Audio generation model by Sony AI
MMAudio
MMAudio is a video-to-audio model from the University of Illinois Urbana-Champaign, Sony AI and Sony Group that generates a synchronized soundtrack for a clip, guided by the video and a text prompt. It is the lowest-priced audio model in the catalog.
01
Price
Per second
Octomind bills the listed provider price plus 5%. Media generation runs on paid plans, per unit, with no minimum.
02
What real work costs
| Workload | Cost |
|---|---|
| 30-second sting 30 s of output | $0.06 |
| 3-minute track 180 s of output | $0.36 |
| One hour of audio 3,600 s of output | $7.20 |
At the listed price, before the 5% platform fee.
03
Specs
- Type
- Audio
- Billed
- Per second of output
- Output
- the input video with a generated soundtrack, 8 s default
- Accepts
- Text, Video
- Weights
- Open — CC BY-NC 4.0
- Released
- December 2024
- Model ID
mmaudio
04
When to pick it
Pick it if
- you need sound effects matched to an existing silent clip
- you want the cheapest per-second audio rate in the roster
- you pair it with a video model that ships without sound
Skip it if
- you need music or speech, where Eleven Music or MiniMax Speech fit
- you have no video to score
05
Use MMAudio in Octomind
Ask the agent for it by name in any Octomind Cloud session on a paid plan — it calls MMAudio through the generate_media tool and bills per second.
Generate a 30-second clip with MMAudio: …06
Compare MMAudio
07
Similar models
Nearest audio models by price.
View all models →08
MMAudio FAQ
How much does MMAudio cost?
MMAudio costs $0.002 / second. 3-minute track comes to $0.36. Octomind bills the listed provider price plus 5%. Media generation runs on paid plans, per unit, with no minimum.
What does MMAudio produce?
The input video with a generated soundtrack, 8 s default. It accepts text, video as input.
Is MMAudio free?
No. MMAudio costs $0.002 / second, and Octomind runs media generation on paid plans only, billed per unit with no minimum.
How do I use MMAudio?
In Octomind Cloud on a paid plan, ask the agent to generate with MMAudio; it runs the job and returns the file to your session.
Use MMAudio in Octomind
Ask the agent to generate it on a paid plan — media is created on request and billed per unit.