Gender Identification Based on Speech Signal | ||
Buhuth Mustaqbaliya Scientific Periodical Journal | ||
Article 1, Volume 1, Issue 1, December 2014, Pages 151-170 | ||
Abstract | ||
Abstract In this paper algorithm has been proposed to identify the speaker's gender based on the signal of her sound. This algorithm relied on deriving the features of speaker's signal in order to gain a representation of this signal and to get benefit from those features in establishing data base that includes the identity of the speaker as well as the drawn features and her gender. This paper present an algorithm confirm the speaker’s gender. The Fourier transformation is adopted here in order to derive some of the factors that depend upon the Fourier transformation factors where those factors have been adjusted to gain an equality of the 15th degree that is too close to the speaker's signal after transforming it to Fourier transformation. The practical application reveals that the amount of the approach of the derived equality is not affected by the equality degree of the polynomial after the 15th degree. This algorithm has been applied on a sample of more than 50 speakers of different groups of age of the two genders where three words of each speaker have been gained then those words are transformed to Fourier transformation where the rate is calculated after adjusting the signal into polynomial of 15th degree. The derived values will be stored after arranging them in decreasing way to be features by which the speaker's gender can be recognized. The practical application of the proposed algorithm resulted in getting on distinction ratio approaches to 72% out of 20 samples. The approaching factor is employed to determine has to do with the distinction ratio. | ||
Statistics Article View: 142 PDF Download: 86 |