WFST 动态解码

WFST:Speech recognition with weighted finite-state transducers

静态解码:

动态解码:Look-Ahead Composition,On-the-fly Rescoring

Look-Ahead Composition

 

二者比较:

《A comparison of dynamic WFST decoding approaches》

http://www.doc88.com/p-9405290980740.html

 

你可能感兴趣的:(ASR)