File size: 4,335 Bytes
55cfea2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language:
- en
- de
datasets:
- mustc
tags:
- audio
- speech-translation
- automatic-speech-recognition
license: MIT
---


# S2T-SMALL-MUSTC-EN-DE-ST

`s2t-small-mustc-en-de-st` is a Speech to Text Transformer (S2T) model trained for end-to-end Speech Translation (ST).
The S2T model was proposed in [this paper](https://arxiv.org/abs/2010.05171) and released in
[this repository](https://github.com/pytorch/fairseq/tree/master/examples/speech_to_text)


## Model description

S2T is a transformer-based seq2seq (encoder-decoder) model designed for end-to-end Automatic Speech Recognition (ASR) and Speech
Translation (ST). It uses a convolutional downsampler to reduce the length of speech inputs by 3/4th before they are
fed into the encoder. The model is trained with standard autoregressive cross-entropy loss and generates the
transcripts/translations autoregressively.

## Intended uses & limitations

This model can be used for end-to-end English speech to German text translation.
See the [model hub](https://huggingface.co/models?filter=speech_to_text) to look for other S2T checkpoints.


### How to use

As this a standard sequence to sequence transformer model, you can use the `generate` method to generate the
transcripts by passing the speech features to the model.

*Note: The `Speech2TextProcessor` object uses [torchaudio](https://github.com/pytorch/audio)  to extract the
filter bank features. Make sure to install the `torchaudio` package before running this example.*

You could either install those as extra speech dependancies with
`pip install transformers"[speech, sentencepiece]"` or install the packages seperatly 
with `pip install torchaudio sentencepiece`.


```python
import torch
from transformers import Speech2TextProcessor, Speech2TextForConditionalGeneration
from datasets import load_dataset
import soundfile as sf

model = Speech2TextForConditionalGeneration.from_pretrained("facebook/s2t-small-mustc-en-de-st")
processor = Speech2TextProcessor.from_pretrained("facebook/s2t-small-mustc-en-de-st")

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

ds = load_dataset(
    "patrickvonplaten/librispeech_asr_dummy",
    "clean",
    split="validation"
)
ds = ds.map(map_to_array)

inputs = processor(
    ds["speech"][0],
    sampling_rate=16_000,
    return_tensors="pt"
)
generated_ids = model.generate(input_ids=inputs["input_features"], attention_mask=inputs["attention_mask"])

translation = processor.batch_decode(generated_ids, skip_special_tokens=True)
```


## Training data

The s2t-small-mustc-en-de-st is trained on English-German subset of [MuST-C](https://ict.fbk.eu/must-c/).
MuST-C is a multilingual speech translation corpus whose size and quality facilitates the training of end-to-end systems
for speech translation from English into several languages. For each target language, MuST-C comprises several hundred
hours of audio recordings from English TED Talks, which are automatically aligned at the sentence level with their manual
transcriptions and translations.


## Training procedure

### Preprocessing

The speech data is pre-processed by extracting Kaldi-compliant 80-channel log mel-filter bank features automatically from
WAV/FLAC audio files via PyKaldi or torchaudio. Further utterance-level CMVN (cepstral mean and variance normalization)
is applied to each example.

The texts are lowercased and tokenized using SentencePiece and a vocabulary size of 8,000.


### Training

The model is trained with standard autoregressive cross-entropy loss and using [SpecAugment](https://arxiv.org/abs/1904.08779).
The encoder receives speech features, and the decoder generates the transcripts autoregressively. To accelerate
model training and for better performance the encoder is pre-trained for English ASR.

## Evaluation results

MuST-C test results for en-de (BLEU score): 22.7



### BibTeX entry and citation info

```bibtex
@inproceedings{wang2020fairseqs2t,
  title = {fairseq S2T: Fast Speech-to-Text Modeling with fairseq},
  author = {Changhan Wang and Yun Tang and Xutai Ma and Anne Wu and Dmytro Okhonko and Juan Pino},
  booktitle = {Proceedings of the 2020 Conference of the Asian Chapter of the Association for Computational Linguistics (AACL): System Demonstrations},
  year = {2020},
}

```