Exploring the Integration of Speech Separation and Recognition with Self-Supervised Learning Representation
Abstract
Neural speech separation has made remarkable progress and its integration with automatic speech recognition (ASR) is an important direction towards realizing multi-speaker ASR. This work provides an insightful investigation of speech separation in reverberant and noisy-reverberant scenarios as an ASR front-end. In detail, we explore multi-channel separation methods, mask-based beamforming and complex spectral mapping, as well as the best features to use in the ASR back-end model. We employ the recent self-supervised learning representation (SSLR) as a feature and improve the recognition performance from the case with filterbank features. To further improve multi-speaker recognition performance, we present a carefully designed training strategy for integrating speech separation and recognition with SSLR. The proposed integration using TF-GridNet-based complex spectral mapping and WavLM-based SSLR achieves a 2.5% word error rate in reverberant WHAMR! test set, significantly outperforming an existing mask-based MVDR beamforming and filterbank integration (28.9%).
Cite
@article{arxiv.2307.12231,
title = {Exploring the Integration of Speech Separation and Recognition with Self-Supervised Learning Representation},
author = {Yoshiki Masuyama and Xuankai Chang and Wangyou Zhang and Samuele Cornell and Zhong-Qiu Wang and Nobutaka Ono and Yanmin Qian and Shinji Watanabe},
journal= {arXiv preprint arXiv:2307.12231},
year = {2023}
}
Comments
Accepted to IEEE WASPAA 2023