(name)
+    vae, stft = vae.cuda(), stft.cuda()
+    print(train_args.model_class)
+    model = getattr(ConDiffusion, train_args.model_class)(
+        scheduler_name=train_args.scheduler_name, 
+        unet_model_config_path=train_args.unet_model_config, 
+        snr_gamma=train_args.snr_gamma
+    ).cuda().eval()
+    
+    # Load Trained Weight #
+    device = vae.device()
+    model.load_state_dict(torch.load(args.model_pt))   
+    scheduler = DDPMScheduler.from_pretrained(train_args.scheduler_name, subfolder="scheduler")
+    
+
+    # Generate #
+    num_steps, guidance, batch_size, num_samples = args.num_steps, args.guidance, args.batch_size, args.num_samples
+    audio_len = 16000 * 10
+    output_dir = os.path.join(CONTROLLABLE_PATH, 
+        f"synthesized/{'-'.join(args.model_pt.split('/')[-3:-1])}_steps-{num_steps}_guidance-{guidance}_samples-{num_samples}_{args.test_file.split('/')[-1].split('.')[0]}/")
+    os.makedirs(output_dir, exist_ok=True)
+    print(f"------Write to files to {output_dir}")
+
+    print("------Diffusion begin!")    
+    if args.num_samples == 1:
+        for batch in tqdm(test_dataloader):
+            idx, onset, event_info, _, _, onset_str = batch # idx, onset, event_info, audios, caption, onset_str
+            with torch.no_grad():
+                latents = model.inference({"onset":onset.to(device), "event_info":event_info.to(device)}, scheduler, num_steps, guidance, num_samples, disable_progress=True)
+                mel = vae.decode_first_stage(latents)
+                wave = vae.decode_to_waveform(mel)
+            for j, wav in enumerate(wave):
+                sf.write(f"{output_dir}/{idx[j]}--{onset_str[j]}.wav", wav[:audio_len], samplerate=16000, subtype='PCM_16')
+    else:
+        print("Clap scorer filter")
+        clap_scorer = laion_clap.CLAP_Module(enable_fusion=False)
+        ckpt_path = 'miniconda3/envs/py3.10.11/lib/python3.10/site-packages/laion_clap/630k-audioset-best.pt'
+        ckpt = clap_load_state_dict(ckpt_path, skip_params=True)
+        del_parameter_key = ["text_branch.embeddings.position_ids"]
+        ckpt = {"model."+k:v for k, v in ckpt.items() if k not in del_parameter_key}
+        clap_scorer.load_state_dict(ckpt)
+        for batch in tqdm(test_dataloader):
+            _, onset, event_info, _, caption, onset_str = batch # idx, onset, event_info, audios, caption, onset_str
+            with torch.no_grad():
+                latents = model.inference({"onset":onset.to(device), "event_info":event_info.to(device)}, scheduler, num_steps, guidance, num_samples, disable_progress=True)
+                mel = vae.decode_first_stage(latents)
+                wave = vae.decode_to_waveform(mel)
+            for j in range(batch_size):
+                text_embed = clap_scorer.get_text_embedding([caption[j], ""], use_tensor=False)[:1]
+                best_idx, best_score = 0, float('-inf')
+                for candidate_idx in range(num_samples):
+                    wav_48k = librosa.core.resample(wave[j*num_samples+candidate_idx].astype(np.float64)[:audio_len] / 32768, orig_sr=16000, target_sr=48000).reshape(1, -1)
+                    audio_embed = clap_scorer.get_audio_embedding_from_data(x = wav_48k)  
+                    pair_similarity = cosine_similarity(audio_embed, text_embed)[0][0]
+                    if pair_similarity > best_score:
+                        best_score = pair_similarity
+                        best_idx = candidate_idx
+                sf.write(f"{output_dir}/{idx[j]}--{onset_str[j]}.wav", wave[j * num_samples + best_idx][:audio_len], samplerate=16000, subtype='PCM_16')
+        
+    
+        
+if __name__ == "__main__":
+    main()
\ No newline at end of file
