Weakly Supervised Data Refinement and Flexible Sequence Compression for Efficient Thai LLM-based ASR

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Shao, Mingchen, Zhu, Xinfa, Wang, Chengyou, Mu, Bingshen, Li, Hai, Yan, Ying, Liu, Junhui, Xie, Danming, Xie, Lei
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916764008316928
author Shao, Mingchen
Zhu, Xinfa
Wang, Chengyou
Mu, Bingshen
Li, Hai
Yan, Ying
Liu, Junhui
Xie, Danming
Xie, Lei
author_facet Shao, Mingchen
Zhu, Xinfa
Wang, Chengyou
Mu, Bingshen
Li, Hai
Yan, Ying
Liu, Junhui
Xie, Danming
Xie, Lei
contents Despite remarkable achievements, automatic speech recognition (ASR) in low-resource scenarios still faces two challenges: high-quality data scarcity and high computational demands. This paper proposes EThai-ASR, the first to apply large language models (LLMs) to Thai ASR and create an efficient LLM-based ASR system. EThai-ASR comprises a speech encoder, a connection module and a Thai LLM decoder. To address the data scarcity and obtain a powerful speech encoder, EThai-ASR introduces a self-evolving data refinement strategy to refine weak labels, yielding an enhanced speech encoder. Moreover, we propose a pluggable sequence compression module used in the connection module with three modes designed to reduce the sequence length, thus decreasing computational demands while maintaining decent performance. Extensive experiments demonstrate that EThai-ASR has achieved state-of-the-art accuracy in multiple datasets. We release our refined text transcripts to promote further research.
format Preprint
id arxiv_https___arxiv_org_abs_2505_22063
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Weakly Supervised Data Refinement and Flexible Sequence Compression for Efficient Thai LLM-based ASR
Shao, Mingchen
Zhu, Xinfa
Wang, Chengyou
Mu, Bingshen
Li, Hai
Yan, Ying
Liu, Junhui
Xie, Danming
Xie, Lei
Sound
Audio and Speech Processing
Despite remarkable achievements, automatic speech recognition (ASR) in low-resource scenarios still faces two challenges: high-quality data scarcity and high computational demands. This paper proposes EThai-ASR, the first to apply large language models (LLMs) to Thai ASR and create an efficient LLM-based ASR system. EThai-ASR comprises a speech encoder, a connection module and a Thai LLM decoder. To address the data scarcity and obtain a powerful speech encoder, EThai-ASR introduces a self-evolving data refinement strategy to refine weak labels, yielding an enhanced speech encoder. Moreover, we propose a pluggable sequence compression module used in the connection module with three modes designed to reduce the sequence length, thus decreasing computational demands while maintaining decent performance. Extensive experiments demonstrate that EThai-ASR has achieved state-of-the-art accuracy in multiple datasets. We release our refined text transcripts to promote further research.
title Weakly Supervised Data Refinement and Flexible Sequence Compression for Efficient Thai LLM-based ASR
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2505.22063