jonatasgrosman/exp_w2v2r_de_xls-r_age_teens-5_sixties-5_s905

.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_s905
+
+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": 0.04359198593692668,
+      "wer": 0.17977329563988256
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.04308514092348837,
+      "wer": 0.17840164514490764
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.04525219822670248,
+      "wer": 0.18673187860546778
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.04124435524335173,
+      "wer": 0.1669407894736842
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.04766574058942093,
+      "wer": 0.19315221436546334
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.04493002700712006,
+      "wer": 0.1870748299319728
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.055054302422723474,
+      "wer": 0.21527581329561526
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.04730831973898858,
+      "wer": 0.2073170731707317
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.04524196327475016,
+      "wer": 0.1797033041132839
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.04787312879798429,
+      "wer": 0.19811813904861475
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.04308953648054219,
+      "wer": 0.1783948047623012
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.0445615355664283,
+      "wer": 0.18066355624675998
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.07376283846872082,
+      "wer": 0.3170731707317073
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.04587867385335096,
+      "wer": 0.18583840139009558
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04357145844661006,
+      "wer": 0.17988380595755624
+    },
+    {
+      "subset_type": "accent",
+      "subset": "canada",
+      "cer": 0.08823529411764706,
+      "wer": 0.375
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.037162162162162164,
+      "wer": 0.20833333333333334
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.05405405405405406,
+      "wer": 0.19230769230769232
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.03888888888888889,
+      "wer": 0.19298245614035087
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.03367003367003367,
+      "wer": 0.1956521739130435
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.052283503868230594,
+      "wer": 0.2158894645941278
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.07506053268765134,
+      "wer": 0.2644628099173554
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.032534246575342464,
+      "wer": 0.15151515151515152
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.12530712530712532,
+      "wer": 0.4915254237288136
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.06825938566552901,
+      "wer": 0.2620320855614973
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.04915978773584906,
+      "wer": 0.19898477157360406
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.03414634146341464,
+      "wer": 0.15151515151515152
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.09715025906735751,
+      "wer": 0.3504273504273504
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.06227544910179641,
+      "wer": 0.216
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.040941503022955315,
+      "wer": 0.16926112694584522
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.07777193904361535,
+      "wer": 0.26501766784452296
+    }
+  ],
+  "test": [
+    {
+      "subset_type": "all",
+      "subset": "all",
+      "cer": 0.050034966749072804,
+      "wer": 0.19514959202175883
+    },
+    {
+      "subset_type": "age",
+      "subset": "",
+      "cer": 0.04920539669934013,
+      "wer": 0.19281799716399034
+    },
+    {
+      "subset_type": "age",
+      "subset": "twenties",
+      "cer": 0.05442631294334701,
+      "wer": 0.2130049686757399
+    },
+    {
+      "subset_type": "age",
+      "subset": "fifties",
+      "cer": 0.044460380064539264,
+      "wer": 0.19057071960297767
+    },
+    {
+      "subset_type": "age",
+      "subset": "teens",
+      "cer": 0.06594697804215319,
+      "wer": 0.23847695390781562
+    },
+    {
+      "subset_type": "age",
+      "subset": "seventies",
+      "cer": 0.06294218101976091,
+      "wer": 0.24027072758037224
+    },
+    {
+      "subset_type": "age",
