AI and machine learning
Stereo vision lab
Place two virtual cameras in a 3D scene, change baseline and focal […]
Sound source localization
Find where a sound comes from using the delay between microphones, the audio cousin of stereo vision. A simulated room in the browser, with a model that learns to estimate direction.
A small version
A small version of the project. The full one is linked below.
How it works
- 01A sound reaches each microphone after a delay of distance divided by the speed of sound. The demo builds the signal each microphone would record, with noise added, in the frequency domain where a delay is just a phase shift.
- 02It then measures the delay between microphones with GCC-PHAT: multiply one spectrum by the conjugate of the other, keep only the phase, and transform back. The peak sits at the delay. This is the same idea as the disparity in the stereo lab, in time instead of pixels.
- 03With three or more microphones each delay says how much farther the source is from one microphone than from another, and the demo finds the point in the room that fits all of them. Add noise and watch the peak blur and the estimate drift.
What it will include
- 01Simulated 2D room where you place sources and a microphone array.
- 02Time difference of arrival computed and drawn live.
- 03A small neural network that estimates direction, compared with the geometric answer.
- 04Optional: use your own microphone.
More in this category
AI and machine learning
3D reconstruction from images
From a handful of photos to a point cloud and a navigable 3D scene […]
AI and machine learning
HARNAIS SSI
A cybersecurity harness for AI coding agents such as Claude Code […]