NTT speaker diarization system for CHiME-7: multi-domain, multi-microphone End-to-end and vector clustering diarization
Abstract
This paper details our speaker diarization system designed for multi-domain, multi-microphone casual conversations. The proposed diarization pipeline uses weighted prediction error (WPE)-based dereverberation as a front end, then applies end-to-end neural diarization with vector clustering (EEND-VC) to each channel separately. It integrates the diarization result obtained from each channel using diarization output voting error reduction plus overlap (DOVER-LAP). To harness the knowledge from the target domain and results integrated across all channels, we apply self-supervised adaptation for each session by retraining the EEND-VC with pseudo-labels derived from DOVER-LAP. The proposed system was incorporated into NTT's submission for the distant automatic speech recognition task in the CHiME-7 challenge. Our system achieved 65 % and 62 % relative improvements on development and eval sets compared to the organizer-provided VC-based baseline diarization system, securing third place in diarization performance.
Cite
@article{arxiv.2309.12656,
title = {NTT speaker diarization system for CHiME-7: multi-domain, multi-microphone End-to-end and vector clustering diarization},
author = {Naohiro Tawara and Marc Delcroix and Atsushi Ando and Atsunori Ogawa},
journal= {arXiv preprint arXiv:2309.12656},
year = {2023}
}
Comments
5 pages, 5 figures, Submitted to ICASSP 2024