jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-8_sixties-2_s945

.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:
+- de
+license: apache-2.0
+tags:
+- automatic-speech-recognition
+- de
+datasets:
+- mozilla-foundation/common_voice_7_0
+---
+# exp_w2v2r_de_xls-r_age_teens-8_sixties-2_s945
+
+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 (de)](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": 38,
+  "xvector_output_dim": 512
+}

cv7_evaluation.json

@@ -0,0 +1,396 @@
+{
+  "validation": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.04461700848840652,
+      "wer": 0.18237604653691422
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.043973804272799205,
+      "wer": 0.18037481995820534
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.04738604172031934,
+      "wer": 0.19312766491096062
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.042870045158053184,
+      "wer": 0.17214912280701755
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.046848613607888,
+      "wer": 0.194268701153703
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.05180456665848269,
+      "wer": 0.21428571428571427
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.057393483709273184,
+      "wer": 0.2164073550212164
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.048939641109298535,
+      "wer": 0.1951219512195122
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.0460788657509969,
+      "wer": 0.18981793661496965
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.049281161997925,
+      "wer": 0.2007318348144276
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.043973381241326895,
+      "wer": 0.18036799600366332
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.04678208505833647,
+      "wer": 0.18792120269569726
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.07376283846872082,
+      "wer": 0.2804878048780488
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.04745499847452456,
+      "wer": 0.19122502172024328
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04444514153201259,
+      "wer": 0.18178019558526923
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.0661764705882353,
+      "wer": 0.3
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.030405405405405407,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.04864864864864865,
+      "wer": 0.21153846153846154
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.044444444444444446,
+      "wer": 0.19298245614035087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.037037037037037035,
+      "wer": 0.15217391304347827
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.05353132018966808,
+      "wer": 0.22625215889464595
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.06779661016949153,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.0410958904109589,
+      "wer": 0.15151515151515152
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.12285012285012285,
+      "wer": 0.4745762711864407
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.07252559726962457,
+      "wer": 0.2887700534759358
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.048275353773584904,
+      "wer": 0.1964467005076142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.05365853658536585,
+      "wer": 0.18181818181818182
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.08549222797927461,
+      "wer": 0.3247863247863248
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.06946107784431138,
+      "wer": 0.24
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.043190773206565954,
+      "wer": 0.1764963823722868
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.08197582764056753,
+      "wer": 0.2862190812720848
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.0506042394306873,
+      "wer": 0.19609055193824007
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.049768668269852816,
+      "wer": 0.19402904993676465
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.056175843750994055,
+      "wer": 0.21386908619572262
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.04467551093581929,
+      "wer": 0.19305210918114143
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.060732907395167804,
+      "wer": 0.2159318637274549
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0670895340326909,
+      "wer": 0.2571912013536379
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.07632157693247536,
+      "wer": 0.2371191135734072
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08073394495412844,
+      "wer": 0.273972602739726
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.05162733372894509,
+      "wer": 0.19829683698296838
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06937427127866304,
+      "wer": 0.23705722070844687
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.049858728399494234,
+      "wer": 0.1942626844629695
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.07083610825210389,
+      "wer": 0.2533500837520938
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.12876254180602006,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.05344068554747275,
+      "wer": 0.20346700750918678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.05046352518033353,
+      "wer": 0.19586493017035697
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.057619816908992996,
+      "wer": 0.21421975992613113
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.06892382103990327,
+      "wer": 0.22821576763485477
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07106598984771574,
+      "wer": 0.2857142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.06690140845070422,
+      "wer": 0.2631578947368421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.058823529411764705,
+      "wer": 0.4444444444444444
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.18699186991869918,
+      "wer": 0.4716981132075472
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10031347962382445,
+      "wer": 0.37254901960784315
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07080504364694472,
+      "wer": 0.26666666666666666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.03535353535353535,
+      "wer": 0.1724137931034483
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.1919191919191919,
+      "wer": 0.5111111111111111
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.11538461538461539,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.06656580937972768,
+      "wer": 0.24870466321243523
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.06961639943170286,
+      "wer": 0.26330532212885155
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.20430107526881722,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.09531772575250837,
+      "wer": 0.33695652173913043
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.010869565217391304,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04705601970156368,
+      "wer": 0.18343248602848386
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.08717948717948718,
+      "wer": 0.39285714285714285
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6278fd087480ec4d1b7d36e28715f86200f10752c5d11ead455555cd3131d3a4
+size 138711641

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:e8cac5cf538f731975b86912baf102e455ce64eb674f75c6dd3b0bf23576de17
+size 1262054513

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/tmpo5jlx3il", "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, "ß": 33, "á": 34, "ä": 35, "ö": 36, "ü": 37}