Fine-Tuning Strategies for Faster Inference Using Speech Self-Supervised Models: A Comparative Study
Author:
Affiliation:
1. Telecom Paris,Palaiseau,France
2. PSL Research University,ENS, INRIA, INSERM, UPEC,France
3. Samsung AI Center,Cambridge,United-Kingdom
4. Mila-Quebec AI Institute, Université de Montréal, Concordia University,Canada
Publisher
IEEE
Link
http://xplorestaging.ieee.org/ielx7/10192576/10192577/10193042.pdf?arnumber=10193042
Reference26 articles.
1. Efficient speech representation learning with low-bit quantization;yeh;arXiv preprint arXiv 2301 07898,2022
2. How Does Pre-Trained Wav2Vec 2.0 Perform on Domain-Shifted Asr? an Extensive Benchmark on Air Traffic Control Communications
3. Speechbrain: A general-purpose speech toolkit;ravanelli;arXiv preprint arXiv 2106 04427,2021
4. Superb @ SLT 2022: Challenge on Generalization and Efficiency of Self-Supervised Speech Representation Learning
Cited by 6 articles. 订阅此论文施引文献 订阅此论文施引文献，注册后可以免费订阅5篇论文的施引文献，订阅后可以查看论文全部施引文献
1. Training Early-Exit Architectures for Automatic Speech Recognition: Fine-Tuning Pre-Trained Models or Training from Scratch;2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW);2024-04-14
2. Skill: Similarity-Aware Knowledge Distillation for Speech Self-Supervised Learning;2024 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW);2024-04-14
3. On The Choice of the Optimal Temporal Support for Audio Classification with Pre-Trained Embeddings;ICASSP 2024 - 2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP);2024-04-14
4. Maize seed variety identification using hyperspectral imaging and self-supervised learning: A two-stage training approach without spectral preprocessing;Expert Systems with Applications;2024-03
5. What Do Self-Supervised Speech Models Know About Words?;Transactions of the Association for Computational Linguistics;2024
1.学者识别学者识别
2.学术分析学术分析
3.人才评估人才评估
"同舟云学术"是以全球学者为主线，采集、加工和组织学术论文而形成的新型学术文献查询和分析系统，可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容，当前同舟云学术共收录了国内外主流学术期刊6万余种，收集的期刊论文及会议论文总量共计约1.5亿篇，并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询！咨询电话：010-8811{复制后删除}0370
www.globalauthorid.com
TOP
Copyright © 2019-2024 北京同舟云网络信息技术有限公司 京公网安备11010802033243号 京ICP备18003416号-3