Single channel AI speech reverberation time modification for room dimension matching

Reverberation is caused by room impulse reflection from walls, ceilings and other plane surfaces contributing an additive echo proportional to the geometry of the environment. Reverberation in speech plays a role in interpretation, giving listeners a sense of the size, shape and materials of the room. In an immersive VR/AR scenario where a speaker is presented in a different virtual environment to their actual environment, a mismatch between the recording and purported room geometries causes psychoacoustic dissonance. This paper presents an AI-based algorithm to estimate and remove room reverberation impulse effects from speech recorded in a source environment, and add reverberation modeled on the target environment. Effectiveness is measured via psychoacoustic assessment.