Deep Learning based Voice assistance in hospitals using Face Recognition

Abstract

Abstract After listening to a wake word or order, voice assistants come in quite compact packages and can accomplish a range of acts. They are able to switch on the lights, answering basic questions, playing music, place online orders, etc. As voice assistants become more robust, their usefulness can also be extended in both the personal and business areas. Various descriptive variables are fused in speech signals, leading to considerable difficulties interpreting any of the variables. There are various effective algorithms for speech and face recognition for the real - world applications. When a human try to communicate with a bot or vice-versa, there occurs so many difficulties in information sharing. This paper deals with one such application of bot assistance in hospitals, where a patient communicates with the bot assistant. Though the task seems to be quite easy, it involves a main threat: surrounding noise, word error rate, accuracy of the speech. An obvious impression is to factorize the frame of audio into various enlightening variables, but it turns out to be extremely difficult. The face of the patient who enters the healthcare premise is captured and analysed through the Deep Neural Network (DNN) based face recognition algorithm. Once identifying the user, the information regarding the patient is gathered from the database and given as a voice over output. Now, it is necessary to measure the accuracy of the word recognition and the word error rate. For this Cascade Deep Factorization (CDF) and Iterative Signal Enhancement (ISE) algorithm are used and discussed in this paper. For several speech processing functions like speaker recognition, this factorization and reconstruction method offers possible values.