Detection of bilingual twins by Teager energy based features

Automatic speaker recognition (ASR) has been an active area of research in speech processing. ASR deals with the identification of a person's voice with the help of machines. An important question which must be answered for the ASR system is how well the system resists effects of determined mimics especially identical twins or triplets. In this paper, a new feature set amalgamating Teager energy operator (TEO) and Mel frequency cepstral coefficients (MFCCs) is developed. We demonstrate the effectiveness of the newly derived feature set for cross-lingual identification of identical twins in Indian languages viz. Marathi and Hindi. The results are also compared for polynomial classifiers of 2nd and 3rd order approximation.