Case study video narrated by an avatar: three scenes, 55 seconds

A written case study becomes a 55-second Sume avatar video in three scenes (the problem, what changed, the result), and it costs $10.12 on Standard, $13.48 on Plus and $30.25 on Max. The narrator is a generated presenter, not the customer, so the video should read as a summary of the written study, with every claim traceable to it.

The video_inputs shape and the 4-60 second window are from Sume's avatar video guide, read 2026-10-04.

Three scenes from a one-page study

Take one sentence from each section of the written case study and turn it into spoken lines. Do not invent a quote or attribute a statement to a person who did not say it.

Cost by tier

Request body

Spoken scenes use voice.type: "text"; the total planned duration must stay between 4 and 60 seconds, and scene backgrounds must resolve to one shared scene.

curl -X POST https://api.sume.com/v1/avatar-1.0/talking-video \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: case-study-001" \
  -d '{"avatar_handle": "narrator", "aspect_ratio": "16:9", "quality": "plus", "video_inputs": [{"id": "problem", "voice": {"type": "text", "script": "The support team handled every refund request by hand, and replies took days.", "duration": 15}, "background": {"type": "prompt", "prompt": "Bright office, neutral wall"}}, {"id": "change", "voice": {"type": "text", "script": "They moved the first reply into a short video that explains the refund steps.", "duration": 20}, "background": {"type": "prompt", "prompt": "Bright office, neutral wall"}}, {"id": "result", "voice": {"type": "text", "script": "Repeat questions dropped, and the team now uses the saved time for harder cases.", "duration": 20}, "background": {"type": "prompt", "prompt": "Bright office, neutral wall"}}]}'

Claims, approval and honesty

Give the customer the final script and keep their approval. Say a number only if it is in the written study, and say it as the study states it. Say in the clip, the caption or the page around it that the presenter is AI-generated; the API does not add that label for you.