Utterance-Wise Meeting Transcription System Using Asynchronous Distributed Microphones

TL;DR

This paper presents a novel meeting transcription system that uses asynchronous distributed microphones, integrating synchronization, diarization, speech enhancement, and recognition to improve accuracy over traditional monaural setups.

Contribution

It introduces a new framework that effectively handles asynchronous microphones and overlapped speech, improving transcription accuracy in real meeting environments.

Findings

Achieved 28.7% CER with 11 distributed microphones.

Reduced CER from 38.2% to 28.7% compared to monaural microphones.

Close to headset microphone transcription accuracy with 21.8% CER.

Abstract

A novel framework for meeting transcription using asynchronous microphones is proposed in this paper. It consists of audio synchronization, speaker diarization, utterance-wise speech enhancement using guided source separation, automatic speech recognition, and duplication reduction. Doing speaker diarization before speech enhancement enables the system to deal with overlapped speech without considering sampling frequency mismatch between microphones. Evaluation on our real meeting datasets showed that our framework achieved a character error rate (CER) of 28.7 % by using 11 distributed microphones, while a monaural microphone placed on the center of the table had a CER of 38.2 %. We also showed that our framework achieved CER of 21.8 %, which is only 2.1 percentage points higher than the CER in headset microphone-based transcription.