/Multilingual_TTS_Augmentation

Multilingual TTS data augmentation through a pivot language for low-resource ASR

Primary LanguagePythonCreative Commons Zero v1.0 UniversalCC0-1.0

Multilingual_TTS_Augmentation

Multilingual TTS data augmentation through a pivot language for low-resource ASR

Phone mappings and data set splits and information, mentioned in When Is TTS Augmentation Through a Pivot Language Useful?, available by Nov. 8, 2022

Needed uploads:

  • Google TTS code (from Swetha Gangu)
  • Microsoft TTS code (from Perez Ogayo)
  • Link to ESPNet recipe (from Nate Robinson)
  • Authentic data (or links thereto) with train/dev/eval split (from Nate Robinson)
  • Text for TTS production (Kiswahili, Guarani, Suba)
  • Phone mappings (from Nate Robinson)

Unfortunately, many of our materials were kept on a server that is currently inaccessible due to operational failure. We are working on recovering them. Please contact nrrobins@cs.cmu.edu.