2017
3DSND
A 32-speaker ambisonic system
3DSND is a low-latency ambisonic rendering system built around 32 distributed amplifier nodes arranged in a six-metre circle. Each node is a Raspberry Pi running JACK audio, networked together to behave as a single spatial loudspeaker — so a sound source can be placed at any point in the listening area rather than being mapped to a fixed channel.
Where traditional stereo and surround systems pin sound to specific speaker positions, ambisonic rendering treats the room itself as the playback medium. That makes it well suited to music production, film post-production, immersive performance and VR — anywhere the spatial relationships between sounds matter as much as the sounds themselves.
The system is interactive. The sound field can be steered by hand through a Kinect sensor or by an HMD inside a Vive scene, so the listener moves through the audio rather than the audio being played at them. Because the architecture is distributed, the array scales: more nodes, larger circle, same software stack.
The system was presented at Sonar+D Hong Kong in 2017 and 2018.