Saved in:
Bibliographic Details
Main Authors: Li, Zhaoyang, Zhou, Haodong, Luo, Longjie, Li, Xiaoxiao, Chen, Yongxin, Li, Lin, Hong, Qingyang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2506.02621
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866908391468695552
author Li, Zhaoyang
Zhou, Haodong
Luo, Longjie
Li, Xiaoxiao
Chen, Yongxin
Li, Lin
Hong, Qingyang
author_facet Li, Zhaoyang
Zhou, Haodong
Luo, Longjie
Li, Xiaoxiao
Chen, Yongxin
Li, Lin
Hong, Qingyang
contents This paper presents the system developed for Task 1 of the Multi-modal Information-based Speech Processing (MISP) 2025 Challenge. We introduce CASA-Net, an embedding fusion method designed for end-to-end audio-visual speaker diarization (AVSD) systems. CASA-Net incorporates a cross-attention (CA) module to effectively capture cross-modal interactions in audio-visual signals and employs a self-attention (SA) module to learn contextual relationships among audio-visual frames. To further enhance performance, we adopt a training strategy that integrates pseudo-label refinement and retraining, improving the accuracy of timestamp predictions. Additionally, median filtering and overlap averaging are applied as post-processing techniques to eliminate outliers and smooth prediction labels. Our system achieved a diarization error rate (DER) of 8.18% on the evaluation set, representing a relative improvement of 47.3% over the baseline DER of 15.52%.
format Preprint
id arxiv_https___arxiv_org_abs_2506_02621
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Cross-attention and Self-attention for Audio-visual Speaker Diarization in MISP-Meeting Challenge
Li, Zhaoyang
Zhou, Haodong
Luo, Longjie
Li, Xiaoxiao
Chen, Yongxin
Li, Lin
Hong, Qingyang
Sound
This paper presents the system developed for Task 1 of the Multi-modal Information-based Speech Processing (MISP) 2025 Challenge. We introduce CASA-Net, an embedding fusion method designed for end-to-end audio-visual speaker diarization (AVSD) systems. CASA-Net incorporates a cross-attention (CA) module to effectively capture cross-modal interactions in audio-visual signals and employs a self-attention (SA) module to learn contextual relationships among audio-visual frames. To further enhance performance, we adopt a training strategy that integrates pseudo-label refinement and retraining, improving the accuracy of timestamp predictions. Additionally, median filtering and overlap averaging are applied as post-processing techniques to eliminate outliers and smooth prediction labels. Our system achieved a diarization error rate (DER) of 8.18% on the evaluation set, representing a relative improvement of 47.3% over the baseline DER of 15.52%.
title Cross-attention and Self-attention for Audio-visual Speaker Diarization in MISP-Meeting Challenge
topic Sound
url https://arxiv.org/abs/2506.02621