How It Works
From audio capture to cloud monitoring — a four-stage pipeline.
Detection pipeline
1
Camera captures audio
Existing cameras stream audio through NX Witness VMS — no new cameras required
2
On-site AI inference
Cochl-certified hardware runs the AI model locally — raw audio never leaves the site
3
Event metadata sent to cloud
Only metadata (sound type, timestamp, camera ID) is transmitted — no audio
4
Dashboard + alerts
View real-time events and receive webhook / email notifications
System components
| Component | Role |
|---|---|
| VMS | Existing video management environment — NX Witness currently supported. Additional VMS platforms will be supported in future updates. |
| Cochl Plugin | Extracts audio streams from the VMS |
| Cochl Hardware | Runs AI inference on-site (required for paid plans) |
| Cloud Dashboard | Real-time event monitoring, history, and notifications |
For the Trial tier, Cochl-certified inference hardware is not required. NX Witness VMS and the Cochl plugin must still be installed.