Music teacher intro video: an avatar, cutaways and a 480p b-roll

A music teacher can publish a 20-second intro with a reusable avatar for about $5.11 on Sume, including the one-time $0.95 avatar creation: an Avatar 1.0 talking video at standard quality, a detached audio spine, one 6-second Wan 3.0 cutaway and a Timeline render. Without the avatar fee, the same video is $4.16.

The trick is to let the avatar talk for the whole 20 seconds but to show it for only part of the video. The detached audio becomes the Timeline spine, and a cutaway shot (hands on a keyboard, a metronome) covers the middle. Avatar 1.0 is English-only, so this recipe is for English lessons. Prices are from the Sume pricing code (read 2026-10-10).

What is the endpoint chain?

The only unusual step is the one that makes the cutaway possible: the avatar video carries its own voice, so you detach it and feed it back as the spine.

• POST /v1/avatar-1.0/generate once to create the teacher avatar from a photo (reusable for every later lesson).
• POST /v1/avatar-1.0/talking-video with avatar_handle, a script of roughly 20 seconds, quality: standard, aspect_ratio: 9:16. Scripts must plan to 4 to 60 seconds.
• POST /v1/audio-detach on that video (wav, the default) to get a media.sume.com audio file.
• POST /v1/videos with wan-3.0, resolution: 480p, duration: 6 for the cutaway.
• POST /v1/timeline-1.0/render: slot 1 is the avatar clip from 0 to 8 s, slot 2 is the cutaway from 8 to 14 s, slot 3 is the avatar clip again from 14 to 20 s with source_in: 14.

What does it cost?

Avatar time is the big number. At standard quality with no product image, Sume bills $0.184 per second, so 20 seconds is $3.68. Plus quality is $0.245 per second and max is $0.55.

What can go wrong?

Timeline allows each later slot to start only after the previous one, and the cutaway must start at least where the first slot ends, so keep the three starts strictly increasing: 0, 8, 14. Use a 0.25-second fade transition on slots after the first if the cut feels hard; transitions are capped at one second and half of the shorter neighbor.

Avatar plans refuse scripts that run under 4 seconds or over 60, and one resolved avatar is allowed for each final video, so a two-teacher duet is two jobs joined in Timeline. See the Avatar video page for scenes and the Audio detach page for the wav shape that Timeline wants.

Where does this stop being the right tool?

If you need a live lesson or a student call, this is a recorded clip, not a live agent. For a course with many lessons, read AI avatar for online course videos and the training explainer plan-first note to price the whole series.

Reuse across lessons

The avatar is the reusable part. After the one-time $0.95, a new lesson intro is only the talking video, a detach, a cutaway and a render, and the script is the only thing that changes.

Keep the intro under 30 seconds. Avatar time is the line that grows, at $0.184 per second on standard, so a 60-second script is $11.04 before anything else is added.

• Read the script aloud once and cut anything you would not say to a student at the door.
• Pick a cutaway that shows the instrument, not hands that must look right; hands are a weak spot for any video model.
• Write the English script with plain words, because the avatar voice is English-only.