AI avatar news anchor: make an AI news presenter video

An AI avatar news anchor is a generated presenter who reads your news script to camera: you write the stories, an avatar tool renders the anchor reading each one in front of a news-desk scene, and you caption the clips and cut them into a bulletin. To do it with Sume, create the anchor once as a reusable avatar, render each story as a 16:9 talking video of up to 60 seconds, caption each story, and join the stories with Timeline 1.0, which renders up to 30 minutes.

Sume facts come from the Generate avatar video, Video captions, and Timeline 1.0 docs, read on 2026-09-27. Anything called current behavior is read from Sume's code.

How do I create the news anchor?

Create the anchor once with POST /v1/avatar-1.0/generate, from a text prompt, a profile of traits, or a reference photo. Give it a stable avatar_handle: the docs recommend one so your app reuses a simple name instead of a generated id, and every story then sends the same handle. Creating an avatar costs $0.95 per avatar. How to create a reusable AI avatar walks through the request.

How do I render one news story?

Send each story to POST /v1/avatar-1.0/talking-video with aspect_ratio: "16:9" and a scene. A prompt scene directs the set, such as an anchor desk; a photo scene, { "type": "photo", "image_url": "https://..." }, uses a public HTTPS picture of your studio as the reference.

• Keep each story inside the window Sume accepts: an estimated 4–60 seconds. Today the estimate counts 2.8 words per second and rounds up clip by clip, so a story holds roughly 165 words at most; how many words fit in 60 seconds explains the count. Split longer stories into several jobs.
• Send the same avatar_handle, scene, aspect_ratio, and quality on every story so the bulletin matches. Each video resolves one avatar and one shared scene.
• Give each story its own Idempotency-Key. Completed results can include public media.sume.com video artifacts.

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: bulletin-0927-story-01" \
  -d '{
    "avatar_handle": "evening_anchor",
    "script": "Good evening. The city council voted tonight to extend the riverside bike lane.",
    "scene": { "type": "prompt", "prompt": "TV news studio, anchor desk, city skyline screen behind" },
    "aspect_ratio": "16:9",
    "quality": "plus"
  }'

How do I add captions to each story?

Caption each story before you join them. Today the caption job refuses a source longer than 60 seconds, so one job fits a story but not a whole bulletin. Two routes work:

• POST /v1/video-captions with the finished story's video_url and a style such as slam. It needs audible speech, which an anchor clip has.
• An avatar video preview created with captions. They are stored and burned in at generate-video, and you approve the desk framing on the first-frame still before paying for the render.
• Not the talking-video create itself: in current code that route refuses a captions field.

How do I join the stories into one bulletin?

Use one POST /v1/timeline-1.0/render. Each captioned story is a video[] slot, and the audio spine is the stories' own sound: detach each story's track with POST /v1/audio-detach and list the files, in order, in audio.parts[]. How to make an AI avatar video longer than 60 seconds shows the full request.

• Timeline renders 1080×1920 by default. For a 16:9 bulletin, set output.width and output.height, for example to 1920 and 1080.
• A fade or dissolve transition on a slot after the first softens the cut between stories. It lasts at most 1 second, and more than 8 adjacent fades are refused, so leave a hard cut in a long run of stories.
• Every URL must already be a media.sume.com file in your workspace. Sume mirrors generated outputs to its own media URLs, so the stories qualify.

What are the limits?

Each step has its own cap, and some things are out of scope:

• It isn't live. Each story is a job you submit and poll, and the result is a file.
• English only, in current code: the clip prompt asks for English speech, and avatar voices are cloned in English. For other languages, the route is TTS 1.0 with a language, then VEED Fabric 1.0 lip sync.
• One anchor per video. For two anchors, render one job per turn and cut them together, as in a two-speaker avatar conversation.
• The avatar speaks the script you send, so fact-checking and any AI disclosure stay with your newsroom.

What does an AI news anchor video cost?

Each story bills per second of video by quality tier: $0.184/s standard, $0.245/s plus, $0.55/s max (no product image). Each caption job reserves a fixed amount for a clip of up to 60 seconds; confirm it in GET /v1/catalog. The join is listed at $0.10 per output minute, reserved as the bulletin's length rounded up to whole minutes, and each audio detach is a flat per-job rate. All are plus a 5.5% agent fee by default. API pricing lists the avatar and Timeline rates.