File size: 4,413 Bytes
fda3753
 
 
 
 
 
5ca11a3
4087411
654c68c
fda3753
4087411
 
 
 
a7c511c
 
8722e95
faf3834
4087411
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29259e2
 
15043b1
29259e2
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
5f42b99
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
4087411
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
---
library_name: fairseq
task: audio-to-audio
tags:
- fairseq
- audio
- audio-to-audio
- speech-to-speech-translation
language: es-en
datasets:
- mtedx
- covost2
- europarl_st
- voxpopuli
widget:
- example_title: Common Voice sample 1
  src: https://huggingface.co/facebook/xm_transformer_600m-es_en-multi_domain/resolve/main/common_voice_es_19966634.flac
---
# xm_transformer_600m-es_en-multi_domain

[W2V2-Transformer](https://aclanthology.org/2021.acl-long.68/) speech-to-text translation model from fairseq S2T ([paper](https://arxiv.org/abs/2010.05171)/[code](https://github.com/pytorch/fairseq/tree/main/examples/speech_to_text)):
- Spanish-English
- Trained on mTEDx, CoVoST 2, EuroParl-ST, VoxPopuli, Multilingual LibriSpeech, Common Voice v7 and CCMatrix
- Speech synthesis with [facebook/fastspeech2-en-ljspeech](https://huggingface.co/facebook/fastspeech2-en-ljspeech)

## Usage
```python
from fairseq.checkpoint_utils import load_model_ensemble_and_task_from_hf_hub
from fairseq.models.text_to_speech.hub_interface import S2THubInterface
from fairseq.models.text_to_speech.hub_interface import TTSHubInterface
import IPython.display as ipd
import torchaudio


models, cfg, task = load_model_ensemble_and_task_from_hf_hub(
    "facebook/xm_transformer_600m-es_en-multi_domain",
    arg_overrides={"config_yaml": "config.yaml"},
)
model = models[0]
generator = task.build_generator(model, cfg)


# requires 16000Hz mono channel audio
audio, _ = torchaudio.load("/path/to/an/audio/file")

sample = S2THubInterface.get_model_input(task, audio)
text = S2THubInterface.get_prediction(task, model, generator, sample)

# speech synthesis
tts_models, tts_cfg, tts_task = load_model_ensemble_and_task_from_hf_hub(
  f"facebook/fastspeech2-en-ljspeech",
  arg_overrides={"vocoder": "griffin_lim", "fp16": False},
)
tts_model = tts_models[0]
TTSHubInterface.update_cfg_with_data_cfg(tts_cfg, tts_task.data_cfg)
tts_generator = tts_task.build_generator([tts_model], tts_cfg)

tts_sample = TTSHubInterface.get_model_input(tts_task, text)
wav, sr = TTSHubInterface.get_prediction(
    tts_task, tts_model, tts_generator, tts_sample
)

ipd.Audio(wav, rate=rate)
```

## Citation
```bibtex
@inproceedings{li-etal-2021-multilingual,
    title = "Multilingual Speech Translation from Efficient Finetuning of Pretrained Models",
    author = "Li, Xian  and
      Wang, Changhan  and
      Tang, Yun  and
      Tran, Chau  and
      Tang, Yuqing  and
      Pino, Juan  and
      Baevski, Alexei  and
      Conneau, Alexis  and
      Auli, Michael",
    booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers)",
    month = aug,
    year = "2021",
    address = "Online",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.acl-long.68",
    doi = "10.18653/v1/2021.acl-long.68",
    pages = "827--838",
}

@inproceedings{wang-etal-2020-fairseq,
    title = "Fairseq {S}2{T}: Fast Speech-to-Text Modeling with Fairseq",
    author = "Wang, Changhan  and
      Tang, Yun  and
      Ma, Xutai  and
      Wu, Anne  and
      Okhonko, Dmytro  and
      Pino, Juan",
    booktitle = "Proceedings of the 1st Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics and the 10th International Joint Conference on Natural Language Processing: System Demonstrations",
    month = dec,
    year = "2020",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2020.aacl-demo.6",
    pages = "33--39",
}

@inproceedings{wang-etal-2021-fairseq,
    title = "fairseq S{\^{}}2: A Scalable and Integrable Speech Synthesis Toolkit",
    author = "Wang, Changhan  and
      Hsu, Wei-Ning  and
      Adi, Yossi  and
      Polyak, Adam  and
      Lee, Ann  and
      Chen, Peng-Jen  and
      Gu, Jiatao  and
      Pino, Juan",
    booktitle = "Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing: System Demonstrations",
    month = nov,
    year = "2021",
    address = "Online and Punta Cana, Dominican Republic",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.emnlp-demo.17",
    doi = "10.18653/v1/2021.emnlp-demo.17",
    pages = "143--152",
}
```