Development of a speech separation system using frequency domain blind source separation technique

Professionals can interact while communicating remotely with teleconferencing. It enables communication between users using computers, smartphones, tablets, and other virtual devices. Even though researchers are adopting a variety of blind source techniques to separate and recognize speech, the problem and the greatest difficulty still lie in assuming that communication comes from multiple speakers. The process of extracting target speech from background noise is known as speech separation. In this research, a speech separation system using the frequency domain Blind source separation technique (BSS technique) is used for the separation of the original speech signals from the user. Frequency domain analysis is used for the comprehensive analysis of the signal properties and along with that the frequency range of the signals in the room is also determined. Determining the frequential spectra of the sources included in a sound recording as well as their temporal activations helps in improving the speech signal of the user. Blind source techniques help in retrieving the original sound by eliminating external noises and selecting the best frequency signal. The total impulse response of the system is evaluated The speech signals along with the Room Impulse Response, magnitude, and Phasor plot are graphically represented for the efficient analysis of the system.