MAVS

The paper presents LAVSS, a location-guided audio-visual spatial audio separator. Existing monaural audio-visual separation (MAVS) methods often overlook the location of the sound source, which is crucial in VR/AR scenarios….