Blind source separation of acoustic mixtures using time-frequency domain independent component analysis

Blind source separation of acoustic mixtures aims at providing a solution to the classical cocktail-party problem. The inherent delays and convolutions in microphone recordings, entails a modification in the independent component analysis (ICA), which achieves separation by making the assumption of statistical independence of source signals that are linearly combined. The proposed algorithm provides a solution for the blind source separation problem by shifting the domain of the problem to the time-frequency domain and applying ICA to each of the frequency components individually. Satisfactory results were achieved for speech-music as well as speech-speech separation by adopting the time-frequency domain ICA.