jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-0_sixties-10_s423

.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-0_sixties-10_s423
+
+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": 0.049865084603708774,
+      "wer": 0.204163042296401
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.048857243868944404,
+      "wer": 0.20152192555219756
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.05643648136565984,
+      "wer": 0.22422874341610233
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.05023582538886101,
+      "wer": 0.19668311403508773
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.05997712044451708,
+      "wer": 0.23706736136955714
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.0542597593911122,
+      "wer": 0.23299319727891157
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.060066833751044275,
+      "wer": 0.23677510608203678
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.052202283849918436,
+      "wer": 0.21951219512195122
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.048638802737163395,
+      "wer": 0.19453809844908967
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.05269008448199199,
+      "wer": 0.2138003136434919
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.04885436883864001,
+      "wer": 0.20152360336358338
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.057621377493413624,
+      "wer": 0.22032141005702435
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.08870214752567694,
+      "wer": 0.3475609756097561
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.0536331740058985,
+      "wer": 0.21420503909643787
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04949554096318913,
+      "wer": 0.20358867630396876
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.09191176470588236,
+      "wer": 0.325
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07094594594594594,
+      "wer": 0.3541666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.04594594594594595,
+      "wer": 0.25
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.07777777777777778,
+      "wer": 0.2807017543859649
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.050505050505050504,
+      "wer": 0.17391304347826086
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.05764911405041178,
+      "wer": 0.23402417962003455
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.07990314769975787,
+      "wer": 0.30578512396694213
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.04708904109589041,
+      "wer": 0.2
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.13267813267813267,
+      "wer": 0.4576271186440678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.08617747440273038,
+      "wer": 0.2994652406417112
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.053287146226415096,
+      "wer": 0.21116751269035533
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.03902439024390244,
+      "wer": 0.15151515151515152
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12435233160621761,
+      "wer": 0.4188034188034188
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.08023952095808383,
+      "wer": 0.264
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04916489862331903,
+      "wer": 0.19792808594606445
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.09826589595375723,
+      "wer": 0.36042402826855124
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.05653508030346536,
+      "wer": 0.2191681090139839
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.05541699942445915,
+      "wer": 0.21573602115509907
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.06625950313325063,
+      "wer": 0.25556275653488875
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.04926496952312657,
+      "wer": 0.20347394540942929
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.0752001175001836,
+      "wer": 0.26152304609218435
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06464991461332033,
+      "wer": 0.25549915397631134
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.08682902989329641,
+      "wer": 0.2847645429362881
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.08623853211009175,
+      "wer": 0.3150684931506849
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.05842881270826676,
+      "wer": 0.22962287104622872
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07228915662650602,
+      "wer": 0.28065395095367845
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.05553858264555881,
+      "wer": 0.21607643624438758
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.08094690319065205,
+      "wer": 0.292713567839196
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.13210702341137123,
+      "wer": 0.3181818181818182
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.061244909009581655,
+      "wer": 0.23598018852851893
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.05623685240853921,
+      "wer": 0.21821193562603527
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.061389337641357025,
+      "wer": 0.24284395198522624
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.08706166868198308,
+      "wer": 0.31950207468879666
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.06598984771573604,
+      "wer": 0.19047619047619047
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.04225352112676056,
+      "wer": 0.23684210526315788
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.04411764705882353,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.16802168021680217,
+      "wer": 0.5094339622641509
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.12695924764890282,
+      "wer": 0.38235294117647056
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.0892337536372454,
+      "wer": 0.3466666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.07575757575757576,
+      "wer": 0.29310344827586204
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.16161616161616163,
+      "wer": 0.4888888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.12179487179487179,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.08774583963691376,
+      "wer": 0.33678756476683935
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07793789324132332,
+      "wer": 0.27170868347338933
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.22580645161290322,
+      "wer": 0.4666666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.09197324414715718,
+      "wer": 0.33695652173913043
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.021739130434782608,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.054298462801382286,
+      "wer": 0.21236704524968453
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.11282051282051282,
+      "wer": 0.5357142857142857
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:07bf0747aec744dd172886bd7f7c136abbb4b13d5539a1dbbdf1cb84a480d112
+size 138458822

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:f2860231d642e1d9977e27b708df1e99ccdec23545f88090a01a7963f8491726
+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/tmp5cvlmkgo", "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}