diff --git a/picoaudio/runner/controllable_train.py b/picoaudio/runner/controllable_train.py
new file mode 100644
index 0000000000000000000000000000000000000000..01f28726a9a92d6d8dfba47e19414813877cd7db
--- /dev/null
+++ b/picoaudio/runner/controllable_train.py
@@ -0,0 +1,393 @@
+import os
+import math
+import json
+import numpy as np
+import pandas as pd
+import random
+import logging
+import argparse
+import diffusers
+import transformers
+from transformers import SchedulerType, get_scheduler
+from tqdm.auto import tqdm
+from datetime import datetime
+
+import torch
+from torch.utils.data import Dataset, DataLoader
+import datasets
+from datasets import load_dataset
+from accelerate import Accelerator
+from accelerate.logging import get_logger
+from accelerate.utils import set_seed
+
+import sys
+import utils.torch_tools as torch_tools
+import models.controllable_diffusion as ConDiffusion
+import models.controllable_dataset as ConDataset
+from data.filter_data import get_event_list
+
+logger = get_logger(__name__)
+
+def parse_args():
+    parser = argparse.ArgumentParser(description="Finetune a diffusion model for text to audio generation task.")
+    parser.add_argument(
+        "--train_file", '-f', type=str, default="data/meta_data/train.json"
+    )
+    parser.add_argument(
+        "--batch_size", '-b', type=int, default=1,
+        help="Batch size (per device) for the training dataloader.",
+    )
+    parser.add_argument(
+        "--learning_rate", '-lr', type=float, default=3e-5,
+        help="Initial learning rate (after the potential warmup period) to use.",
+    )
+    parser.add_argument(
+        "--num_epochs", '-e', type=int, default=40,
+        help="Total number of training epochs to perform."
+    )
+    parser.add_argument(
+        "--output_dir", '-o', type=str, default=None,
+        help="Where to store the final model."
+    )
+    parser.add_argument(
+        "--model_class", '-m', type=str, default="ClapText_Onset_2_Audio_Diffusion", #TextOnset2AudioDiffusion
+        help="name of model_class"
+    )
+    parser.add_argument(
+        "--dataset_class", '-dc', type=str, default="Clap_Onset_2_Audio_Dataset", #Text_Onset2AudioDataset
+        help="name of model_class"
+    )
+    parser.add_argument(
+        "--duration", '-d', type=float, default=10,
+        help="Audio duration."
+    )
+    parser.add_argument(
+        "--num_examples", '-n', type=int, default=-1,
+        help="How many examples to use for training.",
+    )
+    parser.add_argument(     
+        "--scheduler_name", type=str, 
+        default="stabilityai/stable-diffusion-2-1",
+        help="Scheduler identifier.",
+    )
+    parser.add_argument(
+        "--unet_model_config", type=str, default="utils/configs/frequency.json",
+        help="UNet model config json path.",
+    )
+    parser.add_argument(
+        "--text_column", type=str, default="captions",
+        help="The name of the column in the datasets containing the input texts.",
+    )
+    parser.add_argument(
+        "--onset_column", type=str, default="onset",
+        help="The name of the column in the datasets containing the osnet.",
+    )
+    parser.add_argument(
+        "--audio_column", type=str, default="location",
+        help="The name of the column in the datasets containing the audio paths.",
+    )   
+    if True:
+        parser.add_argument(
+            "--augment", action="store_true", default=False,
+            help="Augment training data.",
+        )
+        parser.add_argument(
+            "--uncondition", action="store_true", default=False,
+            help="10% uncondition for training.",
+        )
+        parser.add_argument(
+            "--weight_decay", type=float, default=1e-8,
+            help="Weight decay to use."
