De-Noising Process in Room Impulse Response with Generalized Spectral Subtraction

The generalized spectral subtraction algorithm (GBSS), which has extraordinary ability in background noise reduction, is historically one of the first approaches used for speech enhancement and dereverberation. However, not been applied to de-noise room impulse response (RIR) extend reverberation decay range. application GBSS this study stated as an optimization problem, that is, subtracting level from RIR while maintaining signal quality. process conducted measurements RIRs with artificial natural ambient aims determine optimal sets factors achieve best reduction results regarding largest dynamic range improvement. are set variables determined by estimated SNRs filtered octave band. acoustic parameters, time (RT), early (EDT), improvement energy curve were control measures evaluation criteria ensure reliability algorithm. de-noising compared compensation methods. With achieved factors, contributes a significant effect terms decreases estimation errors RTs caused levels.