jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-5_sixties-5_s200

.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-5_sixties-5_s200
+
+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": 0.04461897590213297,
+      "wer": 0.18366722844405783
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.04397861435858708,
+      "wer": 0.1822064590254015
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.04913358596078143,
+      "wer": 0.19701529972410334
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.042870045158053184,
+      "wer": 0.171875
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.0479925913820341,
+      "wer": 0.19203572757722367
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.04861281610606433,
+      "wer": 0.20748299319727892
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.05476190476190476,
+      "wer": 0.2082036775106082
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.04730831973898858,
+      "wer": 0.21951219512195122
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.04548811106188155,
+      "wer": 0.18577208361429534
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.04713205869275233,
+      "wer": 0.1955044432828019
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.04397939379071999,
+      "wer": 0.1822163017234202
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.04757245013172751,
+      "wer": 0.1868843960601348
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.0765639589169001,
+      "wer": 0.2926829268292683
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.04728338248754195,
+      "wer": 0.18992180712423978
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04464497330154614,
+      "wer": 0.18403014938086093
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.06985294117647059,
+      "wer": 0.325
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.05405405405405406,
+      "wer": 0.3125
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.03513513513513514,
+      "wer": 0.23076923076923078
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.05,
+      "wer": 0.22807017543859648
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.04040404040404041,
+      "wer": 0.1956521739130435
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.05128525081108061,
+      "wer": 0.21934369602763384
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.07263922518159806,
+      "wer": 0.2727272727272727
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.04280821917808219,
+      "wer": 0.1696969696969697
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.11547911547911548,
+      "wer": 0.3898305084745763
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.07849829351535836,
+      "wer": 0.25133689839572193
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.04621167452830189,
+      "wer": 0.18883248730964466
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.01951219512195122,
+      "wer": 0.12121212121212122
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10233160621761658,
+      "wer": 0.3162393162393162
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.06706586826347305,
+      "wer": 0.232
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04191459154210602,
+      "wer": 0.17238544178908133
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.09038360483447189,
+      "wer": 0.3215547703180212
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.05094740380642328,
+      "wer": 0.19887909008489243
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.050128938997131216,
+      "wer": 0.1969033840493619
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.054871648058020805,
+      "wer": 0.21149276301577014
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.046324847615632846,
+      "wer": 0.19106699751861042
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06433135051773518,
+      "wer": 0.23196392785571143
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0653818004391315,
+      "wer": 0.2436548223350254
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.07949824875783987,
+      "wer": 0.2554016620498615
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.07339449541284404,
+      "wer": 0.2602739726027397
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.05076003104030675,
+      "wer": 0.19495133819951338
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06548775748153907,
+      "wer": 0.23841961852861035
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.050214416170301965,
+      "wer": 0.19716658749252955
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.07162317612157171,
+      "wer": 0.25963149078726966
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.13210702341137123,
+      "wer": 0.29545454545454547
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.0535230830802552,
+      "wer": 0.20378654737178464
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.0508959717217566,
+      "wer": 0.19917025059647736
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.05263866451265482,
+      "wer": 0.20313942751615882
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.07980652962515115,
+      "wer": 0.26141078838174275
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.07614213197969544,
+      "wer": 0.38095238095238093
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.04929577464788732,
+      "wer": 0.18421052631578946
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.07352941176470588,
+      "wer": 0.4444444444444444
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.15989159891598917,
+      "wer": 0.4716981132075472
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10501567398119123,
+      "wer": 0.3235294117647059
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.06886517943743937,
+      "wer": 0.2733333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.045454545454545456,
+      "wer": 0.1896551724137931
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.20202020202020202,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.09615384615384616,
+      "wer": 0.18518518518518517
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.06656580937972768,
+      "wer": 0.27979274611398963
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.06819565658615791,
+      "wer": 0.25770308123249297
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.21505376344086022,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.10535117056856187,
+      "wer": 0.3804347826086957
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.010869565217391304,
+      "wer": 0.07142857142857142
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04676473314179035,
+      "wer": 0.18135929331170003
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.09230769230769231,
+      "wer": 0.4642857142857143
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b7f05e104f873d73b50c21d430898d1847d6f7d09a04c91526e1061647c0b21
+size 138807463

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 sha256:96456ddcbeef01bf0e3a74a15438a14ba8f9fe61c715a93f311a6e52300971bc
+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/tmpx03kxsiu", "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}