Application of a Threshold Methods for Compression of Vocal Signals

Abstract

Types of thresholds used for compression of vocal signals are considered. It is shown that quality of the restored speech depends straight on the choice of threshold characteristic. The mathematical description of spectral composition of the signal subjected to threshold is given, its connection with the size of the threshold is revealed and the nonlinear distortion coefficient is calculated. The graphs of revealed dependencies on which it is possible to judge on type selection and threshold level are built.