+        )
+        parser.add_argument(
+            "--snr_gamma", type=float, 
+            #default=None,
+            default=5.0,
+            help="SNR weighting gamma to be used if rebalancing the loss. Recommended value is 5.0. "
+            "More details here: https://arxiv.org/abs/2303.09556.",
+        )
+        parser.add_argument(
+            "--max_train_steps", type=int, default=None,
+            help="Total number of training steps to perform. If provided, overrides num_epochs.",
+        )
+        parser.add_argument(
+            "--gradient_accumulation_steps", type=int, default=4,
+            help="Number of updates steps to accumulate before performing a backward/update pass.",
+        )
+        parser.add_argument(
+            "--lr_scheduler_type", type=SchedulerType, default="linear",
+            help="The scheduler type to use.",
+            choices=["linear", "cosine", "cosine_with_restarts", "polynomial", "constant", "constant_with_warmup"],
+        )
+        parser.add_argument(
+            "--num_warmup_steps", type=int, default=0,
+            help="Number of steps for the warmup in the lr scheduler."
+        )
+        parser.add_argument(
+            "--adam_beta1", type=float, default=0.9,
+            help="The beta1 parameter for the Adam optimizer."
+        )
+        parser.add_argument(
+            "--adam_beta2", type=float, default=0.999,
+            help="The beta2 parameter for the Adam optimizer."
+        )
+        parser.add_argument(
+            "--adam_weight_decay", type=float, default=1e-2,
+            help="Weight decay to use."
+        )
+        parser.add_argument(
+            "--adam_epsilon", type=float, default=1e-08,
+            help="Epsilon value for the Adam optimizer"
+        )
+        parser.add_argument(
+            "--seed", type=int, default=0,
+            help="A seed for reproducible training."
+        )
+        parser.add_argument(
+            "--checkpointing_steps", type=str, default="best",
+            help="Whether the various states should be saved at the end of every 'epoch' or 'best' whenever validation loss decreases.",
+        )
+        parser.add_argument(
+            "--save_every", type=int, default=40,
+            help="Save model after every how many epochs when checkpointing_steps is set to best."
+        )
+        parser.add_argument(
+            "--resume_from_checkpoint", type=str, default=None,
+            help="If the training should continue from a local checkpoint folder.",
+        )
+        parser.add_argument(
+            "--with_tracking", action="store_true",
+            help="Whether to enable experiment trackers for logging.",
+        )
+        parser.add_argument(
+            "--report_to", type=str, default="all",
+            help=(
+                'The integration to report the results and logs to. Supported platforms are `"tensorboard"`,'
+                ' `"wandb"`, `"comet_ml"` and `"clearml"`. Use `"all"` (default) to report to all integrations.'
+                "Only applicable when `--with_tracking` is passed."
+            ),
+        )
+    args = parser.parse_args()
+
+    return args
+
+
+    
+def main():
+    args = parse_args()
+    args.event_list = get_event_list()
+    print(args)
+    accelerator_log_kwargs = {}
+
+    if args.with_tracking:
+        accelerator_log_kwargs["log_with"] = args.report_to
+        accelerator_log_kwargs["logging_dir"] = args.output_dir
+
+    accelerator = Accelerator(gradient_accumulation_steps=args.gradient_accumulation_steps, **accelerator_log_kwargs)
+    
+    # Make one log on every process with the configuration for debugging.
+    logging.basicConfig(
+        format="%(asctime)s - %(levelname)s - %(name)s - %(message)s",
+        datefmt="%m/%d/%Y %H:%M:%S",
+        level=logging.INFO,
+    )
+    logger.info(accelerator.state, main_process_only=False)
+
+    datasets.utils.logging.set_verbosity_error()
+    diffusers.utils.logging.set_verbosity_error()
+    transformers.utils.logging.set_verbosity_error()
+
+    # If passed along, set the training seed now.
