Privacy-respecting detection and localization of sounds in autonomous driving applications

A method comprising:

• obtaining, using one or more audio detectors of a vehicle, a first sound recording of a sound in a driving environment of the vehicle;
• generating, using a processing device and the first sound recording, a second sound recording, wherein the second sound recording renders imperceptible a speech present or potentially present in the first sound recording;
• processing, using the processing device, the second sound recording to obtain an indication of presence of a sound-producing object in the driving environment of the vehicle;
• and
• causing, by the processing device, a driving path of the vehicle to be modified in view of the indication of presence of the sound-producing object.

A system comprising:

• a sensing system of a vehicle, the sensing system comprising one or more audio detectors to:
• obtain a sound recording comprising a plurality of elemental sounds (ESs) generated in a driving environment of the vehicle;
• and
• a perception system of the vehicle, wherein the perception system is to:
• process, using a sound separation model, the sound recording to generate, for each ES of the plurality of ESs, a representation of a respective ES;
• identify a content of a first ES of the plurality of ESs using a representation of the first ES;
• and
• cause a driving path of the vehicle to be modified in view of the identified content of the first ES.