Abstract
We investigate what linguistic factors affect the performance of Automatic Speech Recognition (ASR) models. We hypothesize that orthographic and phonological complexities both degrade accuracy. To examine this, we fine-tune the multilingual self-supervised pretrained model Wav2Vec2-XLSR-53 on 25 languages with 15 writing systems, and we compare their ASR accuracy, number of graphemes, unigram grapheme entropy, logographicity (how much word/morpheme-level information is encoded in the writing system), and number of phonemes. The results demonstrate that a high logographicity correlates with low ASR accuracy, while phonological complexity has no significant effect.- Anthology ID:
- 2024.acl-long.827
- Volume:
- Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
- Month:
- August
- Year:
- 2024
- Address:
- Bangkok, Thailand
- Editors:
- Lun-Wei Ku, Andre Martins, Vivek Srikumar
- Venue:
- ACL
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 15493–15503
- Language:
- URL:
- https://preview.aclanthology.org/add_missing_videos/2024.acl-long.827/
- DOI:
- 10.18653/v1/2024.acl-long.827
- Cite (ACL):
- Chihiro Taguchi and David Chiang. 2024. Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn’t. In Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 15493–15503, Bangkok, Thailand. Association for Computational Linguistics.
- Cite (Informal):
- Language Complexity and Speech Recognition Accuracy: Orthographic Complexity Hurts, Phonological Complexity Doesn’t (Taguchi & Chiang, ACL 2024)
- PDF:
- https://preview.aclanthology.org/add_missing_videos/2024.acl-long.827.pdf