Onur, Tuğba Özge | Hacıoğlu, Rıfat

Proceedings | 2013 | 2013 21ST SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU)

The speech signals are formed with echo caused by the reflections and also noise caused by engine, tyres, wind, etc during using hands-free voice communication in automobiles. In this paper, it has been focused on the adaptive filtering algorithms that based on eliminating purpose of not only noise removal but also separation in speech signals with echo and noise. It has been evaluated that the speech signals separation performance by comparing adaptive filtering methods that made with Least Mean Squares (LMS) and Recursive Least Squares (RLS) algorithms. Signals belong to near and far end speakers will be decomposed with each while . . . near-end and far-end speakers' signals affected by noise are decomposing from noise with the proposed adaptive algorithms. Furthermore, convergency and error analyses are performed by comparing the performances of LMS and RLS algorithms Daha fazlası Daha az

Narin, Ali | Özer, Mahmut | İşler, Yalçın

Proceedings | 2013 | 2013 21ST SIGNAL PROCESSING AND COMMUNICATIONS APPLICATIONS CONFERENCE (SIU)

In this study, the performance of different discrimination algorithms in the analysis of heart rate variability that are used in discriminating the patients with congestive heart failure from normal subjects were investigated. Classifier algorithms of linear discriminant analysis, k-nearest neighbors, multilayer perceptron, radial basis functions and support vector machines were examined with different parameter values. As a result, the maximum classification accuracy of 91.56% was achieved by using multilayer perceptron with 11 neurons in hidden layer.