Fine Tuning Based End-to-End Indian English Speech Synthesis System

DOI

10.2991/978-94-6463-740-3_2How to use a DOI?

Keywords

Indian English; end-to-end; fine tuning; Tacotron2; Waveglow; FastSpeech2; speech synthesis

Abstract

Natural-sounding speech synthesis systems with end-to-end models have been designed for Spanish, American English, and Chinese. However, little work has been done on the end-to-end text-to-speech synthesis development for the Indian languages. The lack of good training data has been a challenge of this in the past. In this work, we have used approximately eight hours of training data to construct a human-resembling quality Indian English text-to-speech converting system. The checkpoints of Tacotron2, FastSpeech2, WaveGlow, and Parallel WaveGAN were pre-trained in American English, so we continued training them using the fine-tuning technique. Therefore, as far as the authors are aware, this is the best quality text-to-speech synthesis (TTS) for Indian English that has not been accomplished yet. Our experiment yields a mean opinion score (MOS) of 4.35 ± 0.14 with the Tacotron2 model and MOS of 4.12 ± 0.17 with the FastSpeech2 model.

Copyright

© 2025 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Manisha Gupta
AU  - Amita Dev
AU  - Poonam Bansal
PY  - 2025
DA  - 2025/06/25
TI  - Fine Tuning Based End-to-End Indian English Speech Synthesis System
BT  - Proceedings of the 6th International Conference on Deep Learning, Artificial Intelligence and Robotics (ICDLAIR 2024)
PB  - Atlantis Press
SP  - 3
EP  - 16
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6463-740-3_2
DO  - 10.2991/978-94-6463-740-3_2
ID  - Gupta2025
ER  -