+    #if args.seed is not None:
+    set_seed(args.seed)
+    seed = args.seed
+    random.seed(seed)
+    np.random.seed(seed)
+    torch.manual_seed(seed)    
+    if torch.cuda.is_available():
+        torch.backends.cudnn.deterministic = True
+        torch.backends.cudnn.benchmark = False
+
+    # Handle output directory creation and wandb tracking
+    if accelerator.is_main_process:
+        if args.output_dir is None or args.output_dir == "":
+            args.output_dir = f"ckpts/{args.model_class}_{args.dataset_class}/base"        
+        elif args.output_dir is not None:
+            args.output_dir = f"ckpts/{args.model_class}_{args.dataset_class}/" + args.output_dir
+        os.makedirs(args.output_dir, exist_ok=True)
+
+        with open("{}/summary.jsonl".format(args.output_dir), "w") as f:
+            f.write(json.dumps(dict(vars(args))) + "\n\n")
+
+        accelerator.project_configuration.automatic_checkpoint_naming = False
+
+    accelerator.wait_for_everyone()
+
+
+    # Initialize models
+    pretrained_model_name = "audioldm-s-full"
+    vae, stft = ConDiffusion.build_pretrained_models(pretrained_model_name)
+    #vae, stft, clap, _ =  build_vae_stft_clap_models(pretrained_model_name)
+
+    model = getattr(ConDiffusion, args.model_class)(
+        scheduler_name=args.scheduler_name, unet_model_config_path=args.unet_model_config, 
+        snr_gamma=args.snr_gamma, uncondition=args.uncondition,
+    )
+
+    # Get the datasets
+    extension = args.train_file.split(".")[-1]
+    raw_datasets = load_dataset(extension, data_files={"train": args.train_file})
+    with accelerator.main_process_first():
+        train_dataset = getattr(ConDataset, args.dataset_class)(raw_datasets["train"], args)
+        accelerator.print("Num instances in train: {}".format(train_dataset.get_num_instances()))
+
+    train_dataloader = DataLoader(train_dataset, shuffle=True, batch_size=args.batch_size, collate_fn=train_dataset.collate_fn)
+
+    # Optimizer
+    optimizer_parameters = model.parameters()
+    if hasattr(model, "text_encoder"):
+        for param in model.text_encoder.parameters():
+            param.requires_grad = False
+            model.text_encoder.eval()
+            optimizer_parameters = model.unet.parameters()
+            accelerator.print("Optimizing UNet parameters.")
+    
+    num_trainable_parameters = sum(p.numel() for p in model.parameters() if p.requires_grad)
+    accelerator.print("Num trainable parameters: {}".format(num_trainable_parameters))
+
+    optimizer = torch.optim.AdamW(
+        optimizer_parameters, lr=args.learning_rate,
+        betas=(args.adam_beta1, args.adam_beta2),
+        weight_decay=args.adam_weight_decay,
+        eps=args.adam_epsilon,
+    )
+
+    # Scheduler and math around the number of training steps.
+    overrode_max_train_steps = False
+    num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
+    if args.max_train_steps is None:
+        args.max_train_steps = args.num_epochs * num_update_steps_per_epoch
+        overrode_max_train_steps = True
+
+    lr_scheduler = get_scheduler(
+        name=args.lr_scheduler_type,
+        optimizer=optimizer,
+        num_warmup_steps=args.num_warmup_steps * args.gradient_accumulation_steps,
+        num_training_steps=args.max_train_steps * args.gradient_accumulation_steps,
+    )
+
+    # Prepare everything with our `accelerator`.
+    vae, stft, model, optimizer, lr_scheduler = accelerator.prepare(
+        vae, stft, model, optimizer, lr_scheduler
+    )
+    train_dataloader = accelerator.prepare(
+        train_dataloader
+    )
+
+    # We need to recalculate our total training steps as the size of the training dataloader may have changed.
+    num_update_steps_per_epoch = math.ceil(len(train_dataloader) / args.gradient_accumulation_steps)
+    if overrode_max_train_steps:
+        args.max_train_steps = args.num_epochs * num_update_steps_per_epoch
+    # Afterwards we recalculate our number of training epochs
+    args.num_epochs = math.ceil(args.max_train_steps / num_update_steps_per_epoch)
+
+    # We need to initialize the trackers we use, and also store our configuration.
+    # The trackers initializes automatically on the main process.
