Skip to content


Intelligent Low-Footprint Sound Event Detection and Localization for Automotive Applications

A human driving in the traffic is constantly exposed to sounds that enhance its environmental awareness, and that can direct the driver’s attention to specific events that require its active response. These sounds include for example emergency sirens, horns, crashes or passing vehicles. Microphone arrays can be equipped on autonomous vehicles to exploit these acoustic cues to increase the safety of self-driving cars. In this context, the car should be able to:

  • Analyze the acoustic scene to detect occurrences of specific sounds
  • Identify the direction of arrival of these sound events

The ESR project aims to develop low-power algorithms to solve these problems in both active (i.e. when the car is driving) and passive (i.e. when the car is parked) mode, and to ensure robustness to the noisy and highly dynamic road environments encountered in a real-world scenario. The algorithms will be deployed on embedded hardware for the automotive industry.

Research Team
Early Stage Researcher: Stefano Damiano
Host Institution: KU Leuven
Supervisors: Toon van Waterschoot, Andre Guntoro