Talking avatar for PowerPoint presentations, slide by slide

To add a talking avatar to PowerPoint presentations, generate one short presenter video per slide and insert each MP4 on its slide. With Sume, each clip is a talking avatar video from a script of up to about 60 seconds, rendered in the slide's shape: 16:9 for PowerPoint's default widescreen slides, 4:3 for standard ones, or 1:1 or 3:4 for a presenter beside the slide content. The presenter always has a background behind them; the talking video has no transparent background.

Sume facts come from Generate avatar video and the Sume API reference; anything called current behavior is read from Sume's code. PowerPoint facts come from Microsoft's support pages on inserting a video and slide size. All were read on 2026-09-27. Sume makes the clips; placing them on slides happens in PowerPoint.

How does a talking head video work in PowerPoint?

It plays like any video placed on a slide. Microsoft's page says that, depending on your version of PowerPoint, an inserted video is embedded by default, which is convenient but increases the size of the presentation. In PowerPoint for Windows, a video in Slide Show plays as part of the click sequence by default; you can instead have it start automatically as soon as it appears onscreen, after any animation steps before it.

For the file itself, Microsoft recommends .mp4 files encoded with H.264 video and AAC audio. Sume's avatar video is an MP4, and in current code its audio track is AAC; Sume's docs don't state its video codec.

Which aspect ratio should the presenter clip use?

Match the clip to where the presenter appears. Sume's default is 9:16, a phone shape, so set aspect_ratio on every request:

How long can each slide's clip be?

One avatar video covers an estimated 4 to 60 seconds of script; how many words fit in 60 seconds turns that into a word count. A slide that needs more narration can be split into several clips joined with Timeline 1.0, as in avatar videos longer than 60 seconds. The spoken language is English only in current code.

Can the presenter have a transparent background?

No. The presenter always has a setting behind them, which you can set with scene: a text prompt or a public HTTPS photo. The talking-video request has no transparency option, a scene photo must be an image, and video backgrounds are not supported. Pick a plain scene that sits well with your slide design, and keep the same scene on every slide.

How do I make one clip per slide?

Send one request per slide with the same avatar_handle, aspect_ratio, and scene, so every clip uses the same presenter, frame shape, and scene direction. Talking avatar video API covers polling and options. 720p is the documented resolution.

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: q3-review-slide-01" \
  -d '{
    "avatar_handle": "product_host",
    "aspect_ratio": "16:9",
    "script": "Welcome to the third-quarter review. First, what shipped this quarter.",
    "scene": { "type": "prompt", "prompt": "Plain light-gray studio backdrop" }
  }'

How do I get the files into the deck, and what does it cost?

When a job completes, GET /v1/avatar-videos/{id} returns the clip's video_url; download each MP4 and insert it on its slide. Avatar video is priced per second by quality tier on API pricing: $0.184/s standard, $0.245/s plus, $0.55/s max (no product image), plus a 5.5% agent fee by default, so a deck's cost follows its total seconds of video. AI avatar video pricing works through per-minute costs.