TTS Arena

A crowdsourced, blind benchmark for text-to-speech. Type a line, hear two anonymous models read it back, and pick the one that sounds more human. Votes build an open leaderboard.

Vote in the arena →  ·  Docs →

What's here

A Bun monorepo:

• apps/web - the arena, leaderboard, and admin (Next.js)
• apps/router - service that calls TTS providers (Hono)
• apps/docs - the documentation site
• packages/* - shared types, the rating math, and providers

Running it

See the development guide for setup, configuration, and scripts. The short version:

bun install
cd apps/web && bun run db:migrate && bun run db:seed
cd ../.. && bun run dev

Contributing

Issues and PRs welcome. To get a model added, see Submit a model.

License

Apache 2.0.