jonatasgrosman/exp_w2v2r_de_vp-100k_age_teens-10_sixties-0_s152

.gitattributes

@@ -32,3 +32,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+cv7_transcriptions.json filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,17 @@
+---
+language:
+- de
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- de
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_de_vp-100k_age_teens-10_sixties-0_s152
+
+Fine-tuned [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) for speech recognition using the train split of [Common Voice 7.0 (de)](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0).
+When using this model, make sure that your speech input is sampled at 16kHz.
+
+This model has been fine-tuned by the [HuggingSound](https://github.com/jonatasgrosman/huggingsound) tool.
+    

config.json

@@ -0,0 +1,116 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-100k-voxpopuli",
+  "activation_dropout": 0.05,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.05,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "conv_dim": [
+    512,
+    512,
+    512,
+    512,
+    512,
+    512,
+    512
+  ],
+  "conv_kernel": [
+    10,
+    3,
+    3,
+    3,
+    3,
+    2,
+    2
+  ],
+  "conv_stride": [
+    5,
+    2,
+    2,
+    2,
+    2,
+    2,
+    2
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.05,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.05,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.05,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.05,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "mask_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.18.0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 38,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,396 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.06487153280219721,
+      "wer": 0.24817875394150266
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.0637396492966452,
+      "wer": 0.2452647967296917
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0638350110483673,
+      "wer": 0.2585034013605442
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.06199021207177814,
+      "wer": 0.2804878048780488
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06477093206951026,
+      "wer": 0.2459992556754745
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.06721607004893124,
+      "wer": 0.2631050915475295
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.06167586552935273,
+      "wer": 0.23396381578947367
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07128439915325162,
+      "wer": 0.2724207687120701
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.0943609022556391,
+      "wer": 0.30664780763790667
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.0755891507336594,
+      "wer": 0.276006272869838
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.06373061854703137,
+      "wer": 0.24522521022396138
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07011466490389505,
+      "wer": 0.26120764552562986
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.09990662931839403,
+      "wer": 0.38414634146341464
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.06812194203989462,
+      "wer": 0.25660964230171074
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.06447943824036509,
+      "wer": 0.2472056377413678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.07744107744107744,
+      "wer": 0.2391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.05277777777777778,
+      "wer": 0.24561403508771928
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1213235294117647,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06457479222167276,
+      "wer": 0.2444091208068406
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.0806089343648615,
+      "wer": 0.3082901554404145
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.05136986301369863,
+      "wer": 0.2
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.06574292452830188,
+      "wer": 0.25279187817258886
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.171990171990172,
+      "wer": 0.4915254237288136
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.08742514970059881,
+      "wer": 0.28
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.10238907849829351,
+      "wer": 0.33689839572192515
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.11259079903147699,
+      "wer": 0.36363636363636365
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.10877561744613767,
+      "wer": 0.42402826855123676
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.03783783783783784,
+      "wer": 0.21153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.05067567567567568,
+      "wer": 0.2708333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10362694300518134,
+      "wer": 0.42735042735042733
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.02926829268292683,
+      "wer": 0.12121212121212122
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.07512998726389941,
+      "wer": 0.2703783070963488
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07418676969888952,
+      "wer": 0.2683784923159468
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.11905342766528422,
+      "wer": 0.3756345177664975
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.13761467889908258,
+      "wer": 0.4657534246575342
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08592200925313946,
+      "wer": 0.30360721442885774
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.07398924833794573,
+      "wer": 0.27651760639446965
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.07426849865339846,
+      "wer": 0.2524330900243309
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.06798135532448907,
+      "wer": 0.2555831265508685
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.10963590453693899,
+      "wer": 0.3379501385041551
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.11698406529343179,
+      "wer": 0.3474114441416894
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07430083380349532,
+      "wer": 0.26859187520112787
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07867787273112507,
+      "wer": 0.2752835916280556
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.12792642140468227,
+      "wer": 0.3125
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09808076527214385,
+      "wer": 0.3391959798994975
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07525344021730715,
+      "wer": 0.27084631399784204
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.07791225416036308,
+      "wer": 0.32642487046632124
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.1346153846153846,
+      "wer": 0.25925925925925924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.1827956989247312,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.2255892255892256,
+      "wer": 0.6888888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.056338028169014086,
+      "wer": 0.23684210526315788
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.06771463119709795,
+      "wer": 0.23651452282157676
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09644670050761421,
+      "wer": 0.47619047619047616
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.1505016722408027,
+      "wer": 0.4673913043478261
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06866418631747587,
+      "wer": 0.2505859022895259
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.10281280310378274,
+      "wer": 0.37333333333333335
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.07352941176470588,
+      "wer": 0.2222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.09985792571544551,
+      "wer": 0.3487394957983193
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.23306233062330622,
+      "wer": 0.5849056603773585
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.04292929292929293,
+      "wer": 0.2413793103448276
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.13333333333333333,
+      "wer": 0.6428571428571429
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.08144857296715131,
+      "wer": 0.2742382271468144
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.03260869565217391,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12852664576802508,
+      "wer": 0.4411764705882353
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35b98fa511290b84e3af5c34354a4224229da8670571fcb3d1a549de125370e9
+size 138180739

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:311c4309f03d5adb79e5a64105abd0fe1b3260e7124873a28ba74f461cf4c83e
+size 1262054513

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "<unk>", "pad_token": "<pad>"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "<unk>", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "<pad>", "do_lower_case": false, "word_delimiter_token": "|", "replace_word_delimiter_char": " ", "special_tokens_map_file": null, "name_or_path": "/tmp/tmpw4ze59az", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2Processor"}

vocab.json

@@ -0,0 +1 @@
+{"<pad>": 0, "<s>": 1, "</s>": 2, "<unk>": 3, "|": 4, "'": 5, "-": 6, "a": 7, "b": 8, "c": 9, "d": 10, "e": 11, "f": 12, "g": 13, "h": 14, "i": 15, "j": 16, "k": 17, "l": 18, "m": 19, "n": 20, "o": 21, "p": 22, "q": 23, "r": 24, "s": 25, "t": 26, "u": 27, "v": 28, "w": 29, "x": 30, "y": 31, "z": 32, "ß": 33, "á": 34, "ä": 35, "ö": 36, "ü": 37}