Omniphony - High performance Spatial Audio Engine with MPV integration

[Glider95]

Omniphony is a high-performance, universal spatial audio engine designed for professional and enthusiast speaker environments. Rather than relying on closed-source hardware, Omniphony handles complex spatial rendering entirely in software.

By pairing Omniphony with an external plugin like harletty-bridge, it can be used to decode TrueHD bitstreams, extract object metadata, and render spatial objects directly to your specific speaker layout using VBAP. The system outputs seamlessly to PipeWire, ASIO, or file—all from a lightweight Rust binary.

Key Technical Highlights:

• Performance: Decodes over 130,000 frames and renders 15 spatial objects to 7.1.4 PCM with 48kHz precision.
• Visual Studio: Includes a 3D interface (Tauri + Three.js) to monitor sound object positioning, speaker heatmaps, and OSC remote control in real time.
• Customization: Supports layouts from 2.0 to 9.1.6, featuring multi-band crossovers, room ratio modeling, and distance attenuation.
• Architecture: A pure-Rust core with a flexible plugin system for adding codecs via .so or .dylib bridges.

Details & Source:

• Platforms: Linux, macOS, Windows
• GitHub: https://github.com/Mgth/Omniphony & and directly integrated with MPV: MPV Omniphony
• License: GPL-3.0

If you are looking for a flexible, commodity-hardware solution for real-time spatial audio rendering, Omniphony provides the framework to build it.

This project is not from me, but from @mgth ! I am just helping to spread the word !

 

[mgth]

Thanks for the message. And for everyone here: please keep in mind that the software is still at an early stage, although it is already fully functional. Right now, we are mainly focused on distribution, first-install issues, and default configuration. So if you run into any trouble getting it to work, your feedback would be very valuable to us.

 

[Glider95]

I am surprised that this does not pick up more !

I need to correct that the speaker layout is actually customizable to everyone's liking - even to 80.3.56 if you have the budget.
As well, this software is able - with the right bridge - to decode Dolby TrueHD in real-time which I think is a first ever?

 

[djtwxj]

Hi,

How to use this tool decode to stdout in macos？

 

[djtwxj]

Thanks for the message. And for everyone here: please keep in mind that the software is still at an early stage, although it is already fully functional. Right now, we are mainly focused on distribution, first-install issues, and default configuration. So if you run into any trouble getting it to work, your feedback would be very valuable to us.

Hi,

can this support macos in future?

 

[mgth]

Yes, I think nothing against that. Only that I don't have such a device. I might ask Claude to try a build. I think the mpv version is simpler, because it does not imply to deal with audio device. Why do you need stdout ?

 

[djtwxj]

Hi,
i want to build a wireless dolby atmos system. if the rendered pcm audio direct to the wireless system. i think will be less delay. so need your tool to output to std not soundcard

 

[mgth]

If you use it within mpv, you can output PCM from there. If you use it standalone, we’ve added a file output option that you may want to try, but it isn’t well tested yet.

 

[kemmler3D]

This looks pretty incredible, unfortunately I am still stuck in 2-channel world.

 

[mgth]

This looks pretty incredible, unfortunately I am still stuck in 2-channel world.

Thank you, with two channels you still can try binaural mode with headphones. It's still early, but with a smartphone as a head tracker it can do the trick.