Evaluating OpenAI's Whisper ASR: Performance analysis across diverse accents and speaker traits-Reference-Cited by-同舟云学术

Evaluating OpenAI's Whisper ASR: Performance analysis across diverse accents and speaker traits

Published:2024-02-01 Issue:2 Volume:4 Page:
ISSN:2691-1191
Container-title:JASA Express Letters
language:en
Short-container-title:

Author:

Graham Calbert¹,Roll Nathan²

Affiliation:

1. Phonetics Laboratory, University of Cambridge 1 , CB3 9DP, Cambridge, United Kingdom

2. University of California, Santa Barbara 2 , Santa Barbara, California 95405, USA crg29@cam.ac.uk , nroll@ucsb.edu

Abstract

This study investigates Whisper's automatic speech recognition (ASR) system performance across diverse native and non-native English accents. Results reveal superior recognition in American compared to British and Australian English accents with similar performance in Canadian English. Overall, native English accents demonstrate higher accuracy than non-native accents. Exploring connections between speaker traits [sex, native language (L1) typology, and second language (L2) proficiency] and word error rate uncovers notable associations. Furthermore, Whisper exhibits enhanced performance in read speech over conversational speech with modifications based on speaker gender. The implications of these findings are discussed.

Funder

Leverhulme Trust

Cambridge Language Sciences Incubator Fund

Publisher

Acoustical Society of America (ASA)

Link

https://pubs.aip.org/asa/jel/article-pdf/doi/10.1121/10.0024876/19692982/025206_1_10.0024876.pdf

Reference40 articles.

1. Do speech recognizers prefer female speakers?,2005

2. Non-native speech recognition using multi condition training,2013

3. Boersma, P., and Weenink, D. (2010). “ Praat: Doing phonetics by computer (version 6.3.17) [computer program],” http://www.praat.org/ (Last viewed February 5, 2024).

4. Interlingual identification and the role of foreign language experience in L2 vowel perception;Appl. Psycholinguist.,1990

5. Fully automated non-native speech recognition using confusion-based acoustic model integration,2005

Cited by 3 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. The Impact of Foreign Accents on the Performance of Whisper Family Models Using Medical Speech in Polish;International Conference on Information Systems Development;2024-09-09

2. Putting GPT-4o to the Sword: A Comprehensive Evaluation of Language, Vision, Speech, and Multimodal Proficiency;Applied Sciences;2024-09-03

3. A Voice User Interface on the Edge for People with Speech Impairments;Electronics;2024-04-07