Real-time spectrogram and signal waveform
The live spectrogram shows how sound changes over time, alongside a signal waveform. The interface lets the operator follow surrounding sounds and the acoustic detection process in one place.
Acoustic Detection System.
Drone detection through sound.
Acoustic Detection System is a TUKAN.army desktop application for acoustic drone detection. A microphone mounted on a parabolic dish captures surrounding sounds, which the application compares with a library of acoustic signatures to trigger a drone detection alarm. Sound analysis runs locally, without the cloud.
Drone recognition using acoustic signatures
The application compares captured sound with acoustic signatures from a library of around 2,700 recordings. Recognising a drone sound triggers an alarm to notify the operator.
Estimating the direction of captured sound
A sonar-style display shows the forward sector and left-to-right sectors. The “Maximum” indicator retains the strongest direction to help the operator follow changes in the sound.
Filtering noise in the spectrogram
An unwanted tone can be selected in the spectrogram and suppressed with a click. Filtering helps reduce noise such as persistent hum, electronic interference or wind noise.
Quiet mode for local background sound
Quiet mode learns the ambient sound at the location and uses it to filter the background. Processing remains within the local application.
Recording library and feedback
The sound library supports WAV imports and selected audio excerpts from YouTube. Yes/No alarm feedback helps progressively refine recognition.
Dataset packs and AI model extensions
The application supports expandable dataset packs and plug-in AI models. These provide a basis for further development of the library and recognition capabilities.
Tested in real-world conditions
Tested with real drones in everyday operating conditions, using different microphones and parabolic dishes.
