تمييز جنس المتكلم باعتماد العزوم السبعة لاشارة الكلام

Abstract

In this research, a proposed approach has been adopted to
represent the speech signal in two-dimensional arrays to make it
suitable to evaluate the seven moments because the seven moments
have stability in rotation, scaling and translation, which was adapted in
many recognition fields.
Many samples have been adopted for both male and female
and a threshold value for both genders of the seven moments have been calculated after truncating the high frequencies from the speech
signal by applying Fourier transformation, Then finding the
correlation between the input samples and the threshold of both
genders. The proposed algorithm contains two phases, in the first one
the threshold values were built and the other phase works to test the
gender of spooking person by evaluating the correlated distance
between the speech parameters of the spooking person and the
threshold values which were got in the first phase.
The applied algorithm has given correct result up to 80% for
male samples and 85% for female ones.