Audibility threshold of auditory-adapted exponential transfer-function smoothing (AAS) applied to loudspeaker impulse responses

A reverberant acoustical system's transfer function may show deep notches or pronounced peaks, requiring large linear amplification in the play-back system when used, for example in auralization or for convolution reverb. It is common practice to apply spectral smoothing, with the aim of reducing spectral fluctuation without degrading auditory-relevant information. A procedure referred to as auditory-adapted exponential smoothing (AAS) was proposed earlier, adapted to the spectral properties of the hearing system by implementing frequency-dependent smoothing bandwidths. This contribution presents listening experiments aimed at determining the audibility threshold of auditory-adapted exponential smoothing, which is the maximum amount of spectral smoothing allowed without being audible. As the results depend on the specific acoustic system, parametrization guidelines are proposed.