joaoalvarenga commited on
Commit
89a63da
0 Parent(s):

first version

Browse files
.gitattributes ADDED
@@ -0,0 +1 @@
 
 
1
+ pytorch_model.bin filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: pt
3
+ datasets:
4
+ - common_voice
5
+ metrics:
6
+ - wer
7
+ tags:
8
+ - audio
9
+ - speech
10
+ - wav2vec2
11
+ - pt
12
+ - apache-2.0
13
+ - portuguese-speech-corpus
14
+ - automatic-speech-recognition
15
+ - speech
16
+ - xlsr-fine-tuning-week
17
+ - PyTorch
18
+ license: apache-2.0
19
+ model-index:
20
+ - name: JoaoAlvarenga XLSR Wav2Vec2 Large 53 Portuguese
21
+ results:
22
+ - task:
23
+ name: Speech Recognition
24
+ type: automatic-speech-recognition
25
+ dataset:
26
+ name: Common Voice
27
+ type: common_voice
28
+ args: es
29
+ metrics:
30
+ - name: Test WER
31
+ type: wer
32
+ value: 15.734702%
33
+ ---
34
+
35
+
36
+ # Wav2Vec2-Large-XLSR-53-Spanish
37
+
38
+ Fine-tuned [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on Spanish using the [Common Voice](https://huggingface.co/datasets/common_voice) dataset.
39
+
40
+ ## Usage
41
+
42
+ The model can be used directly (without a language model) as follows:
43
+
44
+ ```python
45
+ import torch
46
+ import torchaudio
47
+ from datasets import load_dataset
48
+ from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
49
+
50
+ test_dataset = load_dataset("common_voice", "pt", split="test[:2%]")
51
+
52
+ processor = Wav2Vec2Processor.from_pretrained("joorock12/wav2vec2-large-xlsr-53-portuguese")
53
+ model = Wav2Vec2ForCTC.from_pretrained("joorock12/wav2vec2-large-xlsr-53-portuguese")
54
+
55
+ resampler = torchaudio.transforms.Resample(48_000, 16_000)
56
+
57
+ # Preprocessing the datasets.
58
+ # We need to read the aduio files as arrays
59
+ def speech_file_to_array_fn(batch):
60
+ tspeech_array, sampling_rate = torchaudio.load(batch["path"])
61
+ \tbatch["speech"] = resampler(speech_array).squeeze().numpy()
62
+ \treturn batch
63
+
64
+ test_dataset = test_dataset.map(speech_file_to_array_fn)
65
+ inputs = processor(test_dataset["speech"][:2], sampling_rate=16_000, return_tensors="pt", padding=True)
66
+
67
+ with torch.no_grad():
68
+ \tlogits = model(inputs.input_values, attention_mask=inputs.attention_mask).logits
69
+
70
+ predicted_ids = torch.argmax(logits, dim=-1)
71
+
72
+ print("Prediction:", processor.batch_decode(predicted_ids))
73
+ print("Reference:", test_dataset["sentence"][:2])
74
+ ```
75
+
76
+
77
+ ## Evaluation
78
+
79
+ The model can be evaluated as follows on the Portuguese test data of Common Voice.
80
+
81
+
82
+ ```python
83
+ import torch
84
+ import torchaudio
85
+ from datasets import load_dataset, load_metric
86
+ from transformers import Wav2Vec2ForCTC, Wav2Vec2Processor
87
+ import re
88
+
89
+ test_dataset = load_dataset("common_voice", "pt", split="test")
90
+ wer = load_metric("wer")
91
+
92
+ processor = Wav2Vec2Processor.from_pretrained("joorock12/wav2vec2-large-xlsr-53-portuguese")
93
+ model = Wav2Vec2ForCTC.from_pretrained("joorock12/wav2vec2-large-xlsr-53-portuguese")
94
+ model.to("cuda")
95
+
96
+ chars_to_ignore_regex = '[\\\\\\\\\\\\\\\\,\\\\\\\\\\\\\\\\?\\\\\\\\\\\\\\\\.\\\\\\\\\\\\\\\\!\\\\\\\\\\\\\\\\-\\\\\\\\\\\\\\\\;\\\\\\\\\\\\\\\\:\\\\\\\\\\\\\\\\"\\\\\\\\\\\\\\\\“]' # TODO: adapt this list to include all special characters you removed from the data
97
+ resampler = torchaudio.transforms.Resample(48_000, 16_000)
98
+
99
+ # Preprocessing the datasets.
100
+ # We need to read the aduio files as arrays
101
+ def speech_file_to_array_fn(batch):
102
+ \tbatch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower()
103
+ \tspeech_array, sampling_rate = torchaudio.load(batch["path"])
104
+ \tbatch["speech"] = resampler(speech_array).squeeze().numpy()
105
+ \treturn batch
106
+
107
+ test_dataset = test_dataset.map(speech_file_to_array_fn)
108
+
109
+ # Preprocessing the datasets.
110
+ # We need to read the aduio files as arrays
111
+ def evaluate(batch):
112
