arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2104.05055cs.CLcs.SDeess.AS
收藏

NeMo逆向文本规范化:从开发到生产

NeMo Inverse Text Normalization: From Development To Production

NVIDIA(英伟达);Google(谷歌);The Graduate Center, City University of New York(纽约市立大学研究生中心)

Yang Zhang, Evelina Bakhturina, Kyle Gorman, Boris Ginsburg

摘要:

反向文本规范化(ITN)将口语域自动语音识别(ASR)输出转换为书面域文本,以提高ASR输出的可读性。许多国家的最先进的ITN系统使用手写加权有限状态传感器(WFST)的语法,因为这项任务具有极低的容忍度不可恢复的错误。我们为ITN引入了一个基于开源Python WFST的库,它可以实现从开发到生产的无缝路径。我们描述了ITN语法规则的英语规范,但该库可以适用于其他语言。它也可以用于书面到口语的文本规范化。我们使用Google Text标准化数据集的修改版本来评估NeMo ITN库。

英文摘要:

Inverse text normalization (ITN) converts spoken-domain automatic speech recognition (ASR) output into written-domain text to improve the readability of the ASR output. Many state-of-the-art ITN systems use hand-written weighted finite-state transducer(WFST) grammars since this task has extremely low tolerance to unrecoverable errors. We introduce an open-source Python WFST-based library for ITN which enables a seamless path from development to production. We describe the specification of ITN grammar rules for English, but the library can be adapted for other languages. It can also be used for written-to-spoken text normalization. We evaluate the NeMo ITN library using a modified version of the Google Text normalization dataset.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