DSP techniques for telecommunication systems

Includes bibliographical references (pages [119]-122)Three digital signal processing (DSP) techniques for telecommunication systems are presented in this thesis. First, a single-channel speech enhancement technique using spectral subtraction algorithm is presented. This method suppresses the noise in speech by subtracting the estimated noise spectrum from the noisy speech. Computer simulations and real-time experiments on the TMS320C50 system are presented to show that this algorithm achieves good speech enhancement in both white noise and periodic noise cases. Second, a dual-channel acoustic noise and echo cancellation microphone (ANECM) system is developed for hands-free telephone system. In the traditional adaptive noise canceling (ANC) and acoustic echo cancellation (AEC) systems, a very high order finite impulse response (FIR) filter is required to achieve satisfactory performance. However, high order filter results in high cost, slow convergence, high excess mean-square error, and numerical error. In the unique structure of ANECM system, two low order filters are required to cancel the noise and echo. A comparison of system performance shows that the new ANECM system has a better overall performance than the traditional AEC and ANC systems. Finally, a dual-tone multifrequency (DTMF) receiver implemented on the TMS320C50 system is presented. With a modified Goertzel algorithm, the energies of the received signal at eight DTMF frequencies are efficiently computed. Five system tests are performed to determined whether a valid DTMF tone pair has been received. Real-time experiments are conducted to evaluate the system performance with the noise and speech signal present. The results show that this receiver not only correctly detects the input digits but it has also met the requirements of CCITT standard.M.S. (Master of Science