Saved in:
Bibliographic Details
Main Authors: Luo, Longjie, Lu, Shenghui, Li, Lin, Hong, Qingyang
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2505.24446
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911018403233792
author Luo, Longjie
Lu, Shenghui
Li, Lin
Hong, Qingyang
author_facet Luo, Longjie
Lu, Shenghui
Li, Lin
Hong, Qingyang
contents This paper presents our system for the MISP-Meeting Challenge Track 2. The primary difficulty lies in the dataset, which contains strong background noise, reverberation, overlapping speech, and diverse meeting topics. To address these issues, we (a) designed G-SpatialNet, a speech enhancement (SE) model to improve Guided Source Separation (GSS) signals; (b) proposed TLS, a framework comprising time alignment, level alignment, and signal-to-noise ratio filtering, to generate signal-level pseudo labels for real-recorded far-field audio data, thereby facilitating SE models' training; and (c) explored fine-tuning strategies, data augmentation, and multimodal information to enhance the performance of pre-trained Automatic Speech Recognition (ASR) models in meeting scenarios. Finally, our system achieved character error rates (CERs) of 5.44% and 9.52% on the Dev and Eval sets, respectively, with relative improvements of 64.8% and 52.6% over the baseline, securing second place.
format Preprint
id arxiv_https___arxiv_org_abs_2505_24446
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Pseudo Labels-based Neural Speech Enhancement for the AVSR Task in the MISP-Meeting Challenge
Luo, Longjie
Lu, Shenghui
Li, Lin
Hong, Qingyang
Sound
Audio and Speech Processing
This paper presents our system for the MISP-Meeting Challenge Track 2. The primary difficulty lies in the dataset, which contains strong background noise, reverberation, overlapping speech, and diverse meeting topics. To address these issues, we (a) designed G-SpatialNet, a speech enhancement (SE) model to improve Guided Source Separation (GSS) signals; (b) proposed TLS, a framework comprising time alignment, level alignment, and signal-to-noise ratio filtering, to generate signal-level pseudo labels for real-recorded far-field audio data, thereby facilitating SE models' training; and (c) explored fine-tuning strategies, data augmentation, and multimodal information to enhance the performance of pre-trained Automatic Speech Recognition (ASR) models in meeting scenarios. Finally, our system achieved character error rates (CERs) of 5.44% and 9.52% on the Dev and Eval sets, respectively, with relative improvements of 64.8% and 52.6% over the baseline, securing second place.
title Pseudo Labels-based Neural Speech Enhancement for the AVSR Task in the MISP-Meeting Challenge
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2505.24446