Auditory Time-Frequency Masking: Psychoacoustical Data and Application to Audio Representations

Abstract

In this paper, the results of psychoacoustical experiments on auditory time-frequency (TF) masking using stimuli (masker and target) with maximal concentration in the TF plane are presented. The target was shifted either along the time axis, the frequency axis, or both relative to the masker. The results show that a simple superposition of spectral and temporal masking functions does not provide an accurate representation of the measured TF masking function. This confirms the inaccuracy of simple models of TF masking currently implemented in some perceptual audio codecs. In the context of audio signal processing, the present results constitute a crucial basis for the prediction of auditory masking in the TF representations of sounds. An algorithm that removes the inaudible components in the wavelet transform of a sound while causing no audible difference to the original sound after re-synthesis is proposed. Preliminary results are promising, although further development is required.

Designing Gabor windows using convex optimization

“…Such modification is ubiquitous in signal processing, e.g. in denoising [54], signal detection [55], time-stretching and pitch-shifting [56,57,58,59], modification of the spectrogram [60,61], irrelevance filtering [62,63], speech recognition [64], to name a few. Furthermore, jointly optimizing support and smoothness of the synthesis window can provide block-processing algorithms with reduced delay and blocking artifacts, possibly even at lower redundancy than usual.…”

Frame Theory for Signal Processing in Psychoacoustics

“…For more results and discussion on the origins of masking the interested reader is referred to e.g. [32,62,64].…”

“…To best predict masking in the time-frequency decompositions of sounds, it seems intuitive to have data on the time-frequency spread of masking for such elementary atoms, as this will provide a good match between the masking model and the sound decomposition. This has been investigated in [64]. Precisely, spectral, forward, and time-frequency masking have been measured using Gabor atoms of the form s i (t) = sin(2πξ i t + π/4)e −π(Γt) 2 with Γ = 600 s −1 as masker and target.…”

“…The masker level was fixed at L M = 80 dB. The target was defined by s T (t + ∆T ) = [64] illustrated in the time-F ERB plane. The gray circle symbolizes the masker atom s M (t).…”

A Perceptually Motivated Filter Bank with Perfect Reconstruction for Audio Signal Processing

“…To further reduce the redundancy of the AUDlet representation and improve its perceptual relevance, future work includes introducing perceptual sparsity in the transform domain. Specifically, based on the perceptual irrelevance filter proposed in [5] and recent data on auditory TF masking [34], a binary mask will be computed and applied to the sub-channel coefficients in order to re-synthesize only the audible TF components. Furthermore, future work will focus on how to combine the AUDlet FB and knowledge of TF masking to possibly improve audio codecs.…”