File size: 12,985 Bytes
8961f1e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d3f6416
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8961f1e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
---
language: pt
datasets:
- common_voice 
- mls
- cetuc
- lapsbm
- voxforge
- tedx
- sid
metrics:
- wer
tags:
- audio
- speech
- wav2vec2
- pt
- portuguese-speech-corpus
- automatic-speech-recognition
- speech
- PyTorch
- hf-asr-leaderboard
model-index:
- name: bp500-base10k_voxpopuli
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice 
      type: common_voice
      args: pt
    metrics:
    - name: Test WER
      type: wer
      value: 24.9
license: apache-2.0
---

# bp500-base10k_voxpopuli: Wav2vec 2.0 with Brazilian Portuguese (BP) Dataset

This is a the demonstration of a fine-tuned Wav2vec model for Brazilian Portuguese using the following datasets:

- [CETUC](http://www02.smt.ufrj.br/~igor.quintanilha/alcaim.tar.gz): contains approximately 145 hours of Brazilian Portuguese speech distributed among 50 male and 50 female speakers, each pronouncing approximately 1,000 phonetically balanced sentences selected from the [CETEN-Folha](https://www.linguateca.pt/cetenfolha/) corpus.
- [Common Voice 7.0](https://commonvoice.mozilla.org/pt):  is a project proposed by Mozilla Foundation with the goal to create a wide open dataset in different languages. In this project, volunteers donate and validate speech using the [oficial site](https://commonvoice.mozilla.org/pt). 
- [Lapsbm](https://github.com/falabrasil/gitlab-resources): "Falabrasil - UFPA" is a dataset used by the Fala Brasil group to benchmark ASR systems in Brazilian Portuguese. Contains 35 speakers (10 females), each one pronouncing 20 unique sentences, totalling  700 utterances in Brazilian Portuguese. The audios were recorded in 22.05 kHz without environment control.
- [Multilingual Librispeech (MLS)](https://arxiv.org/abs/2012.03411): a massive dataset available in many languages. The MLS is based on audiobook recordings in public domain like [LibriVox](https://librivox.org/). The dataset contains a total of 6k hours of transcribed data in many languages. The set in Portuguese [used in this work](http://www.openslr.org/94/) (mostly Brazilian variant) has approximately 284 hours of speech, obtained from 55 audiobooks read by 62 speakers.
- [Multilingual TEDx](http://www.openslr.org/100): a collection of audio recordings from TEDx talks in 8 source languages. The Portuguese set (mostly Brazilian Portuguese variant) contains 164 hours of transcribed speech. 
- [Sidney](https://igormq.github.io/datasets/) (SID): contains 5,777 utterances recorded by 72 speakers (20 women) from 17 to 59 years old with fields such as place of birth, age, gender, education, and occupation;
- [VoxForge](http://www.voxforge.org/): is a project with the goal to build open datasets for acoustic models. The corpus contains approximately 100 speakers and 4,130 utterances of Brazilian Portuguese, with sample rates varying from 16kHz to 44.1kHz.

These datasets were combined to build a larger Brazilian Portuguese dataset. All data was used for training except Common Voice dev/test sets, that were used for validation/test respectively. We also made test sets for all the gathered datasets.

| Dataset                        |  Train | Valid |  Test |
|--------------------------------|-------:|------:|------:|
| CETUC                          | 94.0h  |    -- |  5.4h |
| Common Voice                   | 37.8h  |  8.9h |  9.5h |
| LaPS BM                        | 0.8h   |    -- |  0.1h |
| MLS                            | 161.0h |    -- |  3.7h |
| Multilingual TEDx (Portuguese) | 148.9h |    -- |  1.8h |
| SID                            | 7.2h   |    -- |  1.0h |
| VoxForge                       | 3.9h   |    -- |  0.1h |
| Total                          | 453.6h |  8.9h | 21.6h |

The original model was fine-tuned using [fairseq](https://github.com/pytorch/fairseq). This notebook uses a converted version of the original one. The link to the original fairseq model is available [here](https://drive.google.com/file/d/19kkENi8uvczmw9OLSdqnjvKqBE53cl_W/view?usp=sharing).

