AEGIS · Acoustic intelligence
Crowds, machines and streets leave patterns in ordinary microphone audio. AEGIS is the engine we are building to read those patterns, with no cameras and no special sensors, and turn them into signals people and machines can use. It is in research.
The engine
One pipeline underpins every application. The output changes by market; the listening does not.
Ordinary microphone audio — a phone, a stadium feed, a microphone already on the plant floor.
Spectral analysis against a baseline that adapts to each environment.
A trained model separates the patterns that matter from ambient sound.
A haptic cue, a machine alert or a marked crowd passage — one engine, a different output per market.
Applications
Each market carries its stage in plain words. Nothing here is described as finished before it is.
A phone-native awareness layer for blind and low-vision pedestrians, designed to flag crowd surges, quiet vehicles and a changing sound environment. It supplements, and never replaces, a cane, a dog guide or orientation and mobility skills.
Acoustic bearing-fault detection for manufacturing, using microphone audio on hardware a plant already owns. It has not yet been tested on a running production machine, and we make no accuracy claim until it has.
Identifying the crowd-dominant passages in broadcast audio. It has been tested on recorded match audio. It does not call goals, and it doesn’t yet separate the crowd from the commentary.
Work with us
We are looking for transit and accessibility organisations, manufacturers and broadcasters that want to test AEGIS on their own audio. Every conversation starts with the founder.