VOICED-UNVOICED CLASSIFICATION OF SPEECH USING AUTOCORRELATION MATRIX


Senturk Z., Yetgin O. E. , SALOR DURNA Ö.

22nd IEEE Signal Processing and Communications Applications Conference (SIU), Trabzon, Türkiye, 23 - 25 Nisan 2014, ss.1802-1805 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Cilt numarası:
  • Doi Numarası: 10.1109/siu.2014.6830601
  • Basıldığı Şehir: Trabzon
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.1802-1805

Özet

In this paper, a fast method for voiced-unvoiced classification of speech signals is introduced. The suggested method makes the V-UV decision, using signal energy, the peak-to-peak differerence of the autocorrelation function, number of zero crossings of the autocorrelation function and the unit delay autocorrelation coefficient all together. This method has been tested on speeches of three speakers, one woman and two men, which include both the speech waveform and the laryngograph signal in stereo form. Having labeled the speech using the laryngograph signal manually, comparison of the hand-labelled decisions and those of the proposed method is achieved. The accuracy of the proposed method is found to be 100% for woman and 98% for men.