COMPARATIVE EVALUATION OF SOME DTMF DETECTION TECHNIQUES

Abstract

This dissertation deals with the implementation of DTMF (Dual Tone Multi Frequency) detection techniques and study the effect of noise and voice signal on DTMF signals. Three algorithms, namely DFT (Discrete Fourier Transform), Goertzel and Filter Banking, have been selected to carry out the work in this dissertation. . With the help of Matlab the simulated model of the telephone system was generated and to meet the limitations of practical installation, additive White Gaussian Noise has been introduced. The order of the 'filter is chosen so that the ITU recommendations are fairly met and simultaneously the detection process remains robust to noise and voice signals. Besides maintaining these stringent requirements, calculation of Power Spectral Density, Keypad simulation and visual inspection of detected waveforms have also been carried out. These algorithms have been tested on large amount of data generated as a rest of power variations of input signal. The results with discussions on performance aspects are presented for different cases. The conclusion of the present work and scope of future work are also given, in the last Chapter of the dissertation. The work has been done in C over DOS and simulated in Matlab. Overall, the work contained in this dissertation adds a further step in the area of DTMF detection algorithms and their application in modem communication systems