Faster, Simpler and More Accurate Hybrid ASR Systems Using Wordpieces
Abstract
In this work, we first show that on the widely used LibriSpeech benchmark, our transformer-based context-dependent connectionist temporal classification (CTC) system produces state-of-the-art results. We then show that using wordpieces as modeling units combined with CTC training, we can greatly simplify the engineering pipeline compared to conventional frame-based cross-entropy training by excluding all the GMM bootstrapping, decision tree building and force alignment steps, while still achieving very competitive word-error-rate. Additionally, using wordpieces as modeling units can significantly improve runtime efficiency since we can use larger stride without losing accuracy. We further confirm these findings on two internal VideoASR datasets: German, which is similar to English as a fusional language, and Turkish, which is an agglutinative language.
Cite
@article{arxiv.2005.09150,
title = {Faster, Simpler and More Accurate Hybrid ASR Systems Using Wordpieces},
author = {Frank Zhang and Yongqiang Wang and Xiaohui Zhang and Chunxi Liu and Yatharth Saraf and Geoffrey Zweig},
journal= {arXiv preprint arXiv:2005.09150},
year = {2020}
}
Comments
In proceedings Interspeech 2020