Generate Seed Voice Preview

Generate a brand new voice from scratch with Seed Audio 1.0.

The voice can be steered by the text prompt alone, by up to 3 reference audio or video clips (uploaded files and/or existing URLs, tagged @Audio1..@Audio3 in the prompt), or by a single reference image. Audio and image references are mutually exclusive. Video uploads have their first audio track extracted before use. No DB record is created — the generated audio is uploaded to the CDN and returned as a preview.