Starstruk Logo MarkStarstruk

v3.1

BenedictKing/ccxv3.1Jul 25, 2024by Flux9665

AI Summary

This release updates the Universal TTS model checkpoints for 7000 languages, addressing limitations from the previous version. It introduces stochastic prosody prediction and expands language coverage through enhanced pretraining and IPA support.

Key Highlights

  • Prosody prediction is now stochastic, sampling from a distribution for pitch, energy, and durations.
  • Added support for more IPA modifiers to cover a wider range of languages.
  • Expanded the number of languages included in the pretraining dataset.
  • Overhauled language similarity prediction modules and their visualization.

New Features

  • Stochastic prosody prediction for pitch, energy, and durations
  • Expanded IPA modifier support for more languages
  • Increased language coverage in pretraining
  • Overhauled language similarity prediction modules and visualization

Full Release Notes

## What's Changed

This release provides new checkpoints and improves some aspects of the previous release that were not included due to time constraints.  For more information on the universal TTS model for 7000 languages, please refer to the previous release v3.0

- Prosody prediction in terms of pitch, energy and durations are now stochastic and sample from a distribution instead of assuming a one-to-one mapping. 
- Added support for more IPA modifiers to cover more languages
- Added more languages into the pretraining 
- Overhauled language similarity prediction modules and visualization

**Full Changelog**: https://github.com/DigitalPhonetics/IMS-Toucan/compare/v3.0...v3.1