File size: 3,962 Bytes
8d88236
 
 
 
 
101761d
 
fd6acee
8d88236
 
0b362f5
8d88236
0b362f5
8d88236
0b362f5
233a80e
 
0b362f5
8d88236
0b362f5
8d88236
0b362f5
8d88236
5dadc39
8d88236
0b362f5
8d88236
0b362f5
 
8d88236
0b362f5
5dadc39
0b362f5
 
e1e8762
8d88236
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
266740b
 
671b8e7
266740b
3eb23cb
266740b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
671b8e7
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
---
language: en
datasets:
- librispeech_asr
tags:
- audio
- automatic-speech-recognition
pipeline_tag: automatic-speech-recognition
license: apache-2.0
---

# Wav2Vec2-Base-960h

[Facebook's Wav2Vec2](https://ai.facebook.com/blog/wav2vec-20-learning-the-structure-of-speech-from-raw-audio/)

The base model pretrained and fine-tuned on 960 hours of Librispeech on 16kHz sampled speech audio. When using the model
make sure that your speech input is also sampled at 16Khz.

[Paper](https://arxiv.org/abs/2006.11477)

Authors: Alexei Baevski, Henry Zhou, Abdelrahman Mohamed, Michael Auli

**Abstract**

We show for the first time that learning powerful representations from speech audio alone followed by fine-tuning on transcribed speech can outperform the best semi-supervised methods while being conceptually simpler. wav2vec 2.0 masks the speech input in the latent space and solves a contrastive task defined over a quantization of the latent representations which are jointly learned. Experiments using all labeled data of Librispeech achieve 1.8/3.3 WER on the clean/other test sets. When lowering the amount of labeled data to one hour, wav2vec 2.0 outperforms the previous state of the art on the 100 hour subset while using 100 times less labeled data. Using just ten minutes of labeled data and pre-training on 53k hours of unlabeled data still achieves 4.8/8.2 WER. This demonstrates the feasibility of speech recognition with limited amounts of labeled data.

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 Wav2Vec2Tokenizer, Wav2Vec2ForMaskedLM
 from datasets import load_dataset
 import soundfile as sf
 import torch
 
 # load model and tokenizer
 tokenizer = Wav2Vec2Tokenizer.from_pretrained("facebook/wav2vec2-base-960h")
 model = Wav2Vec2ForMaskedLM.from_pretrained("facebook/wav2vec2-base-960h")
 
 # define function to read in sound file
 def map_to_array(batch):
     speech, _ = sf.read(batch["file"])
     batch["speech"] = speech
     return batch
     
 # load dummy dataset and read soundfiles
 ds = load_dataset("patrickvonplaten/librispeech_asr_dummy", "clean", split="validation")
 ds = ds.map(map_to_array)
 
 # tokenize
 input_values = tokenizer(ds["speech"][:2], return_tensors="pt", padding="longest").input_values  # Batch size 1
 
 # retrieve logits
 logits = model(input_values).logits
 
 # take argmax and decode
 predicted_ids = torch.argmax(logits, dim=-1)
 transcription = tokenizer.batch_decode(predicted_ids)
 ```
 
 ## Evaluation
 
 This code snippet shows how to evaluate **facebook/wav2vec2-base-960h** on LibriSpeech's "clean" and "other" test data.
 
```python
from datasets import load_dataset
from transformers import Wav2Vec2ForMaskedLM, Wav2Vec2Tokenizer
import soundfile as sf
import torch
from jiwer import wer


librispeech_eval = load_dataset("librispeech_asr", "clean", split="test")

model = Wav2Vec2ForMaskedLM.from_pretrained("facebook/wav2vec2-base-960h").to("cuda")
tokenizer = Wav2Vec2Tokenizer.from_pretrained("facebook/wav2vec2-base-960h")

def map_to_array(batch):
    speech, _ = sf.read(batch["file"])
    batch["speech"] = speech
    return batch

librispeech_eval = librispeech_eval.map(map_to_array)

def map_to_pred(batch):
    input_values = tokenizer(batch["speech"], return_tensors="pt", padding="longest").input_values
    with torch.no_grad():
        logits = model(input_values.to("cuda")).logits

    predicted_ids = torch.argmax(logits, dim=-1)
    transcription = tokenizer.batch_decode(predicted_ids)
    batch["transcription"] = transcription
    return batch

result = librispeech_eval.map(map_to_pred, batched=True, batch_size=16, remove_columns=["speech"])

print("WER:", wer(result["text"], result["transcription"]))
```

| "clean" | "other" |
|---|---|
| 4.1 | 10.0 |