An important issue which must be addressed for the speaker recognition system is how well the system resists the effects of determined mimics such as those based on physiological characteristics especially twins. In this paper, a new feature set based on recently proposed Variable Length Teager Energy Operator (VTEO) and state-of-the-art Mel frequency cepstral coefficients (MFCC) is developed. The effectiveness of the newly derived feature set in identifying twins in Marathi language has been demonstrated. Polynomial classifiers of 2 nd and 3 rd order have been used. The results have been compared with other spectral feature sets such as Linear Prediction Coefficients (LPC), Linear Prediction Cepstral Coefficients (LPCC) and baseline MFCC.