Multi-Channel Sequence-to-Sequence Neural Diarization: Experimental Results for The MISP 2025 Challenge

TL;DR

This paper presents a multi-channel extension of the sequence-to-sequence neural diarization system, achieving state-of-the-art results in the MISP 2025 Challenge by effectively utilizing multi-channel audio data.

Contribution

The paper introduces MC-S2SND, an extension of S2SND that leverages multi-channel audio for improved speaker diarization accuracy.

Findings

Achieved DER of 8.09% on the challenge dataset

Ranked first in the MISP 2025 speaker diarization task

Demonstrated effectiveness of multi-channel audio in diarization

Abstract

This paper describes the speaker diarization system developed for the Multimodal Information-Based Speech Processing (MISP) 2025 Challenge. First, we utilize the Sequence-to-Sequence Neural Diarization (S2SND) framework to generate initial predictions using single-channel audio. Then, we extend the original S2SND framework to create a new version, Multi-Channel Sequence-to-Sequence Neural Diarization (MC-S2SND), which refines the initial results using multi-channel audio. The final system achieves a diarization error rate (DER) of 8.09% on the evaluation set of the competition database, ranking first place in the speaker diarization task of the MISP 2025 Challenge.