Developed with SuperCollider, the Sonic Mirror software instrument performs real-time digital audio recording, analysis, and synthesis. Machine listening algorithms extract features from detected sounds (e.g. loudness, pitch, duration, timbre), and based on the analysis, the system dynamically renders a soundscape from recorded sounds using a variety of audio synthesis techniques (e.g. pitch shifting, time stretching, granular and concatenative synthesis).
This is an ongoing project, and an active development branch of the software can be found here.
Current embedded prototypes feature the Sonic Mirror software operating on a Raspberry Pi, connected to a microphone, digital sound card, and speaker, embedded within a wooden laser-cut enclosure.
In practice, the Sonic Mirror can manifest in a open-ended variety of configurations beyond this singular enclosure design as long as a computer running the software is connected to a microphone and speaker system.
The name and functionality behind this project is inspired by a concept for a cybernetic environmental sound installation initially conceived of by composer and bioacoustic researcher David Dunn (p. 11).