DEEP NEURAL NETWORK USED FOR SPEECH SEPARATION

Abstract
Within the proposed system Deep Neural Network (DNN) is employed to get the speech features of target speaker and interfere for speech separation. This paper focuses on separating the target speech signal from the inputs during this system a regression approaches via deep neural network (DNN) for unsupervised speech separation during a single channel setting. This technique is believe a key assumption that two speakers might be well segregated if they're not too almost like one another. To demonstrate that the space between speakers of various genders is large enough to require possible separation. Then proposed DNN architecture having two outputs, from that one representing the feminine speaker group and another one is male speaker group. Finally the trained and tested DNN dataset performs the speech separation of the target speech.

Authors
Bhagat Anuradha Ramnath, R S Pawase
Amrutvahini College of Engineering

Keywords
Deep neural network (DNN), Regression model, Noise reduction, Speech Separation, Gender mixture detector, Speech separator
Published By :
ICTACT
Published In :
ICTACT Journal on Data Science and Machine Learning
( Volume: 2 , Issue: 2 )
Date of Publication :
March 2021

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.