Speaker recognition using multimodal biometric system

We proposed a model to recognize speaker by face and voice signal. In this model we input voice into model and we get voice and face features corresponding to that voice and face by which we try to recognize the user. To achieve this we used a supervised learning model ANN (Artificial neural network). To train this model we input voice feature and face features in synchronous way into model. The ANN model tries to classify the input with respect to a set of users. The main problem in designing this model is synchronization between voice feature and face feature, extraction of face feature, deciding parameter of ANN like iteration value and the last most difficult is making a model to map the face along with voice features. It´s very difficult to tackle with the above discussed problem but we have designed a model to achieve somewhat realistic model of speaker recognition.