How it works

From acoustic signature to confirmed alerts in seconds

Trained on powerful computers. Deployed on the phone. Detection works without internet.

Evidence-based
Human-verified
Privacy-first
Viet Nguyen, Chief Engineer AI ML Sound

Privacy

Audio stays on the phone by default

Privacy Promise

  • •No one listens in
  • •Detection runs locally on the phone
  • •No continuous recording or audio streaming
  • •Incident clips saved for audit only when your organization enables it

On-Device Detection

Policy-controlled uploads only

SplitSec detects gunshots on your phone, not in the cloud. We do not continuously record or stream audio, and no one listens in. If your organization enables it, incident clips from gunshot detections may be uploaded and saved for audit. If a user reports a false alert, that clip is anonymized to help improve accuracy. We do not sell data or build products that track people.

Behind the Scenes Engineering

Inspired by how the human brain works

SplitSec AI learns sound patterns the way people recognize a familiar song or voice. We train our models on thousands of real world examples using high powered computers, then store a compact version, the “brains and memory,” directly on your phone. That means fast gunshot detection, even offline, with privacy by design.

Inspired by how the human brain works
Trained on thousands of sounds
Compact AI runs on your phone
Book 30 minutes