A Perceptual Model for Sinusoidal Audio Coding Based on Spectral Integration

Abstract

Psychoacoustical models have been used extensively within audio coding applications over the past decades. Recently, parametric coding techniques have been applied to general audio and this has created the need for a psychoacoustical model that is specifically suited for sinusoidal modelling of audio signals. In this paper, we present a new perceptual model that predicts masked thresholds for sinusoidal distortions. The model relies on signal detection theory and incorporates more recent insights about spectral and temporal integration in auditory masking. As a consequence, the model is able to predict the distortion detectability. In fact, the distortion detectability defines a (perceptually relevant) norm on the underlying signal space which is beneficial for optimisation algorithms such as rate-distortion optimisation or linear predictive coding. We evaluate the merits of the model by combining it with a sinusoidal extraction method and compare the results with those obtained with the ISO MPEG-1 Layer I-II recommended model. Listening tests show a clear preference for the new model. More specifically, the model presented here leads to a reduction of more than 20% in terms of number of sinusoids needed to represent signals at a given quality level

On perceptual distortion minimization and nonlinear least-squares frequency estimation

“…Recently, it has been shown that significant improvements are gained by taking spectral integration into account [30], [31]. Using the masking model proposed in [30], [31], which was derived specifically for sinusoidal coding, the distortion D for a particular segment can be written as…”

“…A recently published psychoacoustic masking model for audio coding has been shown to form a distortion measure [30], [31], and this distortion measure has been applied successfully to the sinusoidal estimation problem in [15], [23], [32], [33]. Based on this we define the perceptual frequency estimation problem and its optimal solution.…”

A Low-Complexity Spectro-Temporal Distortion Measure for Audio Processing Applications

“…A mathematical expression approximating the threshold in quiet can be found in [1]. For the auditory filter bank the same gammatone-based approach as in [7] is used. In total 64 filters are used where the center frequencies are linearly spaced on an ERB-scale between 0 and Hz, where denotes the sample rate.…”

“…Van de Par et al introduced a perceptual distortion measure, which we will refer to as the Par-model, including spectral integration [7]. That is, the detectability of a specific frequency component is also determined by off-frequency auditory filters.…”

“…Moreover, it does not need to separate the signal into tonal and noise maskers. It has been shown that the Par-model leads to better coding results compared to the MPEG-1 model for various fixed bit-rates in the field of sinusoidal coding [7]. In addition, the Par-model is defined as a mathematical norm, which allows for incorporating perceptual properties in least squares optimization algorithms.…”