Identification of predominent frequencies in a speech signal using modeling of vocal chord

摘要

A great problem in speech communication is to identify the shape and characteristics of the vocal apparatus. This task is normally done by using a vocal tract model, based on the calculation of the function, area would be performed. We will show that these models have good performance in experiments. A mathematical model of voice box has been obtained. We suggest a new numerical method without saturation, which allows solving the problem in a wide range of frequencies with an arbitrary and controlled precision. The low sound frequencies cause the detection of the membrane at the upper part of the cochlea, whereas high sound frequencies cause the detection of the main volute of the cochlea.