AI music video from audio: a Lyria track, six H3 Max clips, $6.23

"AI music video generator from audio" is on Google's autocomplete today (read 2026-10-07). The straight path on Sume is: make or import the audio, make the picture in clips, and lay the audio under the clips in one Timeline 1.0 render. The track sets the length; the clips follow it.

This page prices the plain version: a 60-second video, six 10-second shots, one song.

The bill, line by line

Costs are provider list x 1.25. At 480p the same six clips are 60 x $0.0625 = $3.75, which puts the whole video at $3.975. At 1080p they are 60 x $0.20 = $12.00.

Make the track fit the picture

Music Router has no duration field and rejects duration outright; ask for length and structure in the prompt instead. Section markers help you line cuts up with the beat: [0:00-0:10] Intro, [0:10-0:20] Verse, and so on, so each shot has its own part of the song. Write the tempo as a number; at 120 BPM a 4-beat bar lasts 2 seconds, so a 10-second shot is five bars.

Then render. The timeline accepts an audio file as audio.url, so the track runs as the audio of the whole render.

curl -X POST https://api.sume.com/v1/music-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: mv-track-001" \
  -d '{"prompt": "Synth-pop, 120 BPM, A minor. [0:00-0:10] Intro: arpeggio only. [0:10-0:30] Verse. [0:30-0:50] Chorus, full band. [0:50-1:00] Outro. A 1-minute track. Instrumental, no vocals."}'

Notes on the clips

• H3 Max reads 5 to 15 seconds per clip at 480p, 768p or 1080p; ten seconds is inside the range.
• MiniMax H3 (not Max) is $0.075 per second at 768p, which would make the six clips $4.50.
• Native audio comes with the clip; decide whether the track or the clip sound is the one you keep, and check the result by ear.