Nine 6-second spokesperson lines: H3 Max 768p $5.40, Fabric $10.13

Price nine 6-second spoken lines on Sume: H3 Max Lip Sync at 480p or 768p against VEED Fabric at 480p or 720p, with the arithmetic.

5 min readSume
All posts

Nine spoken lines of 6 seconds each, 54 audio seconds in total, cost $5.40 on MiniMax H3 Max Lip Sync at 768p and $10.125 on VEED Fabric at 720p, both on Sume (as of 2026-10-09). The cheaper pair is H3 Max at 480p ($3.375) or Fabric at 480p ($5.40). Each line is its own job, so the nine clips still need a timeline or an editor to become one spot.

The scenario is a spokesperson who says nine short lines, one for each product feature. Nothing about the lines is special except that each one is under the 14.8-second ceiling of H3 Max Lip Sync, so either model can take any of them.

The price table

Sume bills lip sync per audio second. The rates below come from the public catalog and the shared price sheet read on 2026-10-09; the totals multiply them by 54 audio seconds.

Nine 6-second lines (54 audio seconds), as of 2026-10-09
Model and resolutionRate per audio secondArithmeticTotal
H3 Max Lip Sync 480p$0.062554 x 0.0625$3.375
H3 Max Lip Sync 768p$0.1054 x 0.10$5.40
Fabric 480p$0.1054 x 0.10$5.40
Fabric 720p$0.187554 x 0.1875$10.125

Why Fabric 720p costs almost twice as much

Sume lists Fabric at $0.1875 per audio second at 720p. The vendor's own model page lists 720p at $0.15 per second and 480p at $0.08 per second (read 2026-10-09). Sume's price is the provider list price times 1.25, so $0.15 x 1.25 = $0.1875 and $0.08 x 1.25 = $0.10. The H3 Max 768p rate of $0.10 sits at the same level as Fabric 480p, which means the real choice at that price is resolution and look, not cost.

One caution on the totals: the catalog says a job's reservation is stored in micro-dollars and rounded up to cents for the balance hold. A 6-second Fabric line at 720p is $1.125, so each hold shows $1.13. Nine holds can add up to a cent or two more than the plain product, and the captured amount is what the job result reports.

What each model needs from you

Both routes take a still image (or a ready avatar handle) plus an audio URL. For H3 Max Lip Sync, Sume's models page lists the audio window as 5 to 14.8 seconds, so every 6-second line fits. Anything shorter than 5 seconds has to go to Fabric. Every submit returns a job you poll at the status and result URLs described in Jobs and results.

  • Use the same still for all nine lines so the face and framing match across cuts.
  • Measure each line's duration from the audio file, and send it as duration_seconds.
  • Keep one lip-sync model for the whole run so the nine clips look consistent.

Pick by constraint

If the spot will play full screen on a phone, 720p or 768p is worth the money. If the nine clips are inset as small presenter windows, 480p of either model saves between $2.025 (H3 Max, 768p to 480p) and $4.725 (Fabric, 720p to 480p). Run one line at both resolutions first: 6 seconds costs $0.375 at H3 480p, so the test is nearly free.

Sources

Related posts

More in Media tools

All Media tools posts

Written by Sume