DISCUSSION ABOUT PROSODY AND SYNTHETIC DATA: CONCERN, REASON, AND FUTURE
DOI:
https://doi.org/10.21009/ishel.v2i1.69401Keywords:
literature review, synthetic data, prosody, artificial intelegent, sppechAbstract
This study used a literature review as the backbone for discussion. Talking about three points: concern, reason, and future. As we already understand how to use artificial intelligence (AI) to create a synthetic environment, we can become more familiar with it. With a practice logic like easy to use, no boundaries, no worries about privacy breach, and ready to generate. But with all the benefits, there is some concern: can synthetic data become an alternative to natural data? The main reason for using synthetic data is the manageable time required, unlike using human voices, which involves human-things challenges. But there is a line to think about how the prosody of synthetic data can sound like a natural one, or is it not? The transformer text-to-speech (TTS) model already appears to be publicly available, but what could be one role for it in language research in the future? All those questions are closely addressed in previous literature. This paper does not wrap up the answer but serves as a trigger for future research.
References
Barkat-Defradas, M., Raymond, M., & Suire, A. (2021). Vocal preferences in humans: A systematic review. In B. Weiss, A. Trouvain, & M. Barkat-Defradas (Eds.), Voice attractiveness: Studies on sexy, likable, and charismatic speakers (pp. 55–80). Springer. https://doi.org/10.1007/978-981-15-6627-1_4
Chim, J., Ive, J., & Liakata, M. (2025). Evaluating synthetic data generation from user-generated text. Computational Linguistics, 51(1), 191–248. https://doi.org/10.1162/coli_a_00540
De Wilde, P., Arora, P., Buarque de Lima Neto, F., Chin, Y., Thinyane, M., Stinckwich, S., Fournier-Tombs, E., & Marwala, T. (2024). Recommendations on the use of synthetic data to train AI models. United Nations University.
Guo, Y., Shang, G., Vazirgiannis, M., & Clavel, C. (2024). The curious decline of linguistic diversity: Training language models on synthetic text. In Findings of the Association for Computational Linguistics: NAACL 2024 (pp. 3589–3604).
Karl, A. L., Fernandes, G. S., Pires, L. A., Serpa, Y. R., & Caminha, C. (2025). Synthetic AI data pipeline for domain-specific speech-to-text solutions, in Proceedings of STIL 2025.
Korzekwa, D., Lorenzo-Trueba, J., Drugman, T., & Kostek, B. (2022). Computer-assisted pronunciation training—Speech synthesis is almost all you need. Computer Speech & Language, 76, Article 101390. https://doi.org/10.1016/j.csl.2022.101390
Ladd, D. R. (2008). Intonational phonology (2nd ed.). Cambridge University Press.
Masson, M., & Carson-Berndsen, J. (2024). Investigating the use of synthetic speech data for the analysis of Spanish-accented English pronunciation patterns in ASR. In Proceedings of Synthetic Data’s Transformative Role in Foundational Speech Models (pp. 81–85).
Mizumoto, T., Kojima, A., Fujita, Y., Liu, L., & Sudo, Y. (2025). Is synthetic data truly effective for training speech language models? In Proceedings of Interspeech 2025 (pp. 1808–1812).
Senocak, A., Park, S., Oh, T.-H., & Chung, J. S. (2026). How far can we go with synthetic data for audio-visual sound source localization? In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR 2026).
Dua, K., Mittal, P., Gupta, R., & Patel, H. L. (2025). SpeechWeave: Diverse multilingual synthetic text & audio data generation pipeline for training text to speech models, in Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics: Industry Track (pp. 718–737).
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2026 Gede Satya Hermawan, Endry Boeriswati, Miftahulkhairah Anwar
This work is licensed under a Creative Commons Attribution 4.0 International License.
