Text-independent Speaker Recognition using Deep Neural Networks

In this paper, a deep neural network is proposed for Speaker Verification (SV) system. SV is an authentication process, which involves the examination of knowing, if the speaker is the target person or not? The human voice is a biometric-like fingerprint which is an authentication tool. Neural networks are one of the successful methods to learn from unstructured data such as speech signals. Deep neural networks show plausible results in many applications of speaker verification. In this paper, we propose a Dense neural network trained on the free-spoken digit dataset (FSDD). Simulation results demonstrate the effectiveness of the proposed approach, despite of the simple deep neural network, we achieved a good accuracy of about 93%.