The CUHK-TENCENT speaker diarization system for the ICASSP 2022 multi-channel multi-party meeting transcription challenge

TL;DR

This paper presents a multi-level feature fusion approach for speaker diarization in multi-channel meetings, addressing challenges like unknown speaker count and overlapped speech, achieving high performance in the M2MeT challenge.

Contribution

We introduce a novel multi-level feature fusion mechanism for target-speaker voice activity detection and a data augmentation method to enhance robustness in challenging scenarios.

Findings

Our system ranks second in the M2MeT diarization task.

Fusion of multiple sub-systems improves diarization accuracy.

Data augmentation enhances robustness in small angular differences.

Abstract

This paper describes our speaker diarization system submitted to the Multi-channel Multi-party Meeting Transcription (M2MeT) challenge, where Mandarin meeting data were recorded in multi-channel format for diarization and automatic speech recognition (ASR) tasks. In these meeting scenarios, the uncertainty of the speaker number and the high ratio of overlapped speech present great challenges for diarization. Based on the assumption that there is valuable complementary information between acoustic features, spatial-related and speaker-related features, we propose a multi-level feature fusion mechanism based target-speaker voice activity detection (FFM-TS-VAD) system to improve the performance of the conventional TS-VAD system. Furthermore, we propose a data augmentation method during training to improve the system robustness when the angular difference between two speakers is relatively small. We provide comparisons for different sub-systems we used in M2MeT challenge. Our submission is a fusion of several sub-systems and ranks second in the diarization task.