🇺🇦 Open Source Ukrainian Text-to-Speech datasets

The texts for these datasets are from Texts for the Ukrainian Text-to-Speech dataset

Community

• Discord: https://discord.gg/yVAjkBgmt4
• Speech Recognition: https://t.me/speech_recognition_uk
• Speech Synthesis: https://t.me/speech_synthesis_uk

Dataset

Look https://huggingface.co/datasets/Yehor/opentts-uk

Voices

Female

Lada

• Quality: high
• Duration: 10h37m
• Audio formats: OPUS
• Frequency: 48000 Hz

Listen to DEMO (choose "lada" in the Voice field)

Tetiana

• Quality: high
• Duration: 8h
• Audio formats: OPUS
• Frequency: 48000 Hz

Kateryna

• Quality: high
• Duration: 2h40m
• Audio formats: OPUS
• Frequency: 48000 Hz

Male

Mykyta

• Quality: high
• Duration: 8h10m
• Audio formats: OPUS
• Frequency: 48000 Hz

Listen to DEMO (choose "mykyta" in the Voice field)

Oleksa

• Quality: high
• Duration: 6h
• Audio formats: OPUS
• Frequency: 48000 Hz

Appearance on the web

• Align Text to Audio and Trim Silence: https://github.com/proger/uk
• NVIDIA's Flowtron: https://github.com/egorsmkv/ukrainian-flowtron-tts
• HF demos:
  • https://huggingface.co/spaces/robinhad/ukrainian-tts
  • https://huggingface.co/spaces/theodotus/ukrainian-voices
• Lada: Ukrainian High-Quality Female Text-to-Speech Dataset: https://zenodo.org/record/7396774
• Google Colabs (RADTTS model):
  • https://colab.research.google.com/drive/13aa0o9fQknDcJtpLrGXhxWPvZpeUggCy?usp=sharing
  • https://colab.research.google.com/drive/1pgiBlMm4tk0atKrszStOSy6XaTDnc3v4?usp=sharing
• Lada is in Piper - https://github.com/rhasspy/piper - A fast, local neural text to speech system
• Tetiana in Balacoon - https://balacoon.com/blog/uk_release/
  • Demo: https://huggingface.co/spaces/balacoon/tts