Text-Independent Speaker Recognition using Emotional Features and Generalized Gamma Distribution
K Suri Babu, Y. Srinivas, Suresh Varma Penumatsa · 2012
In this article, a novel methodology of text independent speaker recognition associated with emotional features is proposed. MFCC and LPC features are considered for speaker voice feature extraction. The speaker data is classified using generalized gamma distribution. The experiment is conducted on the emotional speech data base, containing 50 speakers with 5 different emotions namely happy, angry, sad, boredom and neutral. This approach is very much useful in costumer care and call centre applications. The accuracy of the developed model is presented using Confusion matrix. The results show that there is a great influence of the emotion state, while identifying the speaker. Keywords text-independent speaker recognition, generalized gamma