Speech recognition with principal component analysis


Creative Commons License

Özkaynak E. , Görgünoğlu S., Bayır Ş.

2nd World Conference on Information Technology (WCIT-2011), Antalya, Türkiye, 22 - 27 Kasım 2012, ss.354-360

  • Basıldığı Şehir: Antalya
  • Basıldığı Ülke: Türkiye
  • Sayfa Sayıları: ss.354-360

Özet

Various researches on speech recognition are carried out nowadays; moreover, hardware and software are developed on this subject. Operations can easily and rapidly be carried out by establishing interaction between computer and human via verbal communication. Therefore, studies related to speech recognition remain important today. Various algorithms have been used in speech recognition processes. Principal Component Analysis was utilised in the recognition of verbal statements within the scope of this study. Initially, eigenvalues and eigenvectors of 100 different words recorded in computer settings were obtained; afterwards, recognition of verbal statement stored in the system is performed by comparing the eigenvectors of verbal statements. The performance and usability of Principal Component Analysis in the recognition of verbal statements and voice commands is examined within the settings of this study. At the experimental practices, it is seen that significant results were obtained in speech recognition by using Principal Component Analysis method.