jonatasgrosman/exp_w2v2r_en_xls-r_age_teens-2_sixties-8_s717

.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:
+- en
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- en
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_en_xls-r_age_teens-2_sixties-8_s717
+
+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": 0.10921618294435743,
+      "wer": 0.3014066016504126
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.1076185552595086,
+      "wer": 0.29795872672314067
+    },
+    {
+      "subset_type": "age",
+      "subset": "nineties",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06492411467116357,
+      "wer": 0.21770334928229665
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.11827956989247312,
+      "wer": 0.34615384615384615
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.11336442912188197,
+      "wer": 0.31040564373897706
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.1347241607287152,
+      "wer": 0.354888640467796
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.11317131524271233,
+      "wer": 0.3085359166514348
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.07838640043173233,
+      "wer": 0.2482213438735178
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09709962168978563,
+      "wer": 0.27875457875457876
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.0798841764605689,
+      "wer": 0.23576423576423577
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.10770326089745234,
+      "wer": 0.2980767077894454
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.11199308393586922,
+      "wer": 0.3085544373284538
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.11078140454995054,
+      "wer": 0.32934131736526945
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.11813940724478594,
+      "wer": 0.3204796614154714
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.11012663407542202,
+      "wer": 0.3030500311513395
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.16894705532421178,
+      "wer": 0.43252595155709345
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.0865561694290976,
+      "wer": 0.2391304347826087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.16247139588100687,
+      "wer": 0.34415584415584416
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.07407407407407407,
+      "wer": 0.125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.06924643584521385,
+      "wer": 0.22661870503597123
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.03940886699507389,
+      "wer": 0.175
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.08606691736573137,
+      "wer": 0.2558637798827244
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.14506998764923837,
+      "wer": 0.3737221888153939
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.1366520253782333,
+      "wer": 0.36151603498542273
+    },
+    {
+      "subset_type": "accent",
+      "subset": "southatlandtic",
+      "cer": 0.14835164835164835,
+      "wer": 0.42424242424242425
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.10780669144981413,
+      "wer": 0.35
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.07457627118644068,
+      "wer": 0.22916666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.15897435897435896,
+      "wer": 0.3888888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.10099622122981793,
+      "wer": 0.28897715988083417
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.12121212121212122,
+      "wer": 0.35039370078740156
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.07295373665480427,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.061255167230364525,
+      "wer": 0.19230769230769232
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.14188005331998393,
+      "wer": 0.35016020401490877
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.14044360329586633,
+      "wer": 0.34687093510622924
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.11151295961422544,
+      "wer": 0.2684563758389262
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.11428571428571428,
+      "wer": 0.44
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.1523151182885406,
+      "wer": 0.3866575901974132
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.15579278384640846,
+      "wer": 0.385062398780604
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.1320454349883831,
+      "wer": 0.3441463414634146
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.19865609924190214,
+      "wer": 0.40913508260447035
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.14191530724368184,
+      "wer": 0.34116541353383456
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.14033680834001605,
+      "wer": 0.2709251101321586
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.1403794457177674,
+      "wer": 0.3468706515705672
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.15194565763918327,
+      "wer": 0.37236612912288586
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.1787941787941788,
+      "wer": 0.4357976653696498
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.1435276937519558,
+      "wer": 0.3547427891763307
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.14238791536755757,
+      "wer": 0.350956486692672
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.24277456647398843,
+      "wer": 0.55
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.1254180602006689,
+      "wer": 0.32323232323232326
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.1657754010695187,
+      "wer": 0.3939393939393939
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.14382022471910114,
+      "wer": 0.35714285714285715
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.17229394240317775,
+      "wer": 0.32303370786516855
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.07681940700808626,
+      "wer": 0.2647058823529412
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.12340437040242319,
+      "wer": 0.3133119364206022
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.16225274725274724,
+      "wer": 0.40182218033301914
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.19257773319959878,
+      "wer": 0.4277456647398844
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.1747269890795632,
+      "wer": 0.3745173745173745
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.15753424657534246,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.1187010078387458,
+      "wer": 0.3375796178343949
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.1223441051487301,
+      "wer": 0.34143583945732053
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.10885341074020319,
+      "wer": 0.28270042194092826
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.14305750350631136,
+      "wer": 0.3475177304964539
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.11241970021413276,
+      "wer": 0.3089171974522293
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eee1d98d3d2b529df9fef36a86a11d1819db64a4768a204048db1a18554084cc
+size 123167332

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 sha256:83c300e71b390a6a7ae0b6e02aa275216a7223e4616154182db616e8665c4783
+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/tmpm8mdxp32", "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}