+    if args.with_tracking:
+        experiment_config = vars(args)
+        # TensorBoard cannot log Enums, need the raw value
+        experiment_config["lr_scheduler_type"] = experiment_config["lr_scheduler_type"].value
+        accelerator.init_trackers("text_to_audio_diffusion", experiment_config)
+
+    # Train!
+    total_batch_size = args.batch_size * accelerator.num_processes * args.gradient_accumulation_steps
+
+    logger.info("***** Running training *****")
+    logger.info(f"  Num examples = {len(train_dataset)}")
+    logger.info(f"  Num Epochs = {args.num_epochs}")
+    logger.info(f"  Instantaneous batch size per device = {args.batch_size}")
+    logger.info(f"  Total train batch size (w. parallel, distributed & accumulation) = {total_batch_size}")
+    logger.info(f"  Gradient Accumulation steps = {args.gradient_accumulation_steps}")
+    logger.info(f"  Total optimization steps = {args.max_train_steps}")
+
+    # Only show the progress bar once on each machine.
+    progress_bar = tqdm(range(args.max_train_steps), disable=not accelerator.is_local_main_process)
+
+    completed_steps = 0
+    starting_epoch = 0
+    # Potentially load in the weights and states from a previous save
+    if args.resume_from_checkpoint:
+        if args.resume_from_checkpoint is not None or args.resume_from_checkpoint != "":
+            accelerator.load_state(args.resume_from_checkpoint)
+            # path = os.path.basename(args.resume_from_checkpoint)
+            accelerator.print(f"Resumed from local checkpoint: {args.resume_from_checkpoint}")
+        else:
+            # Get the most recent checkpoint
+            dirs = [f.name for f in os.scandir(os.getcwd()) if f.is_dir()]
+            dirs.sort(key=os.path.getctime)
+            # path = dirs[-1]  # Sorts folders by date modified, most recent checkpoint is the last
+            
+    # Duration of the audio clips in seconds
+    duration, best_loss, best_epoch = args.duration, np.inf, 0
+
+    for epoch in range(starting_epoch, args.num_epochs):
+        model.train()
+        total_loss = 0
+        logger.info(f"train epoch {epoch} begin!")
+        for step, batch in enumerate(train_dataloader):
+            with accelerator.accumulate(model):
+                device = model.device
+                
+                _, onset, event_info, audios, _, _ = batch #idx, onset, event_info, audios, caption, onset_str 
+                target_length = int(duration * 102.4)
+                with torch.no_grad():
+                    unwrapped_vae = accelerator.unwrap_model(vae)
+                    mel, _, waveform = torch_tools.wav_to_fbank(audios, target_length, stft)
+                    mel = mel.unsqueeze(1).to(device)    
+                    true_latent = unwrapped_vae.get_first_stage_encoding(unwrapped_vae.encode_first_stage(mel))
+
+                loss = model({"latent":true_latent, "onset":onset, "event_info":event_info}, validation_mode=False)
+                total_loss += loss.detach().float()
+                accelerator.backward(loss)
+                optimizer.step()
+                lr_scheduler.step()
+                optimizer.zero_grad()
+
+            # Checks if the accelerator has performed an optimization step behind the scenes
+            if accelerator.sync_gradients:
+                progress_bar.update(1)
+                completed_steps += 1
+
+            if completed_steps >= args.max_train_steps:
+                break
+        logger.info(f"train epoch {epoch} finish!")
