Hugging Face launched the Open TTS Leaderboard for multilingual speech and voice cloning, giving speech generation a shared yardstick.
What happened
Hugging Face introduced the Open TTS Leaderboard, a scalable evaluation of multilingual text-to-speech and voice cloning. arXiv also carried EmoRES-TTS on emotional speech and a study of demographic disparities in pruned speech LLMs.
Why it matters
TTS quality is often compared on inconsistent terms, so a common leaderboard helps with model selection. How far emotion and fairness are covered by its metrics is not yet confirmed.
What to watch
Watch which models enter the leaderboard, the range of its metrics, and whether fairness gaps flagged in research show up in the evaluation.