Read Aloud Video
Read Aloud Video uses read_aloud_video and creates page-by-page narrated visuals.
{ "type": "read_aloud_video", "prompt": "Explain MCP step by step in six pages", "language": "en", "aspect_ratio": "16:9", "scene_count": 6}Use only catalog-supported page counts. The detail may include PPT or PDF artifacts in addition to the video. Do not send duration_seconds.
Best-fit requests
Use Read Aloud Video when the content should unfold as a deliberate page sequence: a narrated lesson, a first-week guide, a story, or a public procedure. State who is watching, what each page should accomplish, and what terminology must stay exact. This keeps a page-based video from becoming a collection of disconnected slides.
Use the type-specific controls
Read /catalog before creating the request and use its permitted scene_count, language, ratio, style, voice, and quality values. scene_count controls the number of pages; duration_seconds is not a valid field for this type. Submit one POST /creations request, poll the returned job, then retrieve the detail response to find the completed video and any PPT or PDF artifacts.
For changes, first inspect editable_actions and scenes[] on the completed creation. When a job fails, use its structured status to correct an unsupported value or a missing instruction rather than submitting the same request again.
Related guides
Quickstartfor the shared API workflowCreations & Jobsfor the result lifecycleAI PPTwhen the presentation file is the primary output