+        model.uncondition = False
+
+        if accelerator.is_main_process:    
+            result = {}
+            result["epoch"] = epoch,
+            result["step"] = completed_steps
+            result["train_loss"] = round(total_loss.item()/len(train_dataloader), 4)
+
+            if result["train_loss"] < best_loss:
+                best_loss = result["train_loss"]
+                best_epoch = epoch
+                if args.checkpointing_steps == "best":
+                    accelerator.save(accelerator.unwrap_model(model).state_dict(), f"{args.output_dir}/best.pt")
+                    # Save all states -> continue training
+                    # accelerator.save_state("{}/{}".format(args.output_dir, "best"))
+            
+            result["best_eopch"] = best_epoch
+            logger.info(result)
+            result["time"] = datetime.now().strftime("%y-%m-%d-%H-%M-%S")
+
+            with open("{}/summary.jsonl".format(args.output_dir), "a") as f:
+                f.write(json.dumps(result) + "\n\n")
+
+        if args.with_tracking:
+            accelerator.log(result, step=completed_steps)
+
+  
+            
+if __name__ == "__main__":
+    main()
diff --git a/picoaudio/utils/__pycache__/torch_tools.cpython-310.pyc b/picoaudio/utils/__pycache__/torch_tools.cpython-310.pyc
new file mode 100644
index 0000000000000000000000000000000000000000..561ee820f054621285c7458a623c169d974b6f20
Binary files /dev/null and b/picoaudio/utils/__pycache__/torch_tools.cpython-310.pyc differ
diff --git a/picoaudio/utils/configs/frequency.json b/picoaudio/utils/configs/frequency.json
new file mode 100644
index 0000000000000000000000000000000000000000..c4c2cc2c899d7d24b6eadfc3da7537981704a1bb
--- /dev/null
+++ b/picoaudio/utils/configs/frequency.json
@@ -0,0 +1,46 @@
+{
+  "_class_name": "UNet2DConditionModel",
+  "_diffusers_version": "0.10.0.dev0",
+  "act_fn": "silu",
+  "attention_head_dim": [
+    4,
+    8,
+    16,
+    16
+  ],
+  "block_out_channels": [
+    128,
+    256,
+    512,
+    512
+  ],
+  "center_input_sample": false,
+  "cross_attention_dim": 1024,
+  "down_block_types": [
+    "CrossAttnDownBlock2D",
+    "CrossAttnDownBlock2D",
+    "CrossAttnDownBlock2D",
+    "DownBlock2D"
+  ],
+  "downsample_padding": 1,
+  "dual_cross_attention": false,
+  "flip_sin_to_cos": true,
+  "freq_shift": 0,
+  "in_channels": 10,
+  "layers_per_block": 2,
+  "mid_block_scale_factor": 1,
+  "norm_eps": 1e-05,
+  "norm_num_groups": 32,
+  "num_class_embeds": null,
+  "only_cross_attention": false,
+  "out_channels": 8,
+  "sample_size": [32, 2],
+  "up_block_types": [
+    "UpBlock2D",
+    "CrossAttnUpBlock2D",
+    "CrossAttnUpBlock2D",
+    "CrossAttnUpBlock2D"
+  ],
+  "use_linear_projection": true,
+  "upcast_attention": true
+}
diff --git a/picoaudio/utils/torch_tools.py b/picoaudio/utils/torch_tools.py
new file mode 100644
index 0000000000000000000000000000000000000000..50237e48f616780b1f1f5418461129434d03fd86
--- /dev/null
+++ b/picoaudio/utils/torch_tools.py
@@ -0,0 +1,128 @@
+import torch
+import torchaudio
+import random
+import itertools
+import numpy as np
+#from utils.mix import mix
+
+def normalize_wav(waveform):
+    waveform = waveform - torch.mean(waveform)
+    waveform = waveform / (torch.max(torch.abs(waveform)) + 1e-8)
+    return waveform * 0.5
+
+
+def pad_wav(waveform, segment_length):
+    waveform_length = len(waveform)
+    
+    if segment_length is None or waveform_length == segment_length:
+        return waveform
+    elif waveform_length > segment_length:
+        return waveform[:segment_length]
+    else:
+        pad_wav = torch.zeros(segment_length - waveform_length).to(waveform.device)
+        waveform = torch.cat([waveform, pad_wav])
+        return waveform
+    
+    
+def _pad_spec(fbank, target_length=1024):
+    batch, n_frames, channels = fbank.shape
+    p = target_length - n_frames
+    if p > 0:
+        pad = torch.zeros(batch, p, channels).to(fbank.device)
+        fbank = torch.cat([fbank, pad], 1)
+    elif p < 0:
+        fbank = fbank[:, :target_length, :]
+
+    if channels % 2 != 0:
+        fbank = fbank[:, :, :-1]
+
+    return fbank
+
+
+def read_wav_file(filename, segment_length):
+    
+    waveform, sr = torchaudio.load(filename)  # Faster!!!
