30-second holiday UGC ad with a music bed: a $7.59 end-to-end receipt
Avatar Plus 30 s $7.35, voice detach $0.01, Lyria bed $0.13, Timeline mix $0.10. One finished UGC ad with a ducked music bed costs $7.59 on Sume.
One finished 30-second holiday UGC ad with captions and a ducked music bed costs $7.59 on Sume: $7.35 for the avatar video at Plus quality, $0.01 to detach its voice, $0.13 for the music bed and $0.10 for the Timeline mix. Captions are burned in during the avatar job and add no separate charge.
The ad has four steps because a music bed has to be mixed under the voice, and the mix is a Timeline job whose audio spine is the voice.
The four steps
Step 1 renders the avatar video with inline captions. Step 2 detaches its audio as a wav, so the voice becomes a reusable file. Step 3 generates the bed. Step 4 renders a Timeline with the voice as the spine, the avatar video as the one video slot, and the bed as the soundtrack with duck_db set.
| Step | Surface | Rate | Cost |
|---|---|---|---|
| Avatar video, Plus, 30 s, captions inline | talking-video | $0.245 per s | $7.35 |
| Detach voice track to wav | audio detach | $0.01 per job | $0.01 |
| Instrumental bed | Music Router | $0.125 per generation | $0.13 |
| Mix and render, 1 minute | Timeline 1.0 | $0.10 per output minute | $0.10 |
| Total | $7.59 |
Details that cost a retry
Timeline accepts only Sume-hosted media, so the avatar's output and the detached wav are already eligible, while a bed you made elsewhere must be imported first. duck_db needs a real audio spine, which the detached voice is. Timeline's default output is 1080×1920, while avatar video is 720p, so the clip is scaled to fit under the default fit: cover.
Because captions were burned into the avatar video in step 1, the Timeline render keeps them. If the caption stage soft-fails, the job still returns a clean video with captions.status set to failed, and you can add captions later with the $0.20 standalone job.
Scale it to a test
Ten different hooks, each as a 30-second ad with its own bed, cost 10 × $7.59 = $75.90. Keep the bed constant across hooks if the test is about the script, since you then save $1.17 against one bed per hook: reuse one bed and the total is 10 × ($7.35 + $0.01 + $0.10) + $0.13 = $74.73.
When to skip the bed
Most social platforms add their own sound library, and many UGC-style ads perform without music. If the ad runs as a paid placement with sound on, a quiet bed under the voice helps. If your test is about hook wording, skip the three extra steps and pay $7.35 per ad. The $0.24 for the bed, detach and mix is small, but the extra steps are where jobs fail.
Sources
Related posts
More in Use cases
- 30 holiday product clips on Gemini Omni Flash 1.1: $30, $45 or $90
Price a 30-SKU holiday product video batch on Gemini Omni Flash 1.1 at 720p, 1080p and 4K, with the per-second rates and the rounding shown.
- A $50 AI video ad test budget split across four jobs on Sume
A $50 test can cover 40 draft hooks, 6 hero shots, 4 spoken clips and 2 4K cuts on Sume for $29.70, leaving $20.30 for re-runs. Prices per job below.
- 50 gift SKUs: cutout plus a 5-second clip for $33, or $17.50 at 480p
Remove the background ($0.03) and render a 5-second clip per SKU. Omni Flash at 720p totals $33.00 for 50 gift SKUs; Wan 3.0 at 480p totals $17.50.
- 60-second AI video ad from 10-second shots: hero and filler cost
Six 10-second shots, two on Seedance 2.5 at 1080p and four on Wan 3.0 at 480p, make a 60-second spot for $30.96 on Sume. All six on 2.5 would be $85.32.
Written by Sume