Comparison of linear predictive analysis methods for ANN-based speaker identification

Inal M. M., Bütün E., ERKAN K., Yildirim M., Çeken C.

5th Seminar on Neural Network Applications in Electrical Engineering, Belgrade, Sırbistan Ve Karadağ, 25 - 27 Eylül 2000, ss.109-112 identifier identifier

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.1109/neurel.2000.902395
  • Basıldığı Şehir: Belgrade
  • Basıldığı Ülke: Sırbistan Ve Karadağ
  • Sayfa Sayıları: ss.109-112
  • Kocaeli Üniversitesi Adresli: Evet

Özet

This paper studies the parameterization of speech using Linear Predictive (MP) analysis and its application to speaker identification. The task of speaker classification is realized using Multilayer Perceptron (MLP). The system is evaluated on speaker set for speaker identification. The system is also tested with cepstral and reflection coefficients and results are compared.