SAML: Speaker Adaptive Mixture of LoRA Experts for End-to-End ASR

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhao, Qiuming, Sun, Guangzhi, Zhang, Chao, Xu, Mingxing, Zheng, Thomas Fang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917708086378496
author Zhao, Qiuming
Sun, Guangzhi
Zhang, Chao
Xu, Mingxing
Zheng, Thomas Fang
author_facet Zhao, Qiuming
Sun, Guangzhi
Zhang, Chao
Xu, Mingxing
Zheng, Thomas Fang
contents Mixture-of-experts (MoE) models have achieved excellent results in many tasks. However, conventional MoE models are often very large, making them challenging to deploy on resource-constrained edge devices. In this paper, we propose a novel speaker adaptive mixture of LoRA experts (SAML) approach, which uses low-rank adaptation (LoRA) modules as experts to reduce the number of trainable parameters in MoE. Specifically, SAML is applied to the quantised and personalised end-to-end automatic speech recognition models, which combines test-time speaker adaptation to improve the performance of heavily compressed models in speaker-specific scenarios. Experiments have been performed on the LibriSpeech and the TED-LIUM 3 corpora. Remarkably, with a 7x reduction in model size, 29.1% and 31.1% relative word error rate reductions were achieved on the quantised Whisper model and Conformer-based attention-based encoder-decoder ASR model respectively, comparing to the original full precision models.
format Preprint
id arxiv_https___arxiv_org_abs_2406_19706
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle SAML: Speaker Adaptive Mixture of LoRA Experts for End-to-End ASR
Zhao, Qiuming
Sun, Guangzhi
Zhang, Chao
Xu, Mingxing
Zheng, Thomas Fang
Sound
Audio and Speech Processing
Mixture-of-experts (MoE) models have achieved excellent results in many tasks. However, conventional MoE models are often very large, making them challenging to deploy on resource-constrained edge devices. In this paper, we propose a novel speaker adaptive mixture of LoRA experts (SAML) approach, which uses low-rank adaptation (LoRA) modules as experts to reduce the number of trainable parameters in MoE. Specifically, SAML is applied to the quantised and personalised end-to-end automatic speech recognition models, which combines test-time speaker adaptation to improve the performance of heavily compressed models in speaker-specific scenarios. Experiments have been performed on the LibriSpeech and the TED-LIUM 3 corpora. Remarkably, with a 7x reduction in model size, 29.1% and 31.1% relative word error rate reductions were achieved on the quantised Whisper model and Conformer-based attention-based encoder-decoder ASR model respectively, comparing to the original full precision models.
title SAML: Speaker Adaptive Mixture of LoRA Experts for End-to-End ASR
topic Sound
Audio and Speech Processing
url https://arxiv.org/abs/2406.19706