search

Alexander-H-Liu / End-to-end-ASR-Pytorch

This is an open source project (formerly named Listen, Attend and Spell - PyTorch Implementation) for end-to-end ASR implemented with Pytorch, the well known deep learning toolkit.
MIT License
1.18k stars 318 forks source link

Single audio inference #33

Closed iamxiaoyubei closed 5 years ago

iamxiaoyubei commented 5 years ago

How does your code make inference? That is, I pass in an audio, and the code outputs the audio recognition result.

Youyoun commented 5 years ago

Hi @iamxiaoyubei ,

Maybe I'm wrong, but this repo does not provide a means to infer a single audio. You'll have to write your own script to do that.

It shouldn't be that hard, since all the functions you need for converting data are already implemented, all you have to do is use them.

This repo seems more like a research framework, where you preprocess fixed sets of data and train different models on the same data in order to compare them.

If you really want a small turnaround, you could run preprocess_corpus.py on a folder with a structure similar to the training data. and then run the test on that folder.

iamxiaoyubei commented 5 years ago

I understand. Thank you~！

shamil-kadavan commented 4 years ago

Hi @iamxiaoyubei, I am a newbie to this field. Can you please guide me through?