Hacker News new | ask | show | jobs
by whodunser 3391 days ago
It says they trained on 20 hours of a speech corpus subset. Will larger datasets influence the future of TTS?