Seed Audio

Seed Audio brings three usually-separate audio disciplines into a single AI model built on ByteDance's Seed-TTS, Seed-Music, and SeedFoley research. Type a prompt and choose speech, music, or sound design: speech supports zero-shot voice cloning from a short sample plus fine control over emotion and accent; music mode composes melody, instrumentation, and structure from a mood description or reference clip with editable lyrics; sound design layers ambient noise and foley (footsteps, rain, wind, impacts) synced to a scene in one render. Output is broadcast-ready MP3 in seconds, with text, reference-audio, or reference-image input modes. Free to try; paid plans start around $9.99/month equivalent (billed annually). Seed Audio is an independent platform and is not affiliated with or endorsed by ByteDance.