ACCURACY INVESTIGATION OF AUTOMATIC SPEAKER RECOGNITION FOR TELEPHONE SPEECH SIGNAL QUALITY

  • Ivan Jokić Fakultet tehničkih nauka, Univerzitet u Novom Sadu
  • Vlado Delić Fakultet tehničkih nauka, Univerzitet u Novom Sadu
  • Nikša Jakovljević Fakultet tehničkih nauka, Univerzitet u Novom Sadu
  • Milan Dobrović Telekom Srbija
  • Stevan Jokić Fakultet tehničkih nauka, Univerzitet u Novom Sadu
Keywords: Automatic Speaker Recognition, Mel – Frequency Cepstral Coefficients, Gaussian Mixture Models, Hidden Markov Model, HTK, ITU-T STL2005, ITU-T Recommendation G.729, echo in VoIP

Abstract

This work was performed by examining the accuracy of speaker identification on telephone quality voice signals. Implementation of the used speaker recognizer was performed using HTK. Influence of the considered telephone channels on transmitted voice signal is seen through its basic characteristics, types of the applied codecs and the effects caused by the condition of the transmission channel. These effects were observed by a factor of transmission error probability, while the VoIP telephone channels were analyzed and the appearance of echo. Simulation of the appropriate codecs and the probability of various errors made during transmission by using publicly available library of software tools, ITU-T STL2005, while the echo phenomenon was simulated using effect Delay / Echo-Simple suite Sony Sound Forge 9.0.
Published
2019-01-15
Issue
Vol 2010 No 36 (2010)
Section
Articles