Start rendering this audiobook in the background

Returns immediately with a render you poll — no audio in this response, and nothing to hold a connection open for. Narration is produced by your render fleet: your own devices registered as render nodes, plus the Ditto cloud renderer when your plan includes it.

chapters is required and carries each chapter's segment count. Segment boundaries are derived from it deterministically, which is what makes narration reusable: re-requesting the same (audiobook, chapter, voice, speed) addresses audio that already exists instead of re-rendering (and re-charging for) it.

Finished segments become streamable one at a time — read the timeline while the render is still running rather than waiting for completed.