Transformer-based Acoustic Modeling for Hybrid Speech Recognition
Abstract
We propose and evaluate transformer-based acoustic models (AMs) for hybrid speech recognition. Several modeling choices are discussed in this work, including various positional embedding methods and an iterated loss to enable training deep transformers. We also present a preliminary study of using limited right context in transformer models, which makes it possible for streaming applications. We demonstrate that on the widely used Librispeech benchmark, our transformer-based AM outperforms the best published hybrid result by 19% to 26% relative when the standard n-gram language model (LM) is used. Combined with neural network LM for rescoring, our proposed approach achieves state-of-the-art results on Librispeech. Our findings are also confirmed on a much larger internal dataset.
Cite
@article{arxiv.1910.09799,
title = {Transformer-based Acoustic Modeling for Hybrid Speech Recognition},
author = {Yongqiang Wang and Abdelrahman Mohamed and Duc Le and Chunxi Liu and Alex Xiao and Jay Mahadeokar and Hongzhao Huang and Andros Tjandra and Xiaohui Zhang and Frank Zhang and Christian Fuegen and Geoffrey Zweig and Michael L. Seltzer},
journal= {arXiv preprint arXiv:1910.09799},
year = {2020}
}
Comments
to appear in ICASSP 2020