end2end-asr-pytorch

end2end-asr-pytorch

https://github.com/gentaiscool/end2end-asr-pytorch

End-to-End Automatic Speech Recognition on PyTorch.
End-to-End Speech Recognition on Pytorch.
Transformer-based Speech Recognition Model.

end-to-end：adj. 端到端的，端點對端點的 n. 不斷地
automatic speech recognition，ASR：自動語音識別
text to speech，TTS：從文本到語音
speech to text，STT：從語音到文本

PyTorch
https://pytorch.org/

torchaudio: an audio library for PyTorch
https://github.com/pytorch/audio

1. pytorch==1.4.0 torchaudio==0.4.0 torchvision==0.50

1.1 get started

(base) yongqiang@yongqiang:~$ conda create -n pt-1.4_py-3.6 python=3.6
......
# To activate this environment, use
#
#     $ conda activate pt-1.4_py-3.6
#
# To deactivate an active environment, use
#
#     $ conda deactivate

(base) yongqiang@yongqiang:~$
(base) yongqiang@yongqiang:~$ conda activate pt-1.4_py-3.6
(pt-1.4_py-3.6) yongqiang@yongqiang:~$

conda install pytorch torchvision cpuonly -c pytorch

# CUDA 9.2
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=9.2 -c pytorch

# CUDA 10.0
conda install pytorch==1.2.0 torchvision==0.4.0 cudatoolkit=10.0 -c pytorch

# CPU Only
conda install pytorch==1.2.0 torchvision==0.4.0 cpuonly -c pytorch

• conda install pytorch==1.4.0 torchvision torchaudio==0.4.0 cpuonly -c pytorch

(pt-1.4_py-3.6) yongqiang@yongqiang:~$ conda install pytorch==1.4.0 torchvision torchaudio==0.4.0 cpuonly -c pytorch
......
## Package Plan ##

  environment location: /home/yongqiang/miniconda3/envs/pt-1.4_py-3.6

  added / updated specs:
    - cpuonly
    - pytorch==1.4.0
    - torchaudio==0.4.0
    - torchvision
......
Preparing transaction: done
Verifying transaction: done
Executing transaction: done
(pt-1.4_py-3.6) yongqiang@yongqiang:~$

(pt-1.4_py-3.6) yongqiang@yongqiang:~$ python
Python 3.6.10 |Anaconda, Inc.| (default, May  8 2020, 02:54:21)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
>>> import torchvision
>>> import torchaudio
>>>
>>> torch.__version__
'1.4.0'
>>>
>>> torchvision.__version__
'0.5.0'
>>>
>>> torchaudio.__version__
'0.4.0a0+719bcc7'
>>>
>>> exit()
(pt-1.4_py-3.6) yongqiang@yongqiang:~$

• bash requirement.sh

(pt-1.4_py-3.6) yongqiang@yongqiang:~/pytorch_work/end2end-asr-pytorch$ bash requirement.sh
......
(pt-1.4_py-3.6) yongqiang@yongqiang:~/pytorch_work/end2end-asr-pytorch$

2.

References

Code-Switched Language Models Using Neural Based Synthetic Data from Parallel Sentences.
Lightweight and Efficient End-to-End Speech Recognition Using Low-Rank Transformer.
A Comparison of Modeling Units in Sequence-to-Sequence Speech Recognition with the Transformer on Mandarin Chinese.