+ \\\\\\\\tinputs = processor(batch["speech"], sampling_rate=16_000, return_tensors="pt", padding=True)
113
+
114
+ \\\\\\\\twith torch.no_grad():
115
+ \\\\\\\\t\\\\\\\\tlogits = model(inputs.input_values.to("cuda"), attention_mask=inputs.attention_mask.to("cuda")).logits
116
+
117
+ pred_ids = torch.argmax(logits, dim=-1)
118
+ \\\\\\\\tbatch["pred_strings"] = processor.batch_decode(pred_ids)
119
+ \\\\\\\\treturn batch
120
+
121
+ result = test_dataset.map(evaluate, batched=True, batch_size=8)
122
+
123
+ print("WER: {:2f}".format(100 * wer.compute(predictions=result["pred_strings"], references=result["sentence"])))
124
+ ```
125
+
126
+ **Test Result (wer) **: 15.734702%
127
+
128
+
129
+ ## Training
130
+
131
+ The Common Voice `train`, `validation` datasets were used for training.
132
+
133
+ The script used for training can be found at: https://github.com/joaoalvarenga/wav2vec2-large-xlsr-53-spanish/blob/main/fine-tuning.py
added_tokens.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"<s>": 44, "</s>": 45}
config.json ADDED
@@ -0,0 +1,76 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "/home/joaoalvarenga/wav2vec2-large-xlsr-portuguese/checkpoint-2200",
3
+ "activation_dropout": 0.0,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "Wav2Vec2ForCTC"
7
+ ],
8
+ "attention_dropout": 0.1,
9
+ "bos_token_id": 1,
10
+ "conv_bias": true,
11
+ "conv_dim": [
12
+ 512,
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512
19
+ ],
20
+ "conv_kernel": [
21
+ 10,
22
+ 3,
23
+ 3,
24
+ 3,
25
+ 3,
26
+ 2,
27
+ 2
28
+ ],
29
+ "conv_stride": [
30
+ 5,
31
+ 2,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2,
36
+ 2
37
+ ],
38
+ "ctc_loss_reduction": "mean",
39
+ "ctc_zero_infinity": false,
40
+ "do_stable_layer_norm": true,
41
+ "eos_token_id": 2,
42
+ "feat_extract_activation": "gelu",
43
+ "feat_extract_dropout": 0.0,
44
+ "feat_extract_norm": "layer",
45
+ "feat_proj_dropout": 0.0,
46
+ "final_dropout": 0.0,
47
+ "gradient_checkpointing": true,
48
+ "hidden_act": "gelu",
49
+ "hidden_dropout": 0.1,
50
+ "hidden_size": 1024,
51
+ "initializer_range": 0.02,
52
+ "intermediate_size": 4096,
53
+ "layer_norm_eps": 1e-05,
54
+ "layerdrop": 0.1,
55
+ "mask_channel_length": 10,
56
+ "mask_channel_min_space": 1,
57
+ "mask_channel_other": 0.0,
58
+ "mask_channel_prob": 0.0,
59
+ "mask_channel_selection": "static",
60
+ "mask_feature_length": 10,
61
+ "mask_feature_prob": 0.0,
62
+ "mask_time_length": 10,
63
+ "mask_time_min_space": 1,
64
+ "mask_time_other": 0.0,
65
+ "mask_time_prob": 0.05,
66
+ "mask_time_selection": "static",
67
+ "model_type": "wav2vec2",
68
+ "num_attention_heads": 16,
69
+ "num_conv_pos_embedding_groups": 16,
70
+ "num_conv_pos_embeddings": 128,
71
+ "num_feat_extract_layers": 7,
72
+ "num_hidden_layers": 24,
73
+ "pad_token_id": 43,
74
+ "transformers_version": "4.4.0",
75
+ "vocab_size": 44
76
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_size": 1,
4
+ "padding_side": "right",
5
+ "padding_value": 0.0,
6
+ "return_attention_mask": true,
7
+ "sampling_rate": 16000
8
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9ef6e942cc9abb7eff97c2c23ffbee6501bbc5eeebffc13b379435840d2e07fd
3
+ size 1262103207
special_tokens_map.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}
tokenizer_config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "special_tokens_map_file": "/home/joaoalvarenga/wav2vec2-large-xlsr-portuguese/special_tokens_map.json", "tokenizer_file": null, "name_or_path": "/home/joaoalvarenga/wav2vec2-large-xlsr-portuguese"}
vocab.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"g": 0, "â": 1, "o": 2, "i": 3, "ñ": 4, "d": 5, "y": 6, "c": 7, "r": 8, "k": 9, "s": 10, "ç": 11, "l": 12, "'": 13, "t": 14, "b": 15, "e": 16, "í": 17, "z": 18, "ú": 19, "j": 20, "ó": 21, "á": 22, "w": 23, "ê": 24, "v": 26, "ü": 27, "à": 28, "ô": 29, "õ": 30, "m": 31, "ã": 32, "p": 33, "h": 34, "é": 35, "x": 36, "a": 37, "q": 38, "u": 39, "n": 40, "f": 41, "|": 25, "[UNK]": 42, "[PAD]": 43}