+    waveform = torchaudio.functional.resample(waveform, orig_freq=sr, new_freq=16000)[0]
+    try:
+        waveform = normalize_wav(waveform)
+    except:
+        print ("Exception normalizing:", filename)
+        waveform = torch.ones(160000)
+    waveform = pad_wav(waveform, segment_length).unsqueeze(0)
+    waveform = waveform / torch.max(torch.abs(waveform))
+    waveform = 0.5 * waveform
+    return waveform
+
+
+def get_mel_from_wav(audio, _stft):
+    audio = torch.nan_to_num(torch.clip(audio, -1, 1))
+    audio = torch.autograd.Variable(audio, requires_grad=False)
+    melspec, log_magnitudes_stft, energy = _stft.mel_spectrogram(audio)
+    return melspec, log_magnitudes_stft, energy
+
+
+def wav_to_fbank(paths, target_length=1024, fn_STFT=None):
+    assert fn_STFT is not None
+
+    waveform = torch.cat([read_wav_file(path, target_length * 160) for path in paths], 0)  # hop size is 160
+
+    fbank, log_magnitudes_stft, energy = get_mel_from_wav(waveform, fn_STFT)
+    fbank = fbank.transpose(1, 2)
+    log_magnitudes_stft = log_magnitudes_stft.transpose(1, 2)
+
+    fbank, log_magnitudes_stft = _pad_spec(fbank, target_length), _pad_spec(
+        log_magnitudes_stft, target_length
+    )
+
+    return fbank, log_magnitudes_stft, waveform
+
+
+def uncapitalize(s):
+    if s:
+        return s[:1].lower() + s[1:]
+    else:
+        return ""
+
+    
+def mix_wavs_and_captions(path1, path2, caption1, caption2, target_length=1024):
+    sound1 = read_wav_file(path1, target_length * 160)[0].numpy()
+    sound2 = read_wav_file(path2, target_length * 160)[0].numpy()
+    mixed_sound = mix(sound1, sound2, 0.5, 16000).reshape(1, -1)
+    mixed_caption = "{} and {}".format(caption1, uncapitalize(caption2))
+    return mixed_sound, mixed_caption
+
+
+def augment(paths, texts, num_items=4, target_length=1024):
+    mixed_sounds, mixed_captions = [], []
+    combinations = list(itertools.combinations(list(range(len(texts))), 2))
+    random.shuffle(combinations)
+    if len(combinations) < num_items:
+        selected_combinations = combinations
+    else:
+        selected_combinations = combinations[:num_items]
+        
+    for (i, j) in selected_combinations:
+        new_sound, new_caption = mix_wavs_and_captions(paths[i], paths[j], texts[i], texts[j], target_length)
+        mixed_sounds.append(new_sound)
+        mixed_captions.append(new_caption)
+        
+    waveform = torch.tensor(np.concatenate(mixed_sounds, 0))
+    waveform = waveform / torch.max(torch.abs(waveform))
+    waveform = 0.5 * waveform
+    
+    return waveform, mixed_captions
+
+
+def augment_wav_to_fbank(paths, texts, num_items=4, target_length=1024, fn_STFT=None):
+    assert fn_STFT is not None
+    
+    waveform, captions = augment(paths, texts)
+    fbank, log_magnitudes_stft, energy = get_mel_from_wav(waveform, fn_STFT)
+    fbank = fbank.transpose(1, 2)
+    log_magnitudes_stft = log_magnitudes_stft.transpose(1, 2)
+
+    fbank, log_magnitudes_stft = _pad_spec(fbank, target_length), _pad_spec(
+        log_magnitudes_stft, target_length
+    )
+
+    return fbank, log_magnitudes_stft, waveform, captions
\ No newline at end of file