jonatasgrosman/exp_w2v2r_fr_vp-100k_age_teens-2_sixties-8_s724

.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_vp-100k_age_teens-2_sixties-8_s724
+
+Fine-tuned [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) 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 +1,116 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-100k-voxpopuli",
+  "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_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "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_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "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": 0.10343887013253039,
+      "wer": 0.3444175362142889
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.10104674206863742,
+      "wer": 0.3391494937834423
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.10305160568780955,
+      "wer": 0.3505747126436782
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.09812006429519883,
+      "wer": 0.3300513698630137
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.11525029103608847,
+      "wer": 0.3629081947011707
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08571428571428572,
+      "wer": 0.3793103448275862
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.11664750682569335,
+      "wer": 0.37869885695972655
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.11784196780103691,
+      "wer": 0.38250985850150776
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.11903378434465012,
+      "wer": 0.3649122807017544
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.11494397582777288,
+      "wer": 0.36477064220183486
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.10104311238269509,
+      "wer": 0.33914202844311375
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.10884997633696167,
+      "wer": 0.34971098265895956
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.12288952503359234,
+      "wer": 0.3796687808315715
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.11235172027209235,
+      "wer": 0.36578011790648485
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.10218072941856897,
+      "wer": 0.34140249613548274
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.26105263157894737,
+      "wer": 0.6582278481012658
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.12060041407867494,
+      "wer": 0.3446153846153846
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.09060402684563758,
+      "wer": 0.4056603773584906
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.1,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.18951612903225806,
+      "wer": 0.47368421052631576
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.1207069981319155,
+      "wer": 0.3621848739495798
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.0546448087431694,
+      "wer": 0.25396825396825395
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.14878397711015737,
+      "wer": 0.464
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.25435540069686413,
+      "wer": 0.6458333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.10344357149162581,
+      "wer": 0.3478909143379347
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.09187279151943463,
+      "wer": 0.3402777777777778
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.162787622322137,
+      "wer": 0.4754601226993865
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.24646464646464647,
+      "wer": 0.5783132530120482
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.15,
+      "wer": 0.5333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.22972972972972974,
+      "wer": 0.4375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08701721731037693,
+      "wer": 0.27977839335180055
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.16279069767441862,
+      "wer": 0.5777777777777777
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.16974169741697417,
+      "wer": 0.6086956521739131
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.030042918454935622,
+      "wer": 0.15789473684210525
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.14335664335664336,
+      "wer": 0.5
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.2965779467680608,
+      "wer": 0.6818181818181818
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06354515050167224,
+      "wer": 0.21153846153846154
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.11687416175563374,
+      "wer": 0.3653798619605656
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.11540500192082319,
+      "wer": 0.3615137288793514
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.10640699971775332,
+      "wer": 0.3610169491525424
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.11472680490388981,
+      "wer": 0.36442220200181985
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.13078625299191618,
+      "wer": 0.4043075245365322
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.13369400761083858,
+      "wer": 0.4067378800328677
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.12670192415888376,
+      "wer": 0.3895888594164456
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.10973645680819911,
+      "wer": 0.34922737306843266
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.12845613510685996,
+      "wer": 0.38866213151927437
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.12588105307273212,
+      "wer": 0.3966429064152679
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.11532921360869418,
+      "wer": 0.36142797561992845
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.19764590485532124,
+      "wer": 0.5
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.12573783838795033,
+      "wer": 0.38720087681909515
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.11619716643637258,
+      "wer": 0.3633999686533776
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.10031347962382445,
+      "wer": 0.3148148148148148
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.2615470228158041,
+      "wer": 0.6442307692307693
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.04672897196261682,
+      "wer": 0.2
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.12466307277628032,
+      "wer": 0.38866396761133604
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.09210526315789473,
+      "wer": 0.23684210526315788
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09807767752059632,
+      "wer": 0.3543123543123543
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.13734290843806105,
+      "wer": 0.3978494623655914
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.11854103343465046,
+      "wer": 0.3584905660377358
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.14285714285714285,
+      "wer": 0.5714285714285714
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.11567246854920112,
+      "wer": 0.36732423613171167
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.043478260869565216,
+      "wer": 0.21428571428571427
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.17697349672271304,
+      "wer": 0.48898305084745763
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.18,
+      "wer": 0.5194805194805194
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.19557195571955718,
+      "wer": 0.4878048780487805
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.11376146788990826,
+      "wer": 0.3723404255319149
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.11342734518700184,
+      "wer": 0.3722627737226277
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.11363636363636363,
+      "wer": 0.39473684210526316
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.17557251908396945,
+      "wer": 0.4090909090909091
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.2571428571428571,
+      "wer": 0.5555555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.09214501510574018,
+      "wer": 0.3185840707964602
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.1856492027334852,
+      "wer": 0.4965034965034965
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.09663865546218488,
+      "wer": 0.31896551724137934
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7930ec7685b81439550a1fb025b1e5349e3487a24a264fb4c8cfe084a5e2ab8c
+size 129106616

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 sha256:873f709b8eae2e57f711e62a434bf74725ff74b4bae15b30708da7f1dbbccc88
+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/tmpv04h8eon", "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}