NTU Speechlab LLM-Based Multilingual ASR System for Interspeech MLC-SLM Challenge 2025

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Peng, Yizhou, Wang, Bin, Chao, Yi-Wen, Ma, Ziyang, Zhang, Haoyang, Liu, Hexin, Chen, Xie, Chng, Eng Siong
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915371142873088
author Peng, Yizhou
Wang, Bin
Chao, Yi-Wen
Ma, Ziyang
Zhang, Haoyang
Liu, Hexin
Chen, Xie
Chng, Eng Siong
author_facet Peng, Yizhou
Wang, Bin
Chao, Yi-Wen
Ma, Ziyang
Zhang, Haoyang
Liu, Hexin
Chen, Xie
Chng, Eng Siong
contents This report details the NTU Speechlab system developed for the Interspeech 2025 Multilingual Conversational Speech and Language Model (MLC-SLM) Challenge (Task I), where we achieved 5th place. We present comprehensive analyses of our multilingual automatic speech recognition system, highlighting key advancements in model architecture, data selection, and training strategies. In particular, language-specific prompts and model averaging techniques were instrumental in boosting system performance across diverse languages. Compared to the initial baseline system, our final model reduced the average Mix Error Rate from 20.2% to 10.6%, representing an absolute improvement of 9.6% (a relative improvement of 48%) on the evaluation set. Our results demonstrate the effectiveness of our approach and offer practical insights for future Speech Large Language Models.
format Preprint
id arxiv_https___arxiv_org_abs_2506_13339
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle NTU Speechlab LLM-Based Multilingual ASR System for Interspeech MLC-SLM Challenge 2025
Peng, Yizhou
Wang, Bin
Chao, Yi-Wen
Ma, Ziyang
Zhang, Haoyang
Liu, Hexin
Chen, Xie
Chng, Eng Siong
Computation and Language
Audio and Speech Processing
This report details the NTU Speechlab system developed for the Interspeech 2025 Multilingual Conversational Speech and Language Model (MLC-SLM) Challenge (Task I), where we achieved 5th place. We present comprehensive analyses of our multilingual automatic speech recognition system, highlighting key advancements in model architecture, data selection, and training strategies. In particular, language-specific prompts and model averaging techniques were instrumental in boosting system performance across diverse languages. Compared to the initial baseline system, our final model reduced the average Mix Error Rate from 20.2% to 10.6%, representing an absolute improvement of 9.6% (a relative improvement of 48%) on the evaluation set. Our results demonstrate the effectiveness of our approach and offer practical insights for future Speech Large Language Models.
title NTU Speechlab LLM-Based Multilingual ASR System for Interspeech MLC-SLM Challenge 2025
topic Computation and Language
Audio and Speech Processing
url https://arxiv.org/abs/2506.13339