International Journal Information and Communication Technology Research
Volume:1 Issue: 1, Winter 2009

An approximate analytical method for the evaluation of the cumulative distribution function (CDF) of the sum of L independent random variables (RVs) is presented. The proposed method is based on the convergent infinite series approach, which makes it possible to describe the CDF in the form of an infinite series. The computation of the coefficients of this series needs complicated integrations over the RV’s probability density function (PDF). In some cases, the required integrations have closed-form in terms of confluent hypergeometric function and in other cases, the required integrations can not be analytically solved and have not a closed-form solution. In this paper, an approximation method for computation of the coefficients of the CDF series is presented that only needs the mean and the variance of the RV, so it has low computational complexity; it eliminates the need for calculation of complex functions and can be used as a unified tool for determining CDF of a sum of statistically independent RVs. To present an application for the developed approximation method, it is used to find the distribution of the sum of generalized Gamma (GG) RVs. The derived approximate expressions are used in the performance analysis of equal-gain combining (EGC) receivers operating over GG fading channels. The accuracy of the developed approximation method is verified by performing comparisons between exact existing results in the literature and computer simulations results.

Voice conversion is a method used to transform one speaker''s voice into another speaker''s voice. New modification approach for voice conversion is proposed in this paper. We take Mel-frequency Discrete Wavelet coefficients (MFDWC) as the basic feature. This feature copes well with small training sets of high dimension, which is a problem often encountered in voice conversion. The proposed voice conversion system consists of both off-line (training) and on-line (transformation-synthesis) procedures and assumes parallel training data from source and target speakers and uses the theory of wavelets in order to extract speaker feature information. The satisfactory performance of the voice conversion system can be confirmed through ABX listening test and MOS grade.

A new method for speech enhancement based on Kullback-Leibler (K-L) divergence has been presented in this paper. First, the algorithm performs wavelet-packet transform to noisy speech and decomposes it into sub-bands; then we apply a threshold on coefficients in each sub-band to obtain enhanced speech. To determine the threshold, first the distributions of noisy speech, noise and clean speech coefficients are calculated; then a symmetric K-L divergence between the noisy speech and noise distributions is calculated. Finally a speech/noise decision is made based on the calculated distance. We conducted some tests using TIMIT database in order to assess the performance of the proposed method and to compare it to previous speech enhancement methods. The algorithm is evaluated using the Perceptual Evaluation of Speech Quality measure (PESQ) and the output SNR. We obtain an improvement of up to 2.2dB on SNR and 1.2 on PESQ for the proposed method in comparison to the results of the previous wavelet based methods.

Stereophonic acoustic echo cancellers are being used in high quality audio communication systems to reduce the echoes that arise from coupling between the loudspeakers and microphones. These echo cancellers have problems more severe than their monophonic counterparts. The most important problem is the mismatch between the impulse responses of adaptive filters and those of the acoustic paths of the receiving section and therefore a severe divergence of filters in the event of abrupt changes of the transmission section acoustic paths. The main reason for such mismatching is the two-channel nature of the system and strong cross-correlation between these two channels. In this paper، with inspiration from the lattice predictor structure، we present a novel approach for reduction of the interchannel cross-correlation of the input vectors in the wavelet domain that leads to faster reduction of the misalignment error.

Expansion of global networks and storing the extensive amount of information in various websites, create a serious needs for filtering the irrelevant information in a personalized way. Collaborative filtering or recommender system is a filtering technique that allows incorporation of the profiles which can be implicitly learned from previous activities [1]. We have proposed the CoCo-CF[1] as an effective method suitable for collaborating filtering running in a Jini-grid computing[2] platform and operational in a distributed environment. The CoCo-CF generates representative records from stored preferences and seeks for the answer with the best fitness in the recommender system. We have considered the user satisfaction rate, feasibility of available results, user familiarity and average response time as the evaluation factors. Also we have focused on mean absolute deviation, mean square error and ranked evaluation as the performance evaluation parameters. The obtained results confirm that the CoCo-CF is a successful method for collaborative filtering.

Call centers have become one of the most cost effective ways of selling products to customers and giving services to them in different industries. By applying knowledge management solutions, we can meet call centers’ challenges and gain benefits of reduced training costs, improved call handling and greater flexibility. This paper describes an agent mediated knowledge management system in call centers using the Tropos methodology. We use structure-in-5 for architectural design which specifies that our KM system is an aggregation of five sub-structures. Furthermore we develop a formal methodology and technique to verify the validity of communication protocols defined in a multi-agent environment. This is accomplished by examining agent conversations before deploying the system. The methodology leads to the definition of six different classes of agents. Our experiments develop proof of concept module for a call center that automatically verifies some of the important properties identified in this methodology. Results prove the agent’s specification and indicate that our proposed model works accurately. The paper concludes with observations on the call centers and the role of agents in the proposed model.