jonatasgrosman/exp_w2v2r_fr_xls-r_age_teens-2_sixties-8_s598

.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-2_sixties-8_s598
+
+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": 0.07582793019309085,
+      "wer": 0.26568859864919264
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07385368363491106,
+      "wer": 0.26045582890036345
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07429301805400224,
+      "wer": 0.28065134099616856
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.09714285714285714,
+      "wer": 0.2413793103448276
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08907213899030593,
+      "wer": 0.3005504587155963
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.09002730277338698,
+      "wer": 0.3014635188548232
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08552580519984478,
+      "wer": 0.29220579174368455
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08201211919980078,
+      "wer": 0.2731829573934837
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08162787978014267,
+      "wer": 0.28462073764787754
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07121392130826752,
+      "wer": 0.24743150684931506
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08297604799490962,
+      "wer": 0.2836956032581792
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07383845197573652,
+      "wer": 0.2604244635728543
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.0866067203028869,
+      "wer": 0.2774566473988439
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09256879126015073,
+      "wer": 0.3134249471458774
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07462011817282935,
+      "wer": 0.26265010698811325
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.07407407407407407,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.07258064516129033,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.09473684210526316,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.008583690987124463,
+      "wer": 0.07894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.10610766045548654,
+      "wer": 0.31384615384615383
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09354792355223451,
+      "wer": 0.2957983193277311
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.11301859799713877,
+      "wer": 0.4
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07214765100671142,
+      "wer": 0.29245283018867924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.15891472868217055,
+      "wer": 0.6
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.01639344262295082,
+      "wer": 0.1111111111111111
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07634280027927356,
+      "wer": 0.26838535904958877
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.1258741258741259,
+      "wer": 0.44
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.12721502248082517,
+      "wer": 0.379601226993865
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.1951219512195122,
+      "wer": 0.6875
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07538389948813401,
+      "wer": 0.2548476454293629
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.23067173637515842,
+      "wer": 0.5909090909090909
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.18585858585858586,
+      "wer": 0.5180722891566265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.056856187290969896,
+      "wer": 0.19230769230769232
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.17567567567567569,
+      "wer": 0.4583333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.08362779740871613,
+      "wer": 0.3125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.15157894736842106,
+      "wer": 0.4177215189873418
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.14022140221402213,
+      "wer": 0.45652173913043476
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08631732536619567,
+      "wer": 0.284394191949601
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08509732981263464,
+      "wer": 0.2815573800508287
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07705334462320068,
+      "wer": 0.25084745762711863
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09714541921984755,
+      "wer": 0.31700680272108844
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10313603227729128,
+      "wer": 0.32073404546699535
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09795420674705324,
+      "wer": 0.31515812431842966
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.0843338213762811,
+      "wer": 0.2847682119205298
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08478676718802745,
+      "wer": 0.27785145888594165
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.0828666565763584,
+      "wer": 0.2734303912647862
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09358843883574192,
+      "wer": 0.2997625281617244
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09502038328189888,
+      "wer": 0.31133593929639
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.14565963707699853,
+      "wer": 0.38823529411764707
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08502518958992222,
+      "wer": 0.281431780433017
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.0859485907894986,
+      "wer": 0.2834233361355314
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.03508771929824561,
+      "wer": 0.18421052631578946
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.22760155815247635,
+      "wer": 0.5641025641025641
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07277363672028246,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.07899461400359066,
+      "wer": 0.27419354838709675
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.11590296495956873,
+      "wer": 0.340080971659919
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.13740458015267176,
+      "wer": 0.3181818181818182
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.07598784194528875,
+      "wer": 0.22641509433962265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.08184593416873877,
+      "wer": 0.27566004153070306
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.12084592145015106,
+      "wer": 0.3805309734513274
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.09523809523809523,
+      "wer": 0.2857142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.09090909090909091,
+      "wer": 0.2894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.1070110701107011,
+      "wer": 0.3902439024390244
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1409233399829011,
+      "wer": 0.40084745762711865
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07357449417535254,
+      "wer": 0.2773722627737226
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.13553530751708429,
+      "wer": 0.45454545454545453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.08150470219435736,
+      "wer": 0.2037037037037037
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.14666666666666667,
+      "wer": 0.42857142857142855
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.07703081232492998,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.13577981651376148,
+      "wer": 0.32978723404255317
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.011857707509881422,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.028037383177570093,
+      "wer": 0.1
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.2571428571428571,
+      "wer": 0.4444444444444444
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ddc953604ed845a34284a55faf49908261af6dd35daacc70308416d8da3e9932
+size 130462732

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 sha256:5da1e53bbca80ef99740de46aa116f567505dbbd87ae41f1b94b798f9c34171e
+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/tmps4qdlhbu", "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}