Phase spectrum prediction of audio signals
Ali Bahrami Rad, Tuomas I. Virtanen · 2012
Modeling the phases of audio signals has received significantly less attention in comparison to the modeling of magnitudes. This paper proposes to use linear least squares and neural networks to predict phases from the neighboring points only in the phase spectrum. The simulation results show that there is a structure in the phase components which could be used in further analysis algorithms based on the phase spectrum.