Text To Speech
Description
Natural sounding synthetic speech has been possible for many years and is now ubiquitous in many personal assistants made by the largest companies in the world. Notwithstanding, Idiap has expertise in several niche areas of speech synthesis. These include localisation of accents and dialects, particularly those of our own locality in French speaking Switzerland, and of our neighbours in the German speaking region. We also have expertise in synthesis of emphasis and emotions, necessary when the synthetic speech is used in a dialogue manager, or as the output of a speech to speech translation system. Our recent research involves how to combine these qualities into state of the art neural synthesisers. In the 2023 Blizzard Challenge, our system was ranked favourably.
Publications
- Honnet, P-E. and Lazaridis, A. and Garner, P. and Yamagishi, J. (2017) The SIWIS French Speech Synthesis Database – Design and recording of a high quality French database for speech synthesis. Idiap Research Institute.
- Haolin Chen and Philip N. Garner. Diffusion transformer for adaptive text-to-speech. In Proceedings of the 12th ISCA Speech Synthesis Workshop, pages 157--162, Grenoble, France, August 2023.
- Haolin Chen, Mutian He, Louise Coppieters de Gibson, and Philip N. Garner. The Idiap speech synthesis system for the Blizzard Challenge 2023. In Blizzard Challenge 2023, Grenoble, France, August 2023.
Links
- Repository: https://github.com/idiap/IdiapTTS
Advantages
Our key advantage is the ability to localise the accent and dialect using very little example data from a given region.
Applications
- Speech to speech translation
- Assistance for sight impaired readers
- Component in dialogue systems
- Hands free communication
Technology Readiness Level
TRL 5
Contact us for more information
- Interested in using our technologies?
- Interested to know more about the licensing possibilities and conditions?