Writes speech for the named slides from their on-canvas content, following enhancement as instructions. Use this for slides that have no speech yet; use rewrite-speech to change speech that already exists.
Replaces the retired POST /gpt/{videoId} and POST /gpt/{videoId}/{slideId}.
| Time | Status | User Agent | |
|---|---|---|---|
Retrieving recent requests… | |||
Writes speech for the named slides from their on-canvas content, following enhancement as
instructions. Use this for slides that have no speech yet — to change speech that already exists,
use Rewrite speech for slides.
Replaces the retired POST /gpt/{videoId} and POST /gpt/{videoId}/{slideId}.
enhancementis instruction text, not an enumThe old endpoints took one of ten fixed values. Write the change you want in plain language:
{ "slideIds": [1, 2, 3], "enhancement": "Warm, conversational, about 20 seconds per slide." }
enhancementPresetsis optional and recorded for analytics only — it does not affect the output.
enhancementcarries the actual instruction.
slideIds replaces the old per-slide URL: pass one id to act on one slide, or every id to cover
the whole video. Any id that isn't on the video returns 404 naming the missing ids.
Leave fromEditor unset. When false (the default) the generated speech is persisted server-side,
which is what an API caller wants; the editor sets it true because it owns persistence itself.
Requires LLM access on the account — 409 if unavailable.