A model that gives a silent video sound based on what happens in it: a door slams, a car passes, a wave breaks. Use it when a video model returned a clip without audio, because silence kills watch time in ads.
| What it does | sound effects from the clip content |
|---|---|
| Description | optional |
| Source speech | kept |
| Price | $0.02 a second |
| Model | How it works | 5 seconds | Limit |
|---|---|---|---|
| Mirelo SFX 2.0 | from the picture | $0.10 | none |
| ElevenLabs Sound Effects V3 | from your description | $0.02 | clips up to 14 s |
The model works out what should be heard; a description is optional.
If someone talks in the clip, the voice stays and effects are added around it.
PixVerse C1, Seedance Mini and Wan Animate return silent video — Mirelo fixes that.
No. A description helps when you need a specific character of sound.
Mirelo listens to the picture; ElevenLabs makes sound strictly from your text, cheaper, but for clips up to 14 seconds.
The source audio track is kept and new sounds go on top.