jonatasgrosman/exp_w2v2r_de_vp-100k_age_teens-10_sixties-0_s693

.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_s693
+
+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": 0.0648066081492244,
+      "wer": 0.2477710122866152
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.06373363668941036,
+      "wer": 0.24494842270899417
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0643260495948932,
+      "wer": 0.2687074829931973
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.0636215334420881,
+      "wer": 0.25609756097560976
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06335457863485319,
+      "wer": 0.23930033494603647
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.06806224936536551,
+      "wer": 0.26285427639829445
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.060210737581535376,
+      "wer": 0.22807017543859648
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07187515384236695,
+      "wer": 0.2670262980445044
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09302422723475355,
+      "wer": 0.3123055162659123
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07625611382836817,
+      "wer": 0.2932566649242028
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.06372099846800242,
+      "wer": 0.2449088335692282
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07003203498423675,
+      "wer": 0.26064291920069504
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.09897292250233426,
+      "wer": 0.36585365853658536
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.06646593902898006,
+      "wer": 0.25505443234836706
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.06443412754843597,
+      "wer": 0.24701278455888853
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.08754208754208755,
+      "wer": 0.2391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.06666666666666667,
+      "wer": 0.2982456140350877
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1213235294117647,
+      "wer": 0.4
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06437538883660089,
+      "wer": 0.2443543082657312
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08123284252558023,
+      "wer": 0.31088082901554404
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.046232876712328765,
+      "wer": 0.16363636363636364
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.06574292452830188,
+      "wer": 0.2451776649746193
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.15233415233415235,
+      "wer": 0.4406779661016949
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.08982035928143713,
+      "wer": 0.344
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.10921501706484642,
+      "wer": 0.32085561497326204
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.1089588377723971,
+      "wer": 0.35537190082644626
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.10194429847609038,
+      "wer": 0.3674911660777385
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.051351351351351354,
+      "wer": 0.23076923076923078
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.037162162162162164,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10233160621761658,
+      "wer": 0.38461538461538464
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.05365853658536585,
+      "wer": 0.15151515151515152
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.07428858423184706,
+      "wer": 0.2677546086430946
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07346511285508417,
+      "wer": 0.2658030889510597
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.11344230300073188,
+      "wer": 0.3604060913705584
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.10091743119266056,
+      "wer": 0.4246575342465753
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08533450833516927,
+      "wer": 0.3001002004008016
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.07325762636383878,
+      "wer": 0.2734931950745301
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.06947551011092346,
+      "wer": 0.2451338199513382
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.06776622445320904,
+      "wer": 0.2575682382133995
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.10825120143357497,
+      "wer": 0.33518005540166207
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.11795569374271278,
+      "wer": 0.3637602179836512
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07359726332268128,
+      "wer": 0.2660404247820157
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07664147656378746,
+      "wer": 0.2728071576929222
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.12374581939799331,
+      "wer": 0.2784090909090909
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09614336743960768,
+      "wer": 0.3341708542713568
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07448587525631582,
+      "wer": 0.26813367171709496
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.0794251134644478,
+      "wer": 0.31088082901554404
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.1282051282051282,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.16129032258064516,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.2255892255892256,
+      "wer": 0.6222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.0528169014084507,
+      "wer": 0.2631578947368421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.07194679564691657,
+      "wer": 0.24896265560165975
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.08121827411167512,
+      "wer": 0.38095238095238093
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.11371237458193979,
+      "wer": 0.3804347826086957
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06715479232592318,
+      "wer": 0.24995493059311338
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.09408341416100872,
+      "wer": 0.32666666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.07352941176470588,
+      "wer": 0.2222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.09742236655165415,
+      "wer": 0.34173669467787116
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.2222222222222222,
+      "wer": 0.6226415094339622
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.030303030303030304,
+      "wer": 0.1896551724137931
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.1282051282051282,
+      "wer": 0.5
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.08360258481421648,
+      "wer": 0.28162511542012925
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.03260869565217391,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.13166144200626959,
+      "wer": 0.45098039215686275
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2fc4c81985e09e5e4ac5d362be273eddf4ce2c84b54b9cca2bce9459b3c97e55
+size 138292962

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 sha256:0aefe2b8ad336176d9afb989af47bc7ec82b1a1c06f604c79619242fcc76f7d8
+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/tmp65b2kf7h", "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}