jonatasgrosman/exp_w2v2r_fr_vp-100k_age_teens-0_sixties-10_s131

.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-0_sixties-10_s131
+
+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": 0.1059054403581409,
+      "wer": 0.34772107170013866
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.10335042783659086,
+      "wer": 0.34161427702291625
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.10576769451989136,
+      "wer": 0.35344827586206895
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.09455587392550144,
+      "wer": 0.331763698630137
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.11879446384685034,
+      "wer": 0.37707948243992606
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.09142857142857143,
+      "wer": 0.3448275862068966
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.12104828279925277,
+      "wer": 0.3851084285866895
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.12002494834912096,
+      "wer": 0.3813500347947112
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.1186187432555823,
+      "wer": 0.36390977443609024
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.12388266398086366,
+      "wer": 0.38311926605504587
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.10334979672748043,
+      "wer": 0.3416058507984032
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.10932323710364411,
+      "wer": 0.3439306358381503
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.12671613016299585,
+      "wer": 0.3939393939393939
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.11543472661990395,
+      "wer": 0.371270419873093
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.10445376952131039,
+      "wer": 0.34432658407133554
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.23789473684210527,
+      "wer": 0.569620253164557
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.13975155279503104,
+      "wer": 0.38769230769230767
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.09563758389261745,
+      "wer": 0.36792452830188677
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.13333333333333333,
+      "wer": 0.4375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.1532258064516129,
+      "wer": 0.42105263157894735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.11711452794941803,
+      "wer": 0.36050420168067226
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.04644808743169399,
+      "wer": 0.25396825396825395
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.15879828326180256,
+      "wer": 0.488
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.3310104529616725,
+      "wer": 0.7708333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.10676596393518928,
+      "wer": 0.35226780818623443
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.09540636042402827,
+      "wer": 0.3263888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.16609362602486114,
+      "wer": 0.48236196319018404
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.26666666666666666,
+      "wer": 0.6265060240963856
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.14736842105263157,
+      "wer": 0.48333333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.2533783783783784,
+      "wer": 0.5416666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.09073987901349465,
+      "wer": 0.30193905817174516
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.17054263565891473,
+      "wer": 0.5777777777777777
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.16605166051660517,
+      "wer": 0.6521739130434783
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.02575107296137339,
+      "wer": 0.13157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.20279720279720279,
+      "wer": 0.52
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.311787072243346,
+      "wer": 0.7121212121212122
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06020066889632107,
+      "wer": 0.25
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.12009659943867894,
+      "wer": 0.36980371527623734
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.11869471554796952,
+      "wer": 0.3660326504903433
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.1109229466553768,
+      "wer": 0.3610169491525424
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.12108369910700773,
+      "wer": 0.3707916287534122
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.13430881091089736,
+      "wer": 0.4034896401308615
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.13596350465361515,
+      "wer": 0.41550260202684197
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.12827725891751998,
+      "wer": 0.384946949602122
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.11076134699853588,
+      "wer": 0.35275938189845474
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.1311463159467942,
+      "wer": 0.3941043083900227
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.1297291118985027,
+      "wer": 0.4076799264198666
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.11863762455122694,
+      "wer": 0.3659791174297024
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.20892594409024032,
+      "wer": 0.5029411764705882
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.12790555668634235,
+      "wer": 0.38884491262254156
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.11961135507561849,
+      "wer": 0.36856503711725064
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.13166144200626959,
+      "wer": 0.42592592592592593
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.25598219254312743,
+      "wer": 0.6153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.04672897196261682,
+      "wer": 0.15
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.1421832884097035,
+      "wer": 0.44534412955465585
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.11403508771929824,
+      "wer": 0.2631578947368421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.10553158101216163,
+      "wer": 0.34965034965034963
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.12926391382405744,
+      "wer": 0.3978494623655914
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.1276595744680851,
+      "wer": 0.37735849056603776
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.17857142857142858,
+      "wer": 0.7142857142857143
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.11605406336935917,
+      "wer": 0.3639127855235835
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.03162055335968379,
+      "wer": 0.19047619047619047
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.18367056141350813,
+      "wer": 0.49745762711864405
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.18888888888888888,
+      "wer": 0.45454545454545453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.2140221402214022,
+      "wer": 0.43902439024390244
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.1596330275229358,
+      "wer": 0.4574468085106383
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.11526670754138565,
+      "wer": 0.3722627737226277
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.09545454545454546,
+      "wer": 0.34210526315789475
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.13740458015267176,
+      "wer": 0.45454545454545453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.22857142857142856,
+      "wer": 0.5555555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.10574018126888217,
+      "wer": 0.3274336283185841
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.19362186788154898,
+      "wer": 0.5104895104895105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.08403361344537816,
+      "wer": 0.29310344827586204
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1c55a7b750a62b1774979492ff2156b0f34bee29f3074293b238c15733f1330f
+size 129036845

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:c76d076c5c51bf9c1db8c6aff7b7fc68797f6f38e151b3bbdc815e8972b1060d
+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/tmpnd9uwqfq", "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}