jonatasgrosman/exp_w2v2r_fr_xls-r_age_teens-5_sixties-5_s463

.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_s463
+
+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": 0.07542461836618379,
+      "wer": 0.2639889072773628
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07356022871385301,
+      "wer": 0.2593794362198337
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.08656056904727691,
+      "wer": 0.29078089947655167
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08439558726074924,
+      "wer": 0.29691486893992114
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08340447548829388,
+      "wer": 0.2795748613678373
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08724163692205528,
+      "wer": 0.27167919799498746
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06890768048081627,
+      "wer": 0.23972602739726026
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07940565585556798,
+      "wer": 0.29022988505747127
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08856855092534308,
+      "wer": 0.3042201834862385
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08,
+      "wer": 0.3103448275862069
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07355286248542976,
+      "wer": 0.25934069361277445
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.0866067203028869,
+      "wer": 0.28034682080924855
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08224127744027149,
+      "wer": 0.2820755141532784
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09075772623707426,
+      "wer": 0.29721634954193094
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07444080056472419,
+      "wer": 0.2616144925108321
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.18787878787878787,
+      "wer": 0.4819277108433735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1228510975932293,
+      "wer": 0.36273006134969327
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.07538280329799764,
+      "wer": 0.2708333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.16631578947368422,
+      "wer": 0.4936708860759494
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.08880586291133784,
+      "wer": 0.2773109243697479
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.10526315789473684,
+      "wer": 0.35
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.13178294573643412,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.02459016393442623,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.08053691275167785,
+      "wer": 0.3018867924528302
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.10766045548654245,
+      "wer": 0.28615384615384615
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.046822742474916385,
+      "wer": 0.21153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.06854838709677419,
+      "wer": 0.18421052631578946
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07119590507212657,
+      "wer": 0.24376731301939059
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.1258741258741259,
+      "wer": 0.36
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.04814814814814815,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12177121771217712,
+      "wer": 0.3695652173913043
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.008583690987124463,
+      "wer": 0.05263157894736842
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.0751711353112536,
+      "wer": 0.26650954740031746
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.20905923344947736,
+      "wer": 0.625
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.22813688212927757,
+      "wer": 0.6060606060606061
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.17905405405405406,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.0944206008583691,
+      "wer": 0.312
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08593623342197405,
+      "wer": 0.28265126979801114
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08475693076769711,
+      "wer": 0.2792210170137032
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10061436889642841,
+      "wer": 0.32360996987126817
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08748734105997524,
+      "wer": 0.28348806366047746
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09939935871381475,
+      "wer": 0.31624863685932386
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07979502196193265,
+      "wer": 0.27549668874172184
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08233691539276525,
+      "wer": 0.27206551410373064
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0855207451312447,
+      "wer": 0.2966101694915254
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09617396502764908,
+      "wer": 0.3165532879818594
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08464382357084906,
+      "wer": 0.2790757230974105
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.17361451692005886,
+      "wer": 0.4441176470588235
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09275391817626705,
+      "wer": 0.30171101503988307
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09414409265820856,
+      "wer": 0.3104161876293401
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08544580833222919,
+      "wer": 0.2810438425259678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.009345794392523364,
+      "wer": 0.1
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.13111111111111112,
+      "wer": 0.38961038961038963
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.13921345112567682,
+      "wer": 0.4076271186440678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.07210031347962383,
+      "wer": 0.18518518518518517
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.007905138339920948,
+      "wer": 0.047619047619047616
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.11363636363636363,
+      "wer": 0.42105263157894735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07787367595135347,
+      "wer": 0.27505827505827507
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.12213740458015267,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.08814589665653495,
+      "wer": 0.22641509433962265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.1105121293800539,
+      "wer": 0.3643724696356275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.21814134668892599,
+      "wer": 0.5544871794871795
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.07703081232492998,
+      "wer": 0.28448275862068967
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.06578947368421052,
+      "wer": 0.2894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.14285714285714285,
+      "wer": 0.5
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08215818516247701,
+      "wer": 0.3102189781021898
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.08308157099697885,
+      "wer": 0.26548672566371684
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.37142857142857144,
+      "wer": 0.7777777777777778
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.08252295723676112,
+      "wer": 0.27773657668347673
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.1264236902050114,
+      "wer": 0.4125874125874126
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.11808118081180811,
+      "wer": 0.36585365853658536
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.14862385321100918,
+      "wer": 0.44680851063829785
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.09605026929982047,
+      "wer": 0.3010752688172043
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60b4fef321a80cbe75929780698eaebb35b312ffeb1bb9a31df16c3fe0837f60
+size 130479595

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 sha256:8ae52332109cb2defd7630a92f7cb56ceb728f28ded6f15986e9f9bb010b9a27
+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/tmpa607uza4", "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}