Multilingual Speaker Identification using analysis of Pitch and Formant frequencies

Abstract

In the modern digital automated world, speaker identification system plays a very important role in the field of fast growing internet based communications. In India there are many people who are bi-lingual or multilingual, so the requirements to design such system which is used to identify the multilingual speakers. Present paper explores the idea to identify multi-lingual person by basic features. For this the speech signals of three indian languages i.e Hindi, Marathi and Rajasthani are recorded and basic features pitch, first three formant frequency calculated from PRAAT software. The observation has been presented that the pitch and first three formant frequencies F1,F2 and F3 of speaker are increases when speaker change the language from rajasthani to hindi to marathi. The percentage deviation in pitch as well as formant frequencies for Rajasthani and Marathi from hindi are positive and negative respectively for utterance “p”. Similar analysis has been perform for ’k aand >. This observation will help to make such system which is used to identify the speaker in multilingual environments.