jonatasgrosman/exp_w2v2r_en_xls-r_age_teens-2_sixties-8_s246

.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_s246
+
+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": 0.11090163775934413,
+      "wer": 0.3080145036259065
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.10936575966831338,
+      "wer": 0.3039146173079076
+    },
+    {
+      "subset_type": "age",
+      "subset": "nineties",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06745362563237774,
+      "wer": 0.22966507177033493
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.1003584229390681,
+      "wer": 0.28846153846153844
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.11600081119448388,
+      "wer": 0.31716637272192827
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.13393080143980018,
+      "wer": 0.3623699372259008
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.11613045394446893,
+      "wer": 0.32133065253152987
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.08054506206152186,
+      "wer": 0.2592885375494071
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.10384615384615385,
+      "wer": 0.29926739926739926
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07835121785045138,
+      "wer": 0.24275724275724275
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.1094401826718321,
+      "wer": 0.30406973706796825
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.11293618359006602,
+      "wer": 0.3197621225983532
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.10731948565776459,
+      "wer": 0.344311377245509
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.1197375217924711,
+      "wer": 0.3298847872090289
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.11163223404076072,
+      "wer": 0.3090982912265227
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.16597263533610945,
+      "wer": 0.3944636678200692
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.12338858195211787,
+      "wer": 0.34782608695652173
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.16819221967963388,
+      "wer": 0.33116883116883117
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.037037037037037035,
+      "wer": 0.0625
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.08065173116089613,
+      "wer": 0.25539568345323743
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.06896551724137931,
+      "wer": 0.2
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.09027094783531,
+      "wer": 0.2690572846188543
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.1481576780568135,
+      "wer": 0.38604930847865304
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.1376281112737921,
+      "wer": 0.3673469387755102
+    },
+    {
+      "subset_type": "accent",
+      "subset": "southatlandtic",
+      "cer": 0.13736263736263737,
+      "wer": 0.42424242424242425
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.09293680297397769,
+      "wer": 0.3
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.07344632768361582,
+      "wer": 0.2638888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.13333333333333333,
+      "wer": 0.3472222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.0975609756097561,
+      "wer": 0.28897715988083417
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.13306982872200263,
+      "wer": 0.3700787401574803
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.09252669039145907,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.05787298008267568,
+      "wer": 0.2149321266968326
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.1430663647529294,
+      "wer": 0.3566533708232525
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.14151615824394245,
+      "wer": 0.3531922014894151
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.11452682338758288,
+      "wer": 0.3087248322147651
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.10285714285714286,
+      "wer": 0.36
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.1590489243787617,
+      "wer": 0.39312457454050376
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.1565375703409467,
+      "wer": 0.3905877869867581
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.13660614404956545,
+      "wer": 0.36341463414634145
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.19262577532736044,
+      "wer": 0.3965014577259475
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.14030935677457526,
+      "wer": 0.34116541353383456
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.1467522052927025,
+      "wer": 0.2907488986784141
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.14146776366944988,
+      "wer": 0.3531590032924502
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.1536906862793426,
+      "wer": 0.3797269202674608
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.16632016632016633,
+      "wer": 0.3735408560311284
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.14655262334411182,
+      "wer": 0.36990782039845377
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.1434844895671546,
+      "wer": 0.35772170483497184
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hongkong",
+      "cer": 0.2514450867052023,
+      "wer": 0.5416666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "philippines",
+      "cer": 0.12040133779264214,
+      "wer": 0.35353535353535354
+    },
+    {
+      "subset_type": "accent",
+      "subset": "singapore",
+      "cer": 0.13903743315508021,
+      "wer": 0.36363636363636365
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bermuda",
+      "cer": 0.1662921348314607,
+      "wer": 0.38095238095238093
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.1586395233366435,
+      "wer": 0.30758426966292135
+    },
+    {
+      "subset_type": "accent",
+      "subset": "newzealand",
+      "cer": 0.08221024258760108,
+      "wer": 0.2647058823529412
+    },
+    {
+      "subset_type": "accent",
+      "subset": "us",
+      "cer": 0.12475659887494592,
+      "wer": 0.32156503133119363
+    },
+    {
+      "subset_type": "accent",
+      "subset": "indian",
+      "cer": 0.17082417582417583,
+      "wer": 0.4093622368834433
+    },
+    {
+      "subset_type": "accent",
+      "subset": "african",
+      "cer": 0.17352056168505517,
+      "wer": 0.41040462427745666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.21450858034321374,
+      "wer": 0.444015444015444
+    },
+    {
+      "subset_type": "accent",
+      "subset": "wales",
+      "cer": 0.1506849315068493,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "malaysia",
+      "cer": 0.10414333706606943,
+      "wer": 0.28662420382165604
+    },
+    {
+      "subset_type": "accent",
+      "subset": "england",
+      "cer": 0.12036762525941298,
+      "wer": 0.323911814584511
+    },
+    {
+      "subset_type": "accent",
+      "subset": "ireland",
+      "cer": 0.10885341074020319,
+      "wer": 0.31223628691983124
+    },
+    {
+      "subset_type": "accent",
+      "subset": "scotland",
+      "cer": 0.1276297335203366,
+      "wer": 0.3049645390070922
+    },
+    {
+      "subset_type": "accent",
+      "subset": "australia",
+      "cer": 0.11991434689507495,
+      "wer": 0.32802547770700635
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ff5929fd275944a6d55453639de9595e7730760ec0c52810b1b0e03ddb056f2
+size 122816204

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:e734ccb715e6d0b6bad04fe0b12e89168542d74ce28d6253191a1c7ea9366e41
+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/tmpb9l5w9jp", "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}