We present a distant automatic speech recognition (DASR) system developed for the CHiME-8 DASR track. It consists of a diarization first pipeline. For diarization, we use end-to-end diarization with vector clustering (EEND-VC) followed by target speaker voice activity detection (TS-VAD) refinement. To deal with various numbers of speakers, we developed a new multi-channel speaker counting approach. We then apply guided source separation (GSS) with several improvements to the baseline system. Finally, we perform ASR using a combination of systems built from strong pre-trained models. Our proposed system achieves a macro tcpWER of 21.3 % on the dev set, which is a 57 % relative improvement over the baseline.
@article{arxiv.2409.05554,
title = {NTT Multi-Speaker ASR System for the DASR Task of CHiME-8 Challenge},
author = {Naoyuki Kamo and Naohiro Tawara and Atsushi Ando and Takatomo Kano and Hiroshi Sato and Rintaro Ikeshita and Takafumi Moriya and Shota Horiguchi and Kohei Matsuura and Atsunori Ogawa and Alexis Plaquet and Takanori Ashihara and Tsubasa Ochiai and Masato Mimura and Marc Delcroix and Tomohiro Nakatani and Taichi Asami and Shoko Araki},
journal= {arXiv preprint arXiv:2409.05554},
year = {2024}
}