A survey on single channel speech separation

Abstract

Single channel speech separation is a branch of speech separation process, which is an ongoing interesting research topic for the past 40 years and continues till now, but still there is a lack in separating the required signal from the mixture of signals with 100% accuracy and be used by the common people. Many researches have been done in various ways using the parameters like pitch, phase, magnitude, amplitude, frequency and energy, spectrogram of the speech signal. Various issues in single channel speech separation process are surveyed in this paper and the major challenges faced by the speech research community in realizing the system are pointed out as conclusion. © 2012 ICST Institute for Computer Science, Social Informatics and Telecommunications Engineering.