Low-complexity sinusoidal component selection using loudness patterns

Abstract

Sinusoidal modeling of audio at low-bit rates involves selecting a limited number of parameters according to a quantitative or perceptual criterion. Most perceptual sinusoidal component selection strategies are computationally intensive and not suitable for real-time applications. In this paper, a computationally efficient sinusoidal selection algorithm based on a novel hybrid loudness estimation scheme is presented. The hybrid scheme first estimates efficiently the loudness of a multi-tone signal from the loudness patterns of its constituent sinusoidal components. Then it refines this estimate by performing a full evaluation of loudness but only in select critical bands. Experimental results show that the proposed technique maintains a low perceptual sinusoidal synthesis error at a much lower computational complexity.

Sinusoidal component selection based on partial loudness criteria

“…• Firstly, the PLSS algorithm makes use of the partial loudness patterns instead of the excitation/loudness patterns used in the existing techniques [6,9,10].…”

“…This process is combinatorial in nature with O N L combinations and is therefore computationally intensive. Therefore, several iterative algorithms [6,9,10] have been proposed to carry out the sinusoidal component selection task in a computationally efficient manner.…”

“…General structure of a sinusoidal component selection task. [6,7,9,10] have been proposed in the literature. In [7], the signal-to-mask ratio (SMR) based criteria was suggested to select the components.…”

“…This process is computationally expensive and not adequate for practical applications. To this end, a few computationally efficient alternatives [10,11,12] were proposed. In [11], a pruning approach was described to evaluate the auditory model stages in a computationally efficient manner.…”

“…In [11], a pruning approach was described to evaluate the auditory model stages in a computationally efficient manner. In [10], a hybrid approach to loudness estimation for sinusoidal signals was proposed to speed up the sinusoidal component selection task. Specifically, they speed up the auditory model evaluations that are carried out repeatedly in every iteration.…”