Unified Modeling of Multi-Talker Overlapped Speech Recognition and Diarization with a Sidecar Separator

TL;DR

This paper introduces a unified model combining speech recognition and diarization for overlapped multi-talker speech, using a Sidecar separator with minimal additional parameters, improving performance on multiple datasets.

Contribution

The study extends the Sidecar separator approach by adding a diarization branch, enabling joint modeling of ASR and diarization with negligible overhead and improved results.

Findings

Better ASR performance on LibriMix and LibriSpeechMix datasets.

Acceptable diarization results on CALLHOME with minimal adaptation.

Efficient joint modeling with only 768 extra parameters.

Abstract

Multi-talker overlapped speech poses a significant challenge for speech recognition and diarization. Recent research indicated that these two tasks are inter-dependent and complementary, motivating us to explore a unified modeling method to address them in the context of overlapped speech. A recent study proposed a cost-effective method to convert a single-talker automatic speech recognition (ASR) system into a multi-talker one, by inserting a Sidecar separator into the frozen well-trained ASR model. Extending on this, we incorporate a diarization branch into the Sidecar, allowing for unified modeling of both ASR and diarization with a negligible overhead of only 768 parameters. The proposed method yields better ASR results compared to the baseline on LibriMix and LibriSpeechMix datasets. Moreover, without sophisticated customization on the diarization task, our method achieves acceptable diarization results on the two-speaker subset of CALLHOME with only a few adaptation steps.