No abstract

No abstract

Various methods have been exploited in the blind source separation problems, especially in cocktail party problems. The most commonly used method is the independent component analysis (ICA). Many linear and nonlinear ICA methods, such as the radial basis functions (RBF) and self-organizing map (SOM) methods utilise neural networks and genetic algorithms as optimisation methods. For the contrast function, most of the traditional methods, especially the neural networks, use the gradient descent as an objective function for the ICA method. Most of these methods trap in local minima and consume numerous computation requirements. Three metaheuristic optimisation methods, namely particle, quantum particle, and glowworm swarm optimisation methods are introduced in this study to enhance the existing ICA methods. The proposed methods exhibit better results in separation than those in the traditional methods according to the following separation quality measurements: signal-to-noise ratio, signal-to-interference ratio, log-likelihood ratio, perceptual evaluation speech quality and computation time. These methods effectively achieved an independent identical distribution condition when the sampling frequency of the signals is 8 kHz.

Many techniques are introduced as solutions of the Blind Source Separation mechanisms, as an Independent Component Analysis (ICA), which became most commonly used in this field. ICA methods exploit one of two properties: sample independency and/or non-Gaussianity. In our study, cocktail-party problem processed using ICA method. In this paper, we studied the performance of three technics with independent component analysis are standard FastICA, PSO, and QPSO; and compare the results of each algorithm with others according to the number of metrics (objective as SNR and SDR and subjective as signals plotting and playing). The implement of these algorithms were be made with two source signals and three source signals. As in evaluation process, the QPSO gives more accuracy results than other technics in the signals separation process. Many input speech signals of sampling frequency 8KHz, that achieve IID. also well condition, were tested for different speeches for men and/or women, also music.

No abstract