#### Summary

|  | CETUC          | CV             | LaPS           | MLS            | SID            | TEDx           | VF             | AVG            |
|----------------------|---------------|----------------|----------------|----------------|----------------|----------------|----------------|----------------|
| bp\_500-base10k_voxpopuli (demonstration below)              | 0.120 | 0.249 | 0.039 | 0.227 | 0.169 | 0.349 | 0.116 | 0.181 |
| bp\_500-base10k_voxpopuli + 4-gram (demonstration below)       |  0.074 | 0.174 | 0.032 | 0.182 | 0.181 | 0.349 | 0.111 | 0.157 |

#### Transcription examples

| Text                                                                                                       | Transcription                                                                                                |
|------------------------------------------------------------------------------------------------------------|--------------------------------------------------------------------------------------------------------------|
|suco de uva e água misturam bem|suco **deúva** e água **misturão** bem|
|culpa do dinheiro|**cupa** do dinheiro|
|eu amo shooters call of duty é o meu favorito|eu **omo** **shúters cofedete** é meu favorito|
|você pode explicar por que isso acontece|você pode explicar *por* que isso **ontece**|
|no futuro você desejará ter começado a investir hoje|no futuro você desejará **a** ter começado a investir hoje|


## Demonstration


```python
MODEL_NAME = "lgris/bp500-base10k_voxpopuli" 
```

### Imports and dependencies


```python
%%capture
!pip install torch==1.8.2+cu111 torchvision==0.9.2+cu111 torchaudio===0.8.2 -f https://download.pytorch.org/whl/lts/1.8/torch_lts.html
!pip install datasets
!pip install jiwer
!pip install transformers
!pip install soundfile
!pip install pyctcdecode
!pip install https://github.com/kpu/kenlm/archive/master.zip
```


```python
import jiwer
import torchaudio
from datasets import load_dataset, load_metric
from transformers import (
    Wav2Vec2ForCTC,
    Wav2Vec2Processor,
)
from pyctcdecode import build_ctcdecoder
import torch
import re
import sys
```

### Helpers


```python
chars_to_ignore_regex = '[\,\?\.\!\;\:\"]'  # noqa: W605

def map_to_array(batch):
    speech, _ = torchaudio.load(batch["path"])
    batch["speech"] = speech.squeeze(0).numpy() 
    batch["sampling_rate"] = 16_000 
    batch["sentence"] = re.sub(chars_to_ignore_regex, '', batch["sentence"]).lower().replace("’", "'")
    batch["target"] = batch["sentence"]
    return batch
```


```python
def calc_metrics(truths, hypos):
    wers = []
    mers = []
    wils = []
    for t, h in zip(truths, hypos):
        try:
            wers.append(jiwer.wer(t, h))
            mers.append(jiwer.mer(t, h))
            wils.append(jiwer.wil(t, h))
        except: # Empty string?
            pass
    wer = sum(wers)/len(wers)
    mer = sum(mers)/len(mers)
    wil = sum(wils)/len(wils)
    return wer, mer, wil
```


```python
def load_data(dataset):
    data_files = {'test': f'{dataset}/test.csv'}
    dataset = load_dataset('csv', data_files=data_files)["test"]
    return dataset.map(map_to_array)
```

