jonatasgrosman/exp_w2v2r_fr_xls-r_accent_france-2_belgium-8_s587

.gitattributes

@@ -29,3 +29,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zstandard 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_accent_france-2_belgium-8_s587
+
+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": 0.07782538508319367,
+      "wer": 0.26839149131943335
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07585486521111302,
+      "wer": 0.2631156108138465
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07121392130826752,
+      "wer": 0.2521404109589041
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07876657613037226,
+      "wer": 0.28160919540229884
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08521420496628075,
+      "wer": 0.28717234980282996
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08733669641702238,
+      "wer": 0.2917436845348121
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08682659583298746,
+      "wer": 0.27769423558897244
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09303789500188846,
+      "wer": 0.3071559633027523
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.09056617330076161,
+      "wer": 0.305309261831001
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.05714285714285714,
+      "wer": 0.20689655172413793
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07585050100925517,
+      "wer": 0.2630988023952096
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09733013962727113,
+      "wer": 0.3201198026779422
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08415016589149636,
+      "wer": 0.28545069978848836
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.08991954566966398,
+      "wer": 0.2861271676300578
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07676435266974874,
+      "wer": 0.2656655726719615
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.25602027883396705,
+      "wer": 0.6515151515151515
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.0693345742205677,
+      "wer": 0.24376731301939059
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06688963210702341,
+      "wer": 0.21153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09441011639603392,
+      "wer": 0.2865546218487395
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1399100766992859,
+      "wer": 0.4148773006134969
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.18604651162790697,
+      "wer": 0.5333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.13284132841328414,
+      "wer": 0.391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.20608108108108109,
+      "wer": 0.5625
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.08148148148148149,
+      "wer": 0.2916666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.0764872521246459,
+      "wer": 0.26737530662305803
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.09090909090909091,
+      "wer": 0.32
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.19595959595959597,
+      "wer": 0.5060240963855421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06543624161073826,
+      "wer": 0.27358490566037735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.11731044349070101,
+      "wer": 0.368
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.10080645161290322,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.21951219512195122,
+      "wer": 0.6666666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.08947368421052632,
+      "wer": 0.3
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.061248527679623084,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.0273224043715847,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.2063157894736842,
+      "wer": 0.5443037974683544
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.012875536480686695,
+      "wer": 0.10526315789473684
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.11645962732919254,
+      "wer": 0.36615384615384616
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08961977130272386,
+      "wer": 0.28997154283468857
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08810013567333362,
+      "wer": 0.28564418436028327
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.0911154835780233,
+      "wer": 0.29981801637852595
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.09568162574089754,
+      "wer": 0.30847457627118646
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09035670079891978,
+      "wer": 0.2921087533156499
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09948968071173735,
+      "wer": 0.3127044711014177
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08528550512445095,
+      "wer": 0.2891832229580574
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.10312359886414586,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10923387281646875,
+      "wer": 0.3412763626403725
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.0880469687474191,
+      "wer": 0.2855878070374408
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09768735474530423,
+      "wer": 0.31754426304897676
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09932831263993487,
+      "wer": 0.3160202155513609
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.15301618440411965,
+      "wer": 0.4411764705882353
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08900432511215006,
+      "wer": 0.288089708333927
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.15148063781321183,
+      "wer": 0.45454545454545453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.06542056074766354,
+      "wer": 0.3
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.10658307210031348,
+      "wer": 0.3148148148148148
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08583690987124463,
+      "wer": 0.3175182481751825
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.09243697478991597,
+      "wer": 0.3103448275862069
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1677115987460815,
+      "wer": 0.45508474576271185
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.0690466849744998,
+      "wer": 0.2517482517482518
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.13653136531365315,
+      "wer": 0.36585365853658536
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.10687022900763359,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.27380952380952384,
+      "wer": 0.6428571428571429
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.17142857142857143,
+      "wer": 0.5555555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.12844036697247707,
+      "wer": 0.3404255319148936
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.08459214501510574,
+      "wer": 0.2920353982300885
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.16666666666666666,
+      "wer": 0.45454545454545453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.11657681940700809,
+      "wer": 0.3643724696356275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.11220825852782765,
+      "wer": 0.3709677419354839
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.02631578947368421,
+      "wer": 0.13157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.015810276679841896,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.11818181818181818,
+      "wer": 0.34210526315789475
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.11246200607902736,
+      "wer": 0.2830188679245283
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.08472635958541569,
+      "wer": 0.2839661821417977
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.25097384529771843,
+      "wer": 0.6121794871794872
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5e11bd24bbfe119bedb976ddf56c8e520300bc2e290198d3a5bfb7a905c0711e
+size 130571821

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 sha256:61436e2c56f06da0f6697333f37774c594b8c886e195bcc026897893659656a7
+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/tmpmpuey3u1", "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}