initial commit

.gitattributes

@@ -1,35 +1,27 @@
 *.7z filter=lfs diff=lfs merge=lfs -text
 *.arrow filter=lfs diff=lfs merge=lfs -text
 *.bin filter=lfs diff=lfs merge=lfs -text
+*.bin.* filter=lfs diff=lfs merge=lfs -text
 *.bz2 filter=lfs diff=lfs merge=lfs -text
-*.ckpt filter=lfs diff=lfs merge=lfs -text
 *.ftz filter=lfs diff=lfs merge=lfs -text
 *.gz filter=lfs diff=lfs merge=lfs -text
 *.h5 filter=lfs diff=lfs merge=lfs -text
 *.joblib filter=lfs diff=lfs merge=lfs -text
 *.lfs.* filter=lfs diff=lfs merge=lfs -text
-*.mlmodel filter=lfs diff=lfs merge=lfs -text
 *.model filter=lfs diff=lfs merge=lfs -text
 *.msgpack filter=lfs diff=lfs merge=lfs -text
-*.npy filter=lfs diff=lfs merge=lfs -text
-*.npz filter=lfs diff=lfs merge=lfs -text
 *.onnx filter=lfs diff=lfs merge=lfs -text
 *.ot filter=lfs diff=lfs merge=lfs -text
 *.parquet filter=lfs diff=lfs merge=lfs -text
 *.pb filter=lfs diff=lfs merge=lfs -text
-*.pickle filter=lfs diff=lfs merge=lfs -text
-*.pkl filter=lfs diff=lfs merge=lfs -text
 *.pt filter=lfs diff=lfs merge=lfs -text
 *.pth filter=lfs diff=lfs merge=lfs -text
 *.rar filter=lfs diff=lfs merge=lfs -text
-*.safetensors filter=lfs diff=lfs merge=lfs -text
-saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text 
 *.tar.* filter=lfs diff=lfs merge=lfs -text
-*.tar filter=lfs diff=lfs merge=lfs -text
 *.tflite filter=lfs diff=lfs merge=lfs -text
 *.tgz filter=lfs diff=lfs merge=lfs -text
-*.wasm filter=lfs diff=lfs merge=lfs -text
 *.xz filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
-*.zst filter=lfs diff=lfs merge=lfs -text
+*.zstandard filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,63 @@
+---
+language: multilingual
+datasets:
+- common_voice
+tags:
+- speech
+- audio
+- automatic-speech-recognition
+- phoneme-recognition
+widget:
+- example_title: Librispeech sample 1
+  src: https://cdn-media.huggingface.co/speech_samples/sample1.flac
+- example_title: Librispeech sample 2
+  src: https://cdn-media.huggingface.co/speech_samples/sample2.flac
+license: apache-2.0
+---
+
+# Wav2Vec2-Large-LV60 finetuned on multi-lingual Common Voice
+
+This checkpoint leverages the pretrained checkpoint [wav2vec2-large-lv60](https://huggingface.co/facebook/wav2vec2-large-lv60) 
+and is fine-tuned on [CommonVoice](https://huggingface.co/datasets/common_voice) to recognize phonetic labels in multiple languages.
+
+When using the model make sure that your speech input is sampled at 16kHz. 
+Note that the model outputs a string of phonetic labels. A dictionary mapping phonetic labels to words 
+has to be used to map the phonetic output labels to output words.
+
+[Paper: Simple and Effective Zero-shot Cross-lingual Phoneme Recognition](https://arxiv.org/abs/2109.11680)
+
+Authors: Qiantong Xu, Alexei Baevski, Michael Auli
+
+**Abstract**
+Recent progress in self-training, self-supervised pretraining and unsupervised learning enabled well performing speech recognition systems without any labeled data. However, in many cases there is labeled data available for related languages which is not utilized by these methods. This paper extends previous work on zero-shot cross-lingual transfer learning by fine-tuning a multilingually pretrained wav2vec 2.0 model to transcribe unseen languages. This is done by mapping phonemes of the training languages to the target language using articulatory features. Experiments show that this simple method significantly outperforms prior work which introduced task-specific architectures and used only part of a monolingually pretrained model.
+
+The original model can be found under https://github.com/pytorch/fairseq/tree/master/examples/wav2vec#wav2vec-20.