+      "subset": "fourties",
+      "cer": 0.07721756129347561,
+      "wer": 0.25263157894736843
+    },
+    {
+      "subset_type": "age",
+      "subset": "eighties",
+      "cer": 0.06788990825688074,
+      "wer": 0.2602739726027397
+    },
+    {
+      "subset_type": "age",
+      "subset": "thirties",
+      "cer": 0.04897977815310175,
+      "wer": 0.1882603406326034
+    },
+    {
+      "subset_type": "age",
+      "subset": "sixties",
+      "cer": 0.06820831713952584,
+      "wer": 0.23841961852861035
+    },
+    {
+      "subset_type": "gender",
+      "subset": "",
+      "cer": 0.04929899403916357,
+      "wer": 0.19308273441929602
+    },
+    {
+      "subset_type": "gender",
+      "subset": "female",
+      "cer": 0.06926197251316825,
+      "wer": 0.2629815745393635
+    },
+    {
+      "subset_type": "gender",
+      "subset": "other",
+      "cer": 0.12123745819397994,
+      "wer": 0.26136363636363635
+    },
+    {
+      "subset_type": "gender",
+      "subset": "male",
+      "cer": 0.05306401111189585,
+      "wer": 0.20282792778399106
+    },
+    {
+      "subset_type": "accent",
+      "subset": "",
+      "cer": 0.04985080041320212,
+      "wer": 0.1945731957509536
+    },
+    {
+      "subset_type": "accent",
+      "subset": "austria",
+      "cer": 0.05600430802369413,
+      "wer": 0.2123730378578024
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_states",
+      "cer": 0.06952841596130592,
+      "wer": 0.25311203319502074
+    },
+    {
+      "subset_type": "accent",
+      "subset": "belgium",
+      "cer": 0.050761421319796954,
+      "wer": 0.3333333333333333
+    },
+    {
+      "subset_type": "accent",
+      "subset": "luxembourg",
+      "cer": 0.0528169014084507,
+      "wer": 0.21052631578947367
+    },
+    {
+      "subset_type": "accent",
+      "subset": "denmark",
+      "cer": 0.07352941176470588,
+      "wer": 0.4444444444444444
+    },
+    {
+      "subset_type": "accent",
+      "subset": "czechia",
+      "cer": 0.16531165311653118,
+      "wer": 0.49056603773584906
+    },
+    {
+      "subset_type": "accent",
+      "subset": "united_kingdom",
+      "cer": 0.10658307210031348,
+      "wer": 0.3627450980392157
+    },
+    {
+      "subset_type": "accent",
+      "subset": "france",
+      "cer": 0.07274490785645005,
+      "wer": 0.2866666666666667
+    },
+    {
+      "subset_type": "accent",
+      "subset": "slovakia",
+      "cer": 0.03535353535353535,
+      "wer": 0.13793103448275862
+    },
+    {
+      "subset_type": "accent",
+      "subset": "bulgaria",
+      "cer": 0.0,
+      "wer": 0.0
+    },
+    {
+      "subset_type": "accent",
+      "subset": "russia",
+      "cer": 0.18181818181818182,
+      "wer": 0.4888888888888889
+    },
+    {
+      "subset_type": "accent",
+      "subset": "greece",
+      "cer": 0.1346153846153846,
+      "wer": 0.25925925925925924
+    },
+    {
+      "subset_type": "accent",
+      "subset": "netherlands",
+      "cer": 0.06732223903177005,
+      "wer": 0.26424870466321243
+    },
+    {
+      "subset_type": "accent",
+      "subset": "switzerland",
+      "cer": 0.06657195047696367,
+      "wer": 0.24089635854341737
+    },
+    {
+      "subset_type": "accent",
+      "subset": "turkey",
+      "cer": 0.15053763440860216,
+      "wer": 0.4
+    },
+    {
+      "subset_type": "accent",
+      "subset": "poland",
+      "cer": 0.0802675585284281,
+      "wer": 0.2826086956521739
+    },
+    {
+      "subset_type": "accent",
+      "subset": "other",
+      "cer": 0.043478260869565216,
+      "wer": 0.2857142857142857
+    },
+    {
+      "subset_type": "accent",
+      "subset": "germany",
+      "cer": 0.04749294954122367,
+      "wer": 0.1878492879033712
+    },
+    {
+      "subset_type": "accent",
+      "subset": "hungary",
+      "cer": 0.07179487179487179,
+      "wer": 0.35714285714285715
+    }
+  ]
+}

cv7_transcriptions.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:86c7f43687c7774293037efe35c638e4bfe42f439126d2ef865e4818cd527a8a
+size 138742178

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 sha256:81873885b43f9ba78dfb6addd0e9ffd0e955dafcadc512d4d92c006821a20b02
+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/tmpmectjkiy", "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}