Many video models return clips without sound, and silent videos get scrolled past. There are two ways to fix it here: a model picks sounds from what is visible, or makes exactly the sound you described.
| Tool | How it works | 5 seconds | Limits |
|---|---|---|---|
| Mirelo SFX 2.0 | from the picture, text optional | $0.10 | — |
| ElevenLabs Sound Effects V3 | strictly from text | $0.02 | clips up to 14 s |
| Audio from clip | restores the source audio | free | needs the source clip |
Mirelo SFX 2.0 works out what should be heard: doors, cars, water, footsteps.
ElevenLabs Sound Effects V3 makes the specific sound you describe, for clips up to 14 seconds.
If you swapped a face in your clip and the model dropped the speech, Audio from clip brings it back for free.
Yes, both models add effects on top and leave speech alone.
These models make sound effects and ambience. Take music from the platform’s library.
Veo 3.1, Gemini Omni, Kling 4.0 Flash and MiniMax H3 generate sound themselves.