Multilingual TTS Accent Impressions for Accented ASR
No Thumbnail Available
Access rights
openAccess
URL
Journal Title
Journal ISSN
Volume Title
A4 Artikkeli konferenssijulkaisussa
This publication is imported from Aalto University research portal.
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
View publication in the Research portal (opens in new window)
View/Open full text file from the Research portal (opens in new window)
Other link related to publication (opens in new window)
Date
2023
Major/Subject
Mcode
Degree programme
Language
en
Pages
11
317-327
317-327
Series
Text, Speech, and Dialogue - 26th International Conference, TSD 2023, Proceedings, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Volume 14102 LNAI
Abstract
Automatic Speech Recognition (ASR) for high-resource languages like English is often considered a solved problem. However, most high-resource ASR systems favor socioeconomically advantaged dialects. In the case of English, this leaves behind many L2 speakers and speakers of low-resource accents (a majority of English speakers). One way to mitigate this is to fine-tune a pre-trained English ASR model for a desired low-resource accent. However, collecting transcribed accented audio is costly and time-consuming. In this work, we present a method to produce synthetic L2-English speech via pre-trained text-to-speech (TTS) in an L1 language (target accent). This can be produced at a much larger scale and lower cost than authentic speech collection. We present initial experiments applying this augmentation method. Our results suggest that success of TTS augmentation relies on access to more than one hour of authentic training data and a diversity of target-domain prompts for speech synthesis.Description
Publisher Copyright: © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG.
Keywords
accented speech recognition, data augmentation, low-resource speech technologies, speech synthesis
Other note
Citation
Karakasidis, G, Robinson, N, Getman, Y, Ogayo, A, Al-Ghezi, R, Ayasi, A, Watanabe, S, Mortensen, D R & Kurimo, M 2023, Multilingual TTS Accent Impressions for Accented ASR . in K Ekštein, F Pártl & M Konopík (eds), Text, Speech, and Dialogue - 26th International Conference, TSD 2023, Proceedings . Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 14102 LNAI, Springer, pp. 317-327, International Conference on Text, Speech, and Dialogue, Pilsen, Czech Republic, 04/09/2023 . https://doi.org/10.1007/978-3-031-40498-6_28