Twins Voice Verification And Speaker Identification
Mamatha Balipa, Afeefa Farhath
Abstract
The similarity in twins voice has become a serious matter of research recently. The present work is a preliminary study of exploratory character describing the similarities of voices of twins and to give the accuracy by prediction. Speaker identification is a task of identifying persons voice accuracy from their voices. Recently, deep learning has dramatically revolutionized speaker identification. However, there is lack of comprehensive reviews on the exciting progress. Siamese Neural Network (twin neural network) is used in this work. The neural network helps in the speaker identification. In prior works of speaker identification, discriminative neural networks, utterance-level evaluation speaker representations and the accuracy of the voice are usually calculated. In this model the frame-level evaluation representation is mapped into enrollment feature domain and an utterance-level evaluation-enrollment joint vector for final similarity measure is further generated. Feature learning, attention mechanism, and metric learning are jointly optimized using an end- to-end loss function.