PrismAudio
PrismAudio is the first framework to integrate reinforcement learning into video-to-audio (V2A) generation, equipped with a dedicated Chain-of-Thought (CoT) planning mechanism.
text → audio · made by FunAudioLLM
Sold by
Nobody in the catalogue publishes a price for this yet.
About
PrismAudio — an audio model from FunAudioLLM.
It takes text and returns audio. It was published in March 2026. The catalogue files it under speak. Its weights are published under MIT, so you may run it on your own machine, or buy it from whoever serves it cheapest.
Every current figure
- Maker
- FunAudioLLM
- Register
- model
- Takes
- text
- Returns
- audio
- Published
- March 2026
- Licence
- mit
Known as 1 name
FunAudioLLM/PrismAudio