Speech Separation Using Reduced Complexity DUET Algorithm in Real Environment

Abstract

The Degenerate Unmixing Estimation Technique (DUET) is a practical algorithm for source separation in anechoic environment. DUET performs source separation by frequency domain processing and is independent of the number of mixed sources. However performance of this algorithm for long speech signals is very slow. In this research we first down sampled real mixtures and then implemented the modified DUET algorithm. Using the proposed method reduces the implementation time almost one third of the basic algorithm.

Keywords

• Blind Sources Separation
• Mixed Speech Signal
• Weighted histogram