Exploring the Integration of Speech Separation and Recognition with Self-Supervised Learning Representation

TL;DR

This paper investigates integrating speech separation with recognition using self-supervised learning representations, demonstrating significant WER improvements in noisy, reverberant multi-speaker scenarios.

Contribution

It introduces a novel training strategy combining TF-GridNet and WavLM SSLR for improved multi-speaker recognition in challenging environments.

Findings

Achieved 2.5% WER on reverberant WHAMR! test set.

Outperformed previous mask-based MVDR beamforming with filterbank features.

Validated effectiveness of SSLR in multi-speaker ASR tasks.

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%).