jonatasgrosman/exp_w2v2r_fr_xls-r_age_teens-5_sixties-5_s240

.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:
+- fr
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- fr
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_fr_xls-r_age_teens-5_sixties-5_s240
+
+Fine-tuned [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) for speech recognition using the train split of [Common Voice 7.0 (fr)](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 +1,108 @@
+{
+  "_name_or_path": "facebook/wav2vec2-xls-r-300m",
+  "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_feature_length": 10,
+  "mask_feature_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.05,
+  "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": 57,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,444 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.07439935724830954,
+      "wer": 0.26145853967118
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07242260611239452,
+      "wer": 0.25617365801912545
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.08663241845092685,
+      "wer": 0.2932379019335541
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.07944489923205862,
+      "wer": 0.27951751333797265
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08511188720734704,
+      "wer": 0.28758471965495996
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08641155474391965,
+      "wer": 0.28421052631578947
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06869802222377525,
+      "wer": 0.2495719178082192
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07748841667998083,
+      "wer": 0.29118773946360155
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.088379705400982,
+      "wer": 0.29211009174311925
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.07428571428571429,
+      "wer": 0.27586206896551724
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.0724273039059855,
+      "wer": 0.25616735279441116
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.08707998106956934,
+      "wer": 0.2774566473988439
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08201402881512566,
+      "wer": 0.2816254894019171
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.08882981830928317,
+      "wer": 0.301092318534179
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07322724970987168,
+      "wer": 0.2586827897626463
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.1919191919191919,
+      "wer": 0.46987951807228917
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1236445384818831,
+      "wer": 0.3811349693251534
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.08716136631330977,
+      "wer": 0.3055555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.18105263157894738,
+      "wer": 0.4810126582278481
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09081764621353643,
+      "wer": 0.27983193277310925
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.1,
+      "wer": 0.36666666666666664
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.15503875968992248,
+      "wer": 0.5333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.02459016393442623,
+      "wer": 0.15873015873015872
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.05704697986577181,
+      "wer": 0.27358490566037735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.11024844720496894,
+      "wer": 0.31384615384615383
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.04013377926421405,
+      "wer": 0.1346153846153846
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.08870967741935484,
+      "wer": 0.2894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07584923220102373,
+      "wer": 0.24376731301939059
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.0979020979020979,
+      "wer": 0.3
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.05925925925925926,
+      "wer": 0.22916666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.11070110701107011,
+      "wer": 0.43478260869565216
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.012875536480686695,
+      "wer": 0.10526315789473684
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07481803080034348,
+      "wer": 0.26371987879370884
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.16376306620209058,
+      "wer": 0.5625
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.2370088719898606,
+      "wer": 0.6060606060606061
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.16891891891891891,
+      "wer": 0.4375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.10586552217453506,
+      "wer": 0.336
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08510983238270896,
+      "wer": 0.2810414435198154
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08399589576008676,
+      "wer": 0.27813706174882996
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10109577736004768,
+      "wer": 0.32141878937277457
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08489929109935862,
+      "wer": 0.2738726790450928
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09479293681976245,
+      "wer": 0.30725190839694655
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08184480234260615,
+      "wer": 0.27593818984547464
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08188285152111396,
+      "wer": 0.26524112829845314
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07959356477561388,
+      "wer": 0.27288135593220336
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09370796592437602,
+      "wer": 0.3165532879818594
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08391266961071517,
+      "wer": 0.27805135034279904
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.16331535066208927,
+      "wer": 0.42058823529411765
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09158355383675962,
+      "wer": 0.29720513913414115
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09235341181849355,
+      "wer": 0.3030581742929409
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08489452215555261,
+      "wer": 0.28025305273356793
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.018691588785046728,
+      "wer": 0.15
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.14666666666666667,
+      "wer": 0.4025974025974026
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.13451125676831005,
+      "wer": 0.39915254237288134
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.05642633228840126,
+      "wer": 0.18518518518518517
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.019762845849802372,
+      "wer": 0.09523809523809523
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.10454545454545454,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.0690466849744998,
+      "wer": 0.25990675990675993
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.05343511450381679,
+      "wer": 0.18181818181818182
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.08814589665653495,
+      "wer": 0.24528301886792453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.11994609164420485,
+      "wer": 0.3643724696356275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.22426265998887035,
+      "wer": 0.5801282051282052
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.07282913165266107,
+      "wer": 0.29310344827586204
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.07456140350877193,
+      "wer": 0.2631578947368421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.13095238095238096,
+      "wer": 0.42857142857142855
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.06805640711220111,
+      "wer": 0.25547445255474455
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.1042296072507553,
+      "wer": 0.3185840707964602
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.34285714285714286,
+      "wer": 0.8888888888888888
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07960560328910116,
+      "wer": 0.27009789380005933
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.13553530751708429,
+      "wer": 0.3916083916083916
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.13284132841328414,
+      "wer": 0.3902439024390244
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.13761467889908258,
+      "wer": 0.3723404255319149
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.07989228007181329,
+      "wer": 0.26881720430107525
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b7485739aed2b54106168a6aa4ebbf96e28a0ca26c0369ef117716a37ca61549
+size 130218781

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 sha256:9dbedfca441e96b11694297f80bad56a87d2d0b21d1d5dcf09dd0e222da7db5b
+size 1262132401

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/tmp80f9jrai", "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, "à": 33, "á": 34, "â": 35, "ä": 36, "ç": 37, "è": 38, "é": 39, "ê": 40, "ë": 41, "í": 42, "î": 43, "ï": 44, "ó": 45, "ô": 46, "ö": 47, "ù": 48, "û": 49, "ü": 50, "ć": 51, "č": 52, "ł": 53, "ō": 54, "œ": 55, "š": 56}