Cross-lingual Self-Supervised Speech Representations for Improved Dysarthric Speech Recognition

Hernandez A, Pérez Toro PA, Nöth E, Orozco Arroyave JR, Maier A, Yang SH (2022)

Publication Language: English

Publication Type: Conference contribution, Original article

Publication year: 2022

Pages Range: 51-55

Conference Proceedings Title: Proceedings of Interspeech 2022

Event location: Seoul KR

URI: https://www.isca-speech.org/archive/interspeech_2022/hernandez22_interspeech.html

DOI: 10.21437/Interspeech.2022-10674

Open Access Link: https://www.isca-speech.org/archive/interspeech_2022/hernandez22_interspeech.html

Abstract

State-of-the-art automatic speech recognition (ASR) systems perform well on healthy speech. However, the performance on impaired speech still remains an issue. The current study explores the usefulness of using Wav2Vec self-supervised speech representations as features for training an ASR system for dysarthric speech. Dysarthric speech recognition is particularly difficult as several aspects of speech such as articulation, prosody and phonation can be impaired. Specifically, we train an acoustic model with features extracted from Wav2Vec, Hubert, and the cross-lingual XLSR model. Results suggest that speech representations pretrained on large unlabelled data can improve word error rate (WER) performance. In particular, features from the multilingual model led to lower WERs than Fbanks or models trained on a single language. Improvements were seen in English speakers with cerebral palsy caused dysarthria (UASpeech corpus), Spanish speakers with Parkinsonian dysarthria (PC-GITA corpus) and Italian speakers with paralysis-based dysarthria (EasyCall corpus). Compared to using Fbank features, XLSR-based features reduced WERs by 6.8%, 22.0%, and 7.0% for the UASpeech, PC-GITA, and EasyCall corpus, respectively.

Authors with CRIS profile

Abner Hernandez Lehrstuhl für Informatik 5 (Mustererkennung) Paula Andrea Pérez Toro Technische Fakultät Elmar Nöth Professur für Informatik (Mustererkennung) Juan Rafael Orozco Arroyave Professur für Informatik (Mustererkennung) Andreas Maier Lehrstuhl für Informatik 5 (Mustererkennung) Seung Hee Yang Professur für Artificial Intelligence in Biomedical Speech Processing (Stiftungsprofessur)

How to cite

APA:

Hernandez, A., Pérez Toro, P.A., Nöth, E., Orozco Arroyave, J.R., Maier, A., & Yang, S.H. (2022). Cross-lingual Self-Supervised Speech Representations for Improved Dysarthric Speech Recognition. In Proceedings of Interspeech 2022 (pp. 51-55). Seoul, KR.

MLA:

Hernandez, Abner, et al. "Cross-lingual Self-Supervised Speech Representations for Improved Dysarthric Speech Recognition." Proceedings of the Interspeech, Seoul 2022. 51-55.

BibTeX: Download