jonatasgrosman/exp_w2v2r_de_vp-100k_age_teens-0_sixties-10_s50

.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_vp-100k_age_teens-0_sixties-10_s50
+
+Fine-tuned [facebook/wav2vec2-large-100k-voxpopuli](https://huggingface.co/facebook/wav2vec2-large-100k-voxpopuli) 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 +1,116 @@
+{
+  "_name_or_path": "facebook/wav2vec2-large-100k-voxpopuli",
+  "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_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "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_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.05,
+  "mask_time_selection": "static",
+  "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": 0.0676150912437301,
+      "wer": 0.2623409807545939
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.06636956370116862,
+      "wer": 0.25881892582694344
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.06669342699448068,
+      "wer": 0.2532894736842105
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.07166147917593005,
+      "wer": 0.2843700993204391
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06260741468205254,
+      "wer": 0.2568027210884354
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.0705951853492837,
+      "wer": 0.2791638570465273
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.07291858283359699,
+      "wer": 0.28342111863556557
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.06851549755301795,
+      "wer": 0.25609756097560976
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.07539358282943837,
+      "wer": 0.281726832899144
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.09081035923141187,
+      "wer": 0.31343705799151345
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.06637373526023516,
+      "wer": 0.258821080676047
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.07336265636123258,
+      "wer": 0.2784535186794092
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.11391223155929038,
+      "wer": 0.43902439024390244
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.07056831012420023,
+      "wer": 0.2682737169517885
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.06705052955418926,
+      "wer": 0.2606169694729483
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.11398963730569948,
+      "wer": 0.4358974358974359
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.057432432432432436,
+      "wer": 0.2708333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.06341463414634146,
+      "wer": 0.21212121212121213
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.16953316953316952,
+      "wer": 0.576271186440678
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.06849107470448418,
+      "wer": 0.2638675729006797
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.06964917452830188,
+      "wer": 0.2685279187817259
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.05405405405405406,
+      "wer": 0.2692307692307692
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.11259079903147699,
+      "wer": 0.36363636363636365
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.09812286689419795,
+      "wer": 0.3315508021390374
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.05565068493150685,
+      "wer": 0.23636363636363636
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.10059880239520957,
+      "wer": 0.36
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.06944444444444445,
+      "wer": 0.2807017543859649
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.116132422490804,
+      "wer": 0.42402826855123676
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.1111111111111111,
+      "wer": 0.34782608695652173
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.07773895682555527,
+      "wer": 0.30310880829015546
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.09191176470588236,
+      "wer": 0.4
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.07722198933289046,
+      "wer": 0.28407373828951343
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.07625470136568259,
+      "wer": 0.2816770781435634
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.07513580134203679,
+      "wer": 0.2767639902676399
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.09392534764576727,
+      "wer": 0.32318104906937395
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.10221531286436067,
+      "wer": 0.33651226158038144
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.0666905700968089,
+      "wer": 0.2640198511166253
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.08232337691255527,
+      "wer": 0.3039533376539209
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.1779816513761468,
+      "wer": 0.4794520547945205
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.0953220239406624,
+      "wer": 0.3301603206412826
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.10988026390812088,
+      "wer": 0.3490304709141274
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.07636359987244301,
+      "wer": 0.28185482017254854
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.0805494738328978,
+      "wer": 0.29285828407093784
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.1362876254180602,
+      "wer": 0.32954545454545453
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.10243991039535025,
+      "wer": 0.3559463986599665
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.0771413845810135,
+      "wer": 0.2837104691275474
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.13479623824451412,
+      "wer": 0.49019607843137253
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.21548821548821548,
+      "wer": 0.5777777777777777
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.23035230352303523,
+      "wer": 0.6226415094339622
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.043478260869565216,
+      "wer": 0.14285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.04292929292929293,
+      "wer": 0.20689655172413793
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.08948004836759371,
+      "wer": 0.27800829875518673
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.09153643190582504,
+      "wer": 0.33753501400560226
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.25806451612903225,
+      "wer": 0.6
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.1254180602006689,
+      "wer": 0.42391304347826086
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.08279483037156704,
+      "wer": 0.29732225300092335
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.08121827411167512,
+      "wer": 0.23809523809523808
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.0528169014084507,
+      "wer": 0.2631578947368421
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.10256410256410256,
+      "wer": 0.39285714285714285
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.10184287099903007,
+      "wer": 0.38
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.07365577872813696,
+      "wer": 0.27555435370470527
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.04411764705882353,
+      "wer": 0.2222222222222222
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.10897435897435898,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.09757942511346444,
+      "wer": 0.3626943005181347
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:16a36026c71be05e0aa9230c4d782568b6a97df01ce462227befe16150121be0
+size 137573077

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:ea9b0f670c6ea27c45e144f167fbc8fa5148068d14cb1ddf276016bea6ec2df2
+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/tmpyoece5xg", "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}