Several psychoacoustic models used in perceptual audio coding take into account the difference in masking effects of tones and narrowband noise and therefore incorporate some kind of tonality estimation. For possible optimization of these estimators for time-varying signals it is desirable to know the duration above which the auditory system is able to discriminate between tone bursts and narrowband noise bursts. There is some knowledge that this duration is frequency, bandwidth, and loudness dependent, but up to now no systematic studies have been performed. Therefore this paper presents the setup and the results of experiments for determining frequency dependent thresholds.
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.