+
+# Usage
+
+To transcribe audio files the model can be used as a standalone acoustic model as follows:
+
+```python
+ from transformers import Wav2Vec2Processor, Wav2Vec2ForCTC
+ from datasets import load_dataset
+ import torch
+ 
+ # load model and processor
+ processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-lv-60-espeak-cv-ft")
+ model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-lv-60-espeak-cv-ft")
+     
+ # load dummy dataset and read soundfiles
+ ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
+ 
+ # tokenize
+ input_values = processor(ds[0]["audio"]["array"], return_tensors="pt").input_values
+ 
+ # retrieve logits
+ with torch.no_grad():
+   logits = model(input_values).logits
+ 
+ # take argmax and decode
+ predicted_ids = torch.argmax(logits, dim=-1)
+ transcription = processor.batch_decode(predicted_ids)
+ # => should give ['m ɪ s t ɚ k w ɪ l t ɚ ɹ ɪ z ð ɪ ɐ p ɑː s əl ʌ v ð ə m ɪ d əl k l æ s ᵻ z æ n d w iː ɑːɹ ɡ l æ d t ə w ɛ l k ə m h ɪ z ɡ ɑː s p əl']
+ ```

config.json

@@ -0,0 +1,86 @@
+{
+  "activation_dropout": 0.0,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.1,
+  "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": "sum",
+  "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": 0.1,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "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_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.075,
+  "mask_time_selection": "static",
+  "model_type": "wav2vec2",
+  "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,
+  "pad_token_id": 0,
+  "proj_codevector_dim": 768,
+  "torch_dtype": "float32",
+  "transformers_version": "4.13.0.dev0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 392
+}

preprocessor_config.json

@@ -0,0 +1,8 @@
+{
+  "do_normalize": true,
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3173bde9e9ce490fa0f989e413c42f25bc1820c020adc1e6b9b87025b3cfcc5e
+size 1263535127

requirements.txt

@@ -0,0 +1,7 @@
+numpy
+torch
+torchaudio
+transformers
+phonemizer
+flask
+flask_cors

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, "return_attention_mask": false, "do_normalize": true}

vocab.json

@@ -0,0 +1 @@
+{"<s>": 1, "<pad>": 0, "</s>": 2, "<unk>": 3, "n": 4, "s": 5, "t": 6, "ə": 7, "l": 8, "a": 9, "i": 10, "k": 11, "d": 12, "m": 13, "ɛ": 14, "ɾ": 15, "e": 16, "ɪ": 17, "p": 18, "o": 19, "ɐ": 20, "z": 21, "ð": 22, "f": 23, "j": 24, "v": 25, "b": 26, "ɹ": 27, "ʁ": 28, "ʊ": 29, "iː": 30, "r": 31, "w": 32, "ʌ": 33, "u": 34, "ɡ": 35, "æ": 36, "aɪ": 37, "ʃ": 38, "h": 39, "ɔ": 40, "ɑː": 41, "ŋ": 42, "ɚ": 43, "eɪ": 44, "β": 45, "uː": 46, "y": 47, "ɑ̃": 48, "oʊ": 49, "ᵻ": 50, "eː": 51, "θ": 52, "aʊ": 53, "ts": 54, "oː": 55, "ɔ̃": 56, "ɣ": 57, "ɜ": 58, "ɑ": 59, "dʒ": 60, "əl": 61, "x": 62, "ɜː": 63, "ç": 64, "ʒ": 65, "tʃ": 66, "ɔː": 67, "ɑːɹ": 68, "ɛ̃": 69, "ʎ": 70, "ɔːɹ": 71, "ʋ": 72, "aː": 73, "ɕ": 74, "œ": 75, "ø": 76, "oːɹ": 77, "ɲ": 78, "yː": 79, "ʔ": 80, "iə": 81, "i5": 82, "s.": 83, "tɕ": 84, "??": 85, "nʲ": 86, "ɛː": 87, "œ̃": 88, "ɭ": 89, "ɔø": 90, "ʑ": 91, "tʲ": 92, "ɨ": 93, "ɛɹ": 94, "ts.": 95, "rʲ": 96, "ɪɹ": 97, "ɭʲ": 98, "i.5": 99, "ɔɪ": 100, "q": 101, "sʲ": 102, "u5": 103, "ʊɹ": 104, "iɜ": 105, "a5": 106, "iɛ5": 