3D Audio
This report outlines work done for the 2019 Audio Experience Design Installation at Imperial College London. My contribution to this installation involved:
- Generating a mixed synthetic and real 3D audio dataset with over 2000 audio clips
- Training a CNN to make heading predictions based on 2 channel audio vectors
- Utilising a probabilistic filter to smooth heading predictions
- Creating an interactive display, with real-time audio input, and graphic output
The final result of work was: an interactive installation, which utilized live binaural audio data to make sound source heading predictions.
Click for Project Documentation:
https://de3-audio.readthedocs.io/en/latest/intro.html
Skills:
CNN
Signal Processing
Arduino

