BriefGPT.xyz
Nov, 2018
使用发音辅助的子词建模提高端到端语音识别
Improving End-to-end Speech Recognition with Pronunciation-assisted Sub-word Modeling
HTML
PDF
Hainan Xu, Shuoyang Ding, Shinji Watanabe
TL;DR
本文提出一种发音辅助子词建模方法（PASM），该方法利用单词的发音信息提取子词，实验表明该方法可以比基于字符的基准方法和常用的字节对编码方法更好地提高语音识别精度。
Abstract
In recent years,
end-to-end
models have become popular for application in automatic
speech recognition
. Compared to hybrid approaches, which perform the phone-sequence to word conversion based on a lexicon, an
→