jonatasgrosman/exp_w2v2r_en_xls-r_accent_us-10_england-0_s947

.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:
+- en
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- en
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_en_xls-r_accent_us-10_england-0_s947
+
+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 (en)](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": 33,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,390 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.10660689693696249,
+      "wer": 0.2973805951487872
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.1054133831752682,
+      "wer": 0.2941597940067918
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.08363140861863397,
+      "wer": 0.25374625374625376
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.08010118043844856,
+      "wer": 0.25598086124401914
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09533417402269861,
+      "wer": 0.2827838827838828
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.10813448340993514,
+      "wer": 0.3032352403582526
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07568807339449542,
+      "wer": 0.2426877470355731
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.1113364429121882,
+      "wer": 0.31099353321575546
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.12720193932270624,
+      "wer": 0.3425917963711411
+    },
+    {
+      "subset_type": "age",
+      "subset": "nineties",
+      "cer": 0.038461538461538464,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08602150537634409,
+      "wer": 0.2692307692307692
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.10548913461785156,
+      "wer": 0.2942704251841569
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.10303363722099969,
+      "wer": 0.29551692589204026
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.11448311487053658,
+      "wer": 0.3171878673877263
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.09990108803165183,
+      "wer": 0.3083832335329341
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.1074861174295762,
+      "wer": 0.29880084843648275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.12493899463152758,
+      "wer": 0.3673469387755102
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.04938271604938271,
+      "wer": 0.125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.06761565836298933,
+      "wer": 0.2159090909090909
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.13076923076923078,
+      "wer": 0.3611111111111111
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.07457627118644068,
+      "wer": 0.2708333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.06904276985743381,
+      "wer": 0.21942446043165467
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.08305852858933564,
+      "wer": 0.2502255299954894
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.09576427255985268,
+      "wer": 0.2391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.054187192118226604,
+      "wer": 0.175
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.11923583662714098,
+      "wer": 0.35826771653543305
+    },
+    {
+      "subset_type": "accent",
+      "subset": "southatlandtic",
+      "cer": 0.12087912087912088,
+      "wer": 0.30303030303030304
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.10037174721189591,
+      "wer": 0.29
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.14455537258130918,
+      "wer": 0.3773301262778112
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.09575747165922363,
+      "wer": 0.28649453823237336
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.08417888012025554,
+      "wer": 0.25339366515837103
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.16590389016018306,
+      "wer": 0.34415584415584416
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.15288518738845924,
+      "wer": 0.39792387543252594
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.13788571129207824,
+      "wer": 0.3418688288759563
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.13672224492538954,
+      "wer": 0.3388837754162832
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.10849909584086799,
+      "wer": 0.28859060402684567
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.14394546912590217,
+      "wer": 0.3039647577092511
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.13211055701124166,
+      "wer": 0.32095864661654133
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.12709749591257208,
+      "wer": 0.3302439024390244
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.19986216402481047,
+      "wer": 0.4062196307094266
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.15118288540611405,
+      "wer": 0.3815520762423417
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.1483118172790467,
+      "wer": 0.3736305611126989
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.12,
+      "wer": 0.4
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.13666668023671594,
+      "wer": 0.33882581950073376
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.13471367476791488,
+      "wer": 0.3395777579542075
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.1468781733395116,
+      "wer": 0.3639377423161207
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.1704781704781705,
+      "wer": 0.40077821011673154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.13839407670376208,
+      "wer": 0.3425647373799893
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.17251755265797392,
+      "wer": 0.37572254335260113
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.12808988764044943,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.07994389901823282,
+      "wer": 0.24113475177304963
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.09630459126539753,
+      "wer": 0.27388535031847133
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.1643835616438356,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.17030784508440913,
+      "wer": 0.34129213483146065
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.11850930333189096,
+      "wer": 0.30674002751031637
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.24710982658959538,
+      "wer": 0.5416666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.09838274932614555,
+      "wer": 0.27941176470588236
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.09941944847605225,
+      "wer": 0.26582278481012656
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.1918876755070203,
+      "wer": 0.40540540540540543
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.15978021978021978,
+      "wer": 0.3939679547596607
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.12135586520407154,
+      "wer": 0.33352176370830977
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.11777301927194861,
+      "wer": 0.3248407643312102
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.13368983957219252,
+      "wer": 0.30303030303030304
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.1020066889632107,
+      "wer": 0.2828282828282828
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c8eb1ab4bbb2bfd0ff6f238ec3aa9ee3f8868e3332a3bb47a174f870cc5b88d1
+size 123393566

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 sha256:65cf87d7121b5684127f7f288ada2e7db32a97705ca2273b5fc383b10ea92d3e
+size 1262034033

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/tmphp35qadw", "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}