All projects
Air traffic control
Telling overlapping transmissions apart
Detection that spots when two aircraft transmit at once, built on audio features modelled on human hearing and tuned to work on a fraction of the signal.
Results
- fewer wrong decisions, even in heavy noise
- 99%
- of audio needed, down from 1,000 ms
- 300 ms
- simultaneous calls, up from 200
- 600
The challenge
Air traffic control systems needed to detect when several aircraft transmit at the same time, which the existing system couldn't do reliably.
What we built
- A detection algorithm built on audio features modelled on human hearing
- Custom voice detection so only useful audio is processed
- Memory optimisations and reuse of already computed features
Built with
- Python
- C
- Signal processing
Have something similar in mind?
Tell us what you're working on and we'll come back with a clear next step.