Pitch frequency separation method based on a frequency demodulation

Author(s):  D.D. Poleshenkov, no, no, Saint Petersburg National Research University of Information Technologies, Mechanics and Optics, St. Petersburg, Russia, d.poleshenkov@yandex.ru

O.O. Basov, Dr., associate Professor, Saint Petersburg National Research University of Information Technologies, St. Petersburg, Russia, oobasov@mail.ru

Issue:  Volume 46, № 2

Rubric:  Infocommunication technologies

Annotation:  This paper gives a pitch frequency separation method based on experimental data. This method allows simplifying a solution of a speech signal frames classification problem and increasing accuracy of pitch detection algorithms functioning. It has an increased accuracy and decreased computational complexity in comparison with the same algorithms. This method is based on a frequency demodulation process of a speech signal in a pitch band. The analysis of long-spoken vocal phonemes showed that the result of a pitch detector functioning is equal with the frequency demodulation result of the first harmonic band of a speech signal. The algorithm realizing this method was created. The description of proposed algorithm functioning is given in this paper. Primary examination of the proposed algorithm was made in this work. This paper gives the basic research ways of a proposed method improvement. In addition the description of a practical realization ways of a proposed pitch detection method based on frequency demodulation process was given in this paper.

Keywords:  speech signal, pitch frequency, frame, phoneme transition, frequency demodulation.

Full text (PDF):  Download

Downloads count:  213