Hugging Face launches an open, multilingual text-to-speech leaderboard
Hugging Face and collaborators launched the Open TTS Leaderboard to evaluate open-source text-to-speech and voice-cloning models across languages. The authors say open models are underrepresented on arena-style rankings.
A new Open TTS Leaderboard was announced on the Hugging Face blog on September 30, focused on open-source and multilingual text-to-speech, including voice cloning and streaming performance.
The authors note the Hub hosts more than 8,000 TTS models, but evaluation is fragmented. They point out that only 16 of 92 models on Artificial Analysis's voice arena are open-weights, and argue that human-vote arenas cannot keep pace with releases. The leaderboard also lets users compare and vote on outputs.