jonatasgrosman/exp_w2v2r_fr_xls-r_age_teens-8_sixties-2_s187

.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-8_sixties-2_s187
+
+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": 0.07668019176416023,
+      "wer": 0.2673691205694605
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07477153823381506,
+      "wer": 0.2626554139431852
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.08668630550366432,
+      "wer": 0.2927037709646405
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08689042217284527,
+      "wer": 0.29343539781953143
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08542232570172034,
+      "wer": 0.2900492914356131
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.09147505603054702,
+      "wer": 0.2862155388471178
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07198266825075128,
+      "wer": 0.2508561643835616
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07684933695478512,
+      "wer": 0.2796934865900383
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.08982752108775022,
+      "wer": 0.30275229357798167
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.06857142857142857,
+      "wer": 0.2413793103448276
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07478438639611909,
+      "wer": 0.26265437874251496
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.07998106956933271,
+      "wer": 0.2543352601156069
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08347599497023044,
+      "wer": 0.2849106700868548
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09312379505754513,
+      "wer": 0.3060253699788584
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07567203062037578,
+      "wer": 0.2647670248166734
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.19393939393939394,
+      "wer": 0.5180722891566265
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.12351229833377414,
+      "wer": 0.37576687116564417
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.07538280329799764,
+      "wer": 0.2916666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.22105263157894736,
+      "wer": 0.5189873417721519
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09024285098433683,
+      "wer": 0.27899159663865547
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.12105263157894737,
+      "wer": 0.4166666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.12790697674418605,
+      "wer": 0.4222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.030054644808743168,
+      "wer": 0.15873015873015872
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07885906040268456,
+      "wer": 0.330188679245283
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.09368530020703934,
+      "wer": 0.26461538461538464
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06688963210702341,
+      "wer": 0.19230769230769232
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.08870967741935484,
+      "wer": 0.39473684210526316
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07259190321079571,
+      "wer": 0.24653739612188366
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.10139860139860139,
+      "wer": 0.36
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.1,
+      "wer": 0.3125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12915129151291513,
+      "wer": 0.41304347826086957
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.008583690987124463,
+      "wer": 0.07894736842105263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.0763749006893563,
+      "wer": 0.26973209561829636
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.20209059233449478,
+      "wer": 0.625
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.23827629911280102,
+      "wer": 0.6060606060606061
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.1858108108108108,
+      "wer": 0.4375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.10014306151645208,
+      "wer": 0.336
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08689317424323773,
+      "wer": 0.2863018994682503
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08554592712542733,
+      "wer": 0.28253879900686263
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10352574389069735,
+      "wer": 0.327307586962476
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.0918757736018904,
+      "wer": 0.30072944297082227
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.0984961387345888,
+      "wer": 0.31897491821155943
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08440702781844803,
+      "wer": 0.28344370860927154
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08460723475102164,
+      "wer": 0.28525932666060055
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07931131809201242,
+      "wer": 0.26101694915254237
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.096472874009864,
+      "wer": 0.3201814058956916
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08543570460408377,
+      "wer": 0.2824049345498979
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.15890142226581658,
+      "wer": 0.4117647058823529
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09563403215957664,
+      "wer": 0.3090178408329781
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.094296491027546,
+      "wer": 0.31317544263048974
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08642179780810509,
+      "wer": 0.2847911887494122
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.018691588785046728,
+      "wer": 0.1
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.12666666666666668,
+      "wer": 0.37662337662337664
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.13493872898261614,
+      "wer": 0.38728813559322034
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.07523510971786834,
+      "wer": 0.2777777777777778
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.019762845849802372,
+      "wer": 0.09523809523809523
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.12272727272727273,
+      "wer": 0.34210526315789475
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07669674382110632,
+      "wer": 0.2773892773892774
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.08396946564885496,
+      "wer": 0.22727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.0851063829787234,
+      "wer": 0.2830188679245283
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.10377358490566038,
+      "wer": 0.3562753036437247
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.21146355036171396,
+      "wer": 0.5288461538461539
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.07142857142857142,
+      "wer": 0.2672413793103448
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.03070175438596491,
+      "wer": 0.13157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.09523809523809523,
+      "wer": 0.2857142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07786633966891478,
+      "wer": 0.2737226277372263
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.10725075528700906,
+      "wer": 0.3274336283185841
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.17142857142857143,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.084369383785913,
+      "wer": 0.28478196380895876
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.13553530751708429,
+      "wer": 0.42657342657342656
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.13653136531365315,
+      "wer": 0.36585365853658536
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.11376146788990826,
+      "wer": 0.3617021276595745
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.10771992818671454,
+      "wer": 0.3333333333333333
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:12aed0bf595997a5480b9149956caebba6eb4076565a393ac88570f79199d101
+size 130319758

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 sha256:2b9e11f1e2ba082c414cf16049d54650c92f83a2e6c4df9e20a5e94e3b783672
+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/tmp9v8x25rh", "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}