jonatasgrosman/exp_w2v2r_fr_xls-r_age_teens-2_sixties-8_s82

.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_s82
+
+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": 0.07615800908300688,
+      "wer": 0.26539466705857545
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07409672119067719,
+      "wer": 0.2600736314993058
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07844703626777441,
+      "wer": 0.29310344827586204
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.045714285714285714,
+      "wer": 0.1724137931034483
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.0888203449578245,
+      "wer": 0.301651376146789
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.09112300617904871,
+      "wer": 0.3001816045294306
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08521536670547147,
+      "wer": 0.29066543438077636
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08300821781356355,
+      "wer": 0.2621553884711779
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08392780571473123,
+      "wer": 0.29343539781953143
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.0705150604514641,
+      "wer": 0.24529109589041095
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08373354341206236,
+      "wer": 0.2844606453354935
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07409819626330058,
+      "wer": 0.26007360279441116
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.09891150023663038,
+      "wer": 0.29190751445086704
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09210141964129229,
+      "wer": 0.3093727977448908
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.0749231901865282,
+      "wer": 0.26235312930712823
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.05925925925925926,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.0967741935483871,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.11315789473684211,
+      "wer": 0.35
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.02145922746781116,
+      "wer": 0.10526315789473684
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.12060041407867494,
+      "wer": 0.3292307692307692
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09196723667193563,
+      "wer": 0.2865546218487395
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.10014306151645208,
+      "wer": 0.352
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07550335570469799,
+      "wer": 0.3113207547169811
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.18604651162790697,
+      "wer": 0.6222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.02459016393442623,
+      "wer": 0.15873015873015872
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07656750314985274,
+      "wer": 0.26766389303063826
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.13286713286713286,
+      "wer": 0.46
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1248346998148638,
+      "wer": 0.37883435582822084
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.1602787456445993,
+      "wer": 0.5833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08608655188459749,
+      "wer": 0.29362880886426596
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.2370088719898606,
+      "wer": 0.6363636363636364
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.18787878787878787,
+      "wer": 0.43373493975903615
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06020066889632107,
+      "wer": 0.23076923076923078
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.17905405405405406,
+      "wer": 0.3958333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.0800942285041225,
+      "wer": 0.3125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.17894736842105263,
+      "wer": 0.5189873417721519
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12546125461254612,
+      "wer": 0.34782608695652173
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08705319075020371,
+      "wer": 0.2874237075440009
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08590091470086218,
+      "wer": 0.28456755311747
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07705334462320068,
+      "wer": 0.2627118644067797
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.10080705425198028,
+      "wer": 0.3247165532879819
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10343404704048416,
+      "wer": 0.3292248698986579
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09646389378133044,
+      "wer": 0.3105234460196292
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08052708638360176,
+      "wer": 0.2781456953642384
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08726229323731292,
+      "wer": 0.27519893899204245
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08415317087937037,
+      "wer": 0.2829845313921747
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09348666802361083,
+      "wer": 0.3018936856847105
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09643006819827028,
+      "wer": 0.31685444929868933
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.14958312898479648,
+      "wer": 0.4117647058823529
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08583164512070113,
+      "wer": 0.2844390461626265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08671991822982718,
+      "wer": 0.2866363649316786
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.05701754385964912,
+      "wer": 0.21052631578947367
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.2392877017250974,
+      "wer": 0.6217948717948718
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.0745390349156532,
+      "wer": 0.27855477855477856
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.09156193895870736,
+      "wer": 0.2956989247311828
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.10714285714285714,
+      "wer": 0.3562753036437247
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.11450381679389313,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.08206686930091185,
+      "wer": 0.22641509433962265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.08207981486496467,
+      "wer": 0.27566004153070306
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.11178247734138973,
+      "wer": 0.3274336283185841
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.10714285714285714,
+      "wer": 0.35714285714285715
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.10454545454545454,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.1033210332103321,
+      "wer": 0.34146341463414637
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.13878597891137076,
+      "wer": 0.39322033898305087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.0815450643776824,
+      "wer": 0.28102189781021897
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.1264236902050114,
+      "wer": 0.3706293706293706
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.06269592476489028,
+      "wer": 0.2222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.15777777777777777,
+      "wer": 0.4805194805194805
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.08823529411764706,
+      "wer": 0.3103448275862069
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.13761467889908258,
+      "wer": 0.425531914893617
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.011857707509881422,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.009345794392523364,
+      "wer": 0.1
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.2571428571428571,
+      "wer": 0.5555555555555556
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9d32d30680e5e84fb118dcfde534ab49abd7ed9cbcbcf30c955a2ccd8d11de8f
+size 130374270

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 sha256:12fb07ae302a4a861bb1dadd61d16a601e16160cb2414cbfc301663ef9a84452
+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/tmpyv76lr0y", "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}