A Sidecar Separator Can Convert a Single-Talker Speech Recognition System to a Multi-Talker One

TL;DR

This paper introduces a Sidecar separator that enhances a single-talker ASR system to effectively recognize overlapping speech from multiple talkers by leveraging layer-specific speech embeddings.

Contribution

The paper proposes a novel Sidecar separator approach that, when added to a pre-trained ASR model, significantly improves multi-talker speech recognition performance.

Findings

Achieves 10.36% WER on LibriMix 2-speaker dataset

Outperforms previous state-of-the-art results

Maintains comparable performance with limited training data

Abstract

Although automatic speech recognition (ASR) can perform well in common non-overlapping environments, sustaining performance in multi-talker overlapping speech recognition remains challenging. Recent research revealed that ASR model's encoder captures different levels of information with different layers -- the lower layers tend to have more acoustic information, and the upper layers more linguistic. This inspires us to develop a Sidecar separator to empower a well-trained ASR model for multi-talker scenarios by separating the mixed speech embedding between two suitable layers. We experimented with a wav2vec 2.0-based ASR model with a Sidecar mounted. By freezing the parameters of the original model and training only the Sidecar (8.7 M, 8.4% of all parameters), the proposed approach outperforms the previous state-of-the-art by a large margin for the 2-speaker mixed LibriMix dataset, reaching a word error rate (WER) of 10.36%; and obtains comparable results (7.56%) for LibriSpeechMix dataset when limited training.