107, "øː": 108, "ʕ": 109, "ja": 110, "əɜ": 111, "th": 112, "ɑ5": 113, "oɪ": 114, "dʲ": 115, "ə5": 116, "tɕh": 117, "ts.h": 118, "mʲ": 119, "ɯ": 120, "dʑ": 121, "vʲ": 122, "e̞": 123, "tʃʲ": 124, "ei5": 125, "o5": 126, "onɡ5": 127, "ɑu5": 128, "iɑ5": 129, "ai5": 130, "aɪɚ": 131, "kh": 132, "ə1": 133, "ʐ": 134, "i2": 135, "ʉ": 136, "ħ": 137, "t[": 138, "aɪə": 139, "ʲ": 140, "ju": 141, "ə2": 142, "u2": 143, "oɜ": 144, "pː": 145, "iɛɜ": 146, "ou5": 147, "y5": 148, "uɜ": 149, "tː": 150, "uo5": 151, "d[": 152, "uoɜ": 153, "tsh": 154, "ɑɜ": 155, "ɵ": 156, "i̪5": 157, "uei5": 158, "ɟ": 159, "aɜ": 160, "ɑɨ": 161, "i.ɜ": 162, "eʊ": 163, "o2": 164, "ɐ̃": 165, "ä": 166, "pʲ": 167, "kʲ": 168, "n̩": 169, "ɒ": 170, "ph": 171, "ɑu2": 172, "uɨ": 173, "əɪ": 174, "ɫ": 175, "ɬ": 176, "yɜ": 177, "bʲ": 178, "ɑ2": 179, "s̪": 180, "aiɜ": 181, "χ": 182, "ɐ̃ʊ̃": 183, "1": 184, "ə4": 185, "yæɜ": 186, "a2": 187, "ɨː": 188, "t̪": 189, "iouɜ": 190, "ũ": 191, "onɡɜ": 192, "aɨ": 193, "iɛ2": 194, "ɔɨ": 195, "ɑuɜ": 196, "o̞": 197, "ei2": 198, "iou2": 199, "c": 200, "kː": 201, "y2": 202, "ɖ": 203, "oe": 204, "dˤ": 205, "yɛɜ": 206, "əʊ": 207, "S": 208, "ɡʲ": 209, "onɡ2": 210, "u\"": 211, "eiɜ": 212, "ʈ": 213, "ɯᵝ": 214, "iou5": 215, "dZ": 216, "r̝̊": 217, "i.2": 218, "tS": 219, "s^": 220, "ʝ": 221, "yə5": 222, "iɑɜ": 223, "uə5": 224, "pf": 225, "ɨu": 226, "iɑ2": 227, "ou2": 228, "ər2": 229, "fʲ": 230, "ai2": 231, "r̝": 232, "uəɜ": 233, "ɳ": 234, "əɨ": 235, "ua5": 236, "uɪ": 237, "ɽ": 238, "bː": 239, "yu5": 240, "uo2": 241, "yɛ5": 242, "l̩": 243, "ɻ": 244, "ərɜ": 245, "ʂ": 246, "i̪2": 247, "ouɜ": 248, "uaɜ": 249, "a.": 250, "a.ː": 251, "yæ5": 252, "dː": 253, "r̩": 254, "ee": 255, "ɪu": 256, "ər5": 257, "i̪ɜ": 258, "æi": 259, "u:": 260, "i.ː": 261, "t^": 262, "o1": 263, "ɪ^": 264, "ai": 265, "ueiɜ": 266, "æː": 267, "ɛɪ": 268, "eə": 269, "i.": 270, "ɴ": 271, "ie": 272, "ua2": 273, "ɑ1": 274, "o4": 275, "tʃː": 276, "o:": 277, "ɑ:": 278, "u1": 279, "N": 280, "i̪1": 281, "au": 282, "yæ2": 283, "u.": 284, "qː": 285, "yəɜ": 286, "y:": 287, "kʰ": 288, "tʃʰ": 289, "iʊ": 290, "sx": 291, "õ": 292, "uo": 293, "tʰ": 294, "uai5": 295, "bʰ": 296, "u.ː": 297, "uə2": 298, "ʊə": 299, "d^": 300, "s̪ː": 301, "yiɜ": 302, "dʰ": 303, "r.": 304, "oe:": 305, "i1": 306, "ɟː": 307, "yu2": 308, "nʲʲ": 309, "i̪4": 310, "uei2": 311, "tsʲ": 312, "ɸ": 313, "ĩ": 314, "ɑ4": 315, "t̪ː": 316, "eɑ": 317, "u4": 318, "e:": 319, "tsː": 320, "ʈʰ": 321, "ɡʰ": 322, "ɯɯ": 323, "dʒʲ": 324, "ʂʲ": 325, "X": 326, "ɵː": 327, "uaiɜ": 328, "tɕʲ": 329, "ã": 330, "t^ː": 331, "ẽː": 332, "yɛ2": 333, "cː": 334, "i.1": 335, "ɛʊ": 336, "dˤdˤ": 337, "dʒː": 338, "i4": 339, "ɡː": 340, "yi": 341, "ɕʲ": 342, "ɟʰ": 343, "pʰ": 344, "dʑʲ": 345, "yuɜ": 346, "ua1": 347, "ua4": 348, "æiː": 349, "ɐɐ": 350, "ui": 351, "iou1": 352, "ʊː": 353, "a1": 354, "iou4": 355, "cʰ": 356, "iɛ1": 357, "yə2": 358, "ɖʰ": 359, "ẽ": 360, "ʒʲ": 361, "ää": 362, "ər4": 363, "iːː": 364, "ɪː": 365, "iɑ1": 366, "ər1": 367, "œː": 368, "øi": 369, "ɪuː": 370, "cʰcʰ": 371, "əː1": 372, "iː1": 373, "ũ": 374, "kʰː": 375, "o̞o̞": 376, "xʲ": 377, "ou1": 378, "iɛ4": 379, "e̞e̞": 380, "y1": 381, "dzː": 382, "dʲʲ": 383, "dʰː": 384, "ɯᵝɯᵝ": 385, "lː": 386, "uo1": 387, "i.4": 388, "i:": 389, "yɛ5ʲ": 390, "a4": 391}