jonatasgrosman/exp_w2v2r_fr_xls-r_gender_male-2_female-8_s886

.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_gender_male-2_female-8_s886
+
+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": 0.07545858146739702,
+      "wer": 0.2644106352117266
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07345034471257578,
+      "wer": 0.2593014367502301
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06723041442448809,
+      "wer": 0.23929794520547945
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.08116312509985621,
+      "wer": 0.2796934865900383
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08244649748567419,
+      "wer": 0.2878682440269079
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.08420644159875824,
+      "wer": 0.28773875539125077
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08375529177388562,
+      "wer": 0.2781954887218045
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09259725544504595,
+      "wer": 0.30788990825688073
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.08884178761316282,
+      "wer": 0.2964426877470356
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08,
+      "wer": 0.2413793103448276
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07345465071500754,
+      "wer": 0.2593017090818363
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.09166325874861249,
+      "wer": 0.30320648343904155
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.08291544836153741,
+      "wer": 0.2838756131587237
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.08092759110269758,
+      "wer": 0.27167630057803466
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.07426653415684735,
+      "wer": 0.26111952970919033
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.21926489226869456,
+      "wer": 0.5833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07584923220102373,
+      "wer": 0.24930747922437674
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06688963210702341,
+      "wer": 0.19230769230769232
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.09268573070843512,
+      "wer": 0.2823529411764706
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.11954509389050516,
+      "wer": 0.37116564417177916
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.16666666666666666,
+      "wer": 0.4888888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.11070110701107011,
+      "wer": 0.391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.1858108108108108,
+      "wer": 0.4583333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "gabon",
+      "cer": 0.07777777777777778,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07613414761373577,
+      "wer": 0.2700206820258766
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.1048951048951049,
+      "wer": 0.36
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.17575757575757575,
+      "wer": 0.4819277108433735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07214765100671142,
+      "wer": 0.27358490566037735
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.11444921316165951,
+      "wer": 0.368
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.10080645161290322,
+      "wer": 0.34210526315789475
+    },
+    {
+      "subset_type": "accent",
+      "subset": "st_pierre_et_miquelon",
+      "cer": 0.15331010452961671,
+      "wer": 0.5833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.10526315789473684,
+      "wer": 0.36666666666666664
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.08362779740871613,
+      "wer": 0.3055555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.030054644808743168,
+      "wer": 0.19047619047619047
+    },
+    {
+      "subset_type": "accent",
+      "subset": "italy",
+      "cer": 0.21894736842105264,
+      "wer": 0.5569620253164557
+    },
+    {
+      "subset_type": "accent",
+      "subset": "mayotte",
+      "cer": 0.02575107296137339,
+      "wer": 0.10526315789473684
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.10559006211180125,
+      "wer": 0.3076923076923077
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.08510562142199932,
+      "wer": 0.2819857904310405
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.08384271618986486,
+      "wer": 0.27837143045474855
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08438020281519601,
+      "wer": 0.2807097361237489
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.07902907140841095,
+      "wer": 0.2711864406779661
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08804996061663103,
+      "wer": 0.28514588859416445
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.09547035180418191,
+      "wer": 0.31215921483097053
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08140556368960468,
+      "wer": 0.27593818984547464
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.09834105514870722,
+      "wer": 0.32108843537414966
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.10027050570812893,
+      "wer": 0.3244316625582032
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.08376571009381119,
+      "wer": 0.2782562248937213
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.0926963081495028,
+      "wer": 0.3048976776270407
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.09286586606961124,
+      "wer": 0.3035986117030993
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.15448749386954388,
+      "wer": 0.4441176470588235
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.08474191289205202,
+      "wer": 0.2806448855135859
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cote_d_ivoire",
+      "cer": 0.13895216400911162,
+      "wer": 0.43356643356643354
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.037383177570093455,
+      "wer": 0.15
+    },
+    {
+      "subset_type": "accent",
+      "subset": "reunion",
+      "cer": 0.0877742946708464,
+      "wer": 0.25925925925925924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.08828939301042306,
+      "wer": 0.3102189781021898
+    },
+    {
+      "subset_type": "accent",
+      "subset": "morocco",
+      "cer": 0.06862745098039216,
+      "wer": 0.25862068965517243
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1305215161014534,
+      "wer": 0.38389830508474576
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07532365633581797,
+      "wer": 0.28554778554778554
+    },
+    {
+      "subset_type": "accent",
+      "subset": "new_caledonia",
+      "cer": 0.15498154981549817,
+      "wer": 0.4146341463414634
+    },
+    {
+      "subset_type": "accent",
+      "subset": "romania",
+      "cer": 0.0916030534351145,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "burundi",
+      "cer": 0.05952380952380952,
+      "wer": 0.2857142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.37142857142857144,
+      "wer": 0.5555555555555556
+    },
+    {
+      "subset_type": "accent",
+      "subset": "tunisia",
+      "cer": 0.11743119266055047,
+      "wer": 0.3723404255319149
+    },
+    {
+      "subset_type": "accent",
+      "subset": "senegal",
+      "cer": 0.09667673716012085,
+      "wer": 0.2743362831858407
+    },
+    {
+      "subset_type": "accent",
+      "subset": "cameroon",
+      "cer": 0.13777777777777778,
+      "wer": 0.4155844155844156
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.10714285714285714,
+      "wer": 0.32388663967611336
+    },
+    {
+      "subset_type": "accent",
+      "subset": "algeria",
+      "cer": 0.09066427289048475,
+      "wer": 0.27956989247311825
+    },
+    {
+      "subset_type": "accent",
+      "subset": "martinique",
+      "cer": 0.02631578947368421,
+      "wer": 0.13157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "guadeloupe",
+      "cer": 0.015810276679841896,
+      "wer": 0.09523809523809523
+    },
+    {
+      "subset_type": "accent",
+      "subset": "benin",
+      "cer": 0.09545454545454546,
+      "wer": 0.3157894736842105
+    },
+    {
+      "subset_type": "accent",
+      "subset": "monaco",
+      "cer": 0.11246200607902736,
+      "wer": 0.3018867924528302
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.0810581255077673,
+      "wer": 0.27654998516760604
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.2159154145798553,
+      "wer": 0.5737179487179487
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e5adc5934d7e9c4ca18d65adac885bcd905a56e97134e1193713eae7e5b4cd7a
+size 130658499

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 sha256:440836f6dc9fe1c106b3a54d8ebf81f69041f80f6ad5ee8d3e6a97b2d0afe00c
+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/tmpynby_r3z", "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}