### Model


```python
class STT:

    def __init__(self, 
                 model_name, 
                 device='cuda' if torch.cuda.is_available() else 'cpu', 
                 lm=None):
        self.model_name = model_name
        self.model = Wav2Vec2ForCTC.from_pretrained(model_name).to(device)
        self.processor = Wav2Vec2Processor.from_pretrained(model_name)
        self.vocab_dict = self.processor.tokenizer.get_vocab()
        self.sorted_dict = {
            k.lower(): v for k, v in sorted(self.vocab_dict.items(), 
                                            key=lambda item: item[1])
        }
        self.device = device
        self.lm = lm
        if self.lm:            
            self.lm_decoder = build_ctcdecoder(
                list(self.sorted_dict.keys()),
                self.lm
            )

    def batch_predict(self, batch):
        features = self.processor(batch["speech"], 
                                  sampling_rate=batch["sampling_rate"][0], 
                                  padding=True, 
                                  return_tensors="pt")
        input_values = features.input_values.to(self.device)
        with torch.no_grad():
            logits = self.model(input_values).logits
        if self.lm:
            logits = logits.cpu().numpy()
            batch["predicted"] = []
            for sample_logits in logits:
                batch["predicted"].append(self.lm_decoder.decode(sample_logits))
        else:
            pred_ids = torch.argmax(logits, dim=-1)
            batch["predicted"] = self.processor.batch_decode(pred_ids)
        return batch
```

### Download datasets


```python
%%capture
!gdown --id 1HFECzIizf-bmkQRLiQD0QVqcGtOG5upI
!mkdir bp_dataset
!unzip bp_dataset -d bp_dataset/
```


```python
%cd bp_dataset 
```

    /content/bp_dataset


### Tests 


```python
stt = STT(MODEL_NAME)
```

#### CETUC


```python
ds = load_data('cetuc_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("CETUC WER:", wer)
```
    CETUC WER: 0.12096759949218888


#### Common Voice


```python
ds = load_data('commonvoice_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("CV WER:", wer)
```
    CV WER: 0.24977003159495725


#### LaPS


```python
ds = load_data('lapsbm_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("Laps WER:", wer)
```
    Laps WER: 0.039769570707070705


#### MLS


```python
ds = load_data('mls_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("MLS WER:", wer)
```
    MLS WER: 0.2269637077788063


#### SID


```python
ds = load_data('sid_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("Sid WER:", wer)
```
    Sid WER: 0.1691680138494731


#### TEDx


```python
ds = load_data('tedx_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("TEDx WER:", wer)
```
    TEDx WER: 0.34908555859018014


#### VoxForge


```python
ds = load_data('voxforge_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("VoxForge WER:", wer)
```
    VoxForge WER: 0.11649350649350651


### Tests with LM


```python
!rm -rf ~/.cache
!gdown --id 1GJIKseP5ZkTbllQVgOL98R4yYAcIySFP  # trained with wikipedia
stt = STT(MODEL_NAME, lm='pt-BR-wiki.word.4-gram.arpa')
# !gdown --id 1dLFldy7eguPtyJj5OAlI4Emnx0BpFywg  # trained with bp
# stt = STT(MODEL_NAME, lm='pt-BR.word.4-gram.arpa')
```

### Cetuc


```python
ds = load_data('cetuc_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("CETUC WER:", wer)
```
    CETUC WER: 0.07499558425787961


#### Common Voice


```python
ds = load_data('commonvoice_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("CV WER:", wer)
```
    CV WER: 0.17442648452610307


#### LaPS


```python
ds = load_data('lapsbm_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("Laps WER:", wer)
```
    Laps WER: 0.032774621212121206


#### MLS


```python
ds = load_data('mls_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("MLS WER:", wer)
```
    MLS WER: 0.18213620321569274


#### SID


```python
ds = load_data('sid_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("Sid WER:", wer)
```
    Sid WER: 0.18102544972868206


#### TEDx


```python
ds = load_data('tedx_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("TEDx WER:", wer)
```
    TEDx WER: 0.3491402028105601


#### VoxForge


```python
ds = load_data('voxforge_dataset')
result = ds.map(stt.batch_predict, batched=True, batch_size=8) 
wer, mer, wil = calc_metrics(result["sentence"], result["predicted"])
print("VoxForge WER:", wer)
```
    VoxForge WER: 0.11189529220779222