Regenerates the speech on the named drafts, applying enhancement as shared instructions. Mirrors the video generate-speech flow. Drafts with a chapter are regenerated from that chapter; drafts without one are regenerated from the linked video's matching slide.
| Time | Status | User Agent | |
|---|---|---|---|
Retrieving recent requests… | |||
Regenerates the speech on the named drafts, applying enhancement as shared instructions. Mirrors
the video-level Generate speech for slides flow.
Where the content comes from depends on the draft:
| Draft | Regenerated from |
|---|---|
Has a chapter | The chapter's title and points, targeting a duration based on the outline's language |
No chapter | The linked video's slide where slide.id matches draft.id, using that slide's own language, voice and provider |
That second case is why a chapterless draft with no matching video slide returns 400 — there is
nothing to regenerate from.
enhancementis instruction text, not an enumThe retired
/gptendpoints accepted one of ten fixed values (engaging,shorter, and so on).
Write what you want in plain language instead:"Make this more conversational and about a third shorter."