Academic and Open Source Ambisonic Resources

Spatial Audio is hot! There is a bunch of academic research and open source software implementing higher order Ambisonics. These are my notes, with some comparisons of DSP + programming platforms I have evaluated.

ChucK

A strongly timed strongly typed language built upon the RtAudio C++ library. All code executes in a virtual machine and is sample accurate.

Pros

• Small C++ codebase
• Compiles quickly and easily on Linux
• Runs in a web browser
• Runs on mobile devices
• Has plugin system
• Runs in video game engines
• AI Stuff!

Cons

• Ambisonic plugins are slim to none
• C++ codebase
• The virtual machine is 4400 lines of C++, written in 2002

SuperCollider

A DSP engine and object oriented programming language for sound synthesis.

Pros

• Has IEM encoder/decoder up to 3rd order!
• Has an Ambisonic Tool Kit
• Huge international user community
• Supports higher order and single order Ambisonics
• Built in Binaural decoding with measured and synthetic HRTFs
• Packaged in Ubuntu Studio through the sc3-plugins package
• There is a whole class online with a week dedicated to Ambisonics

Cons

• Deep roots in academic language
• Complex architecture
• New jargon beyond yaw/pitch/roll, i.e. incidence, quality, gain. Yikes!
• HRTF data for IEM plugins is pretty old

Pyo

A framework for DSP implemented in Python. It has a Binaural generator that looks pretty simple. Has an advantage of being implemented in the same language as the GenAI text-to-speech I’m looking to use.

Pros

• The Binuaral encoder/decoder does all the HRTF stuff internally
• The example code works first try
• Integrates with the TTS stuff I already know
• Built in GUI classes
• Could integrate something in TouchDesigner to do OSC for x/y/z + distance

Cons

• JACK audio output not built into wheel, requires a custom build
• Python isn’t gonna run on a phone or a watch
• Pyo uses wxPython which is…interesting