End-to-End Automatic Speech Recognition use tensorflow

use tensorflow to implement a end-to-end algorithm according baidu deepspeech paperDeepSpeech,DeepSpeech2 and seq2seq listener-attention-speller modelListen, Attend and Spell Towards better decoding and language model integration in sequence to Sequence model

1 prerequests.

see requirements.txt

2 data and preprocess

2.1 English corpus:LibriSpeech

2.2 Chinese corpus:THCHS-30

2.3 preprocess

3 train

4 test

5 other

refs. zzw922cn/Automatic_Speech_Recognition