An Analysis/Synthesis System of Audio Signal with Utilization of an SN Model

Abstract

An SN (sinusoids plus noise) model is a spectral model, in which the periodic components of the sound are represented by sinusoids with time-varying frequencies, amplitudes and phases. The remaining non-periodic components are represented by a filtered noise. The sinusoidal model utilizes physical properties of musical instruments and the noise model utilizes the human inability to perceive the exact spectral shape or the phase of stochastic signals. SN modeling can be applied in a compression, transformation, separation of sounds, etc. The designed system is based on methods used in the SN modeling. We have proposed a model that achieves good results in audio perception. Although many systems do not save phases of the sinusoids, they are important for better modelling of transients, for the computation of residual and last but not least for stereo signals, too. One of the fundamental properties of the proposed system is the ability of the signal reconstruction not only from the amplitude but from the phase point of view, as well.