Our sound recognition is based on modern AI Deep Learning pattern recognition and audio signal analysis.
AI Sound Event Recognition
AmAISound is our AI Sound Event Recognition System which recognizes the sound events and their respective temporal start and end time in a recording.
Vast Array of Applications
Forced Entry Indicator
Breaking door, glass etc.
Talking when there should not be anyone.
Unauthorized People Activities
Moving Vehicle Detector
Drone, Cars, Tractor etc.
Confirmation of False Alarm
Smart fence is disturbed by construction vehicles.
Indication of Cyber Physical attack
Stuxnet centrifuge failure, power generator made to malfunction.
Anywhere Audio Processing Engine
Optimised for fast modern CPU instead of requiring GPU with the use of Single Instruction Multiple Data (SIMD) and AVX, optimized number libraries.
Use an existing server, just add a low cost microphone.
Use an embedded edge controller. Eg, Raspberry PI
Use AI at the edge, low cost Nvidia Nano.
• Sound event recognition is versatile and useful for physical security monitoring; e.g. glass shattering (forced entry) etc.
• Testing of key sounds showed that the technology is mature and is accurate.
• Many ways to implement cost effectively, e.g. augment CCTV
Many other innovative applications
Support existing CCTV cameras as they will typically support sound streaming.
Add voice recognition as low cost biometrics e.g. for door access, or mobile phone access control.
Support voice transcription and criminal investigation related to background voice.
Review videos for events to support AI vision.