The use of Code Division Multiple Access (CDMA) technology for cellular radio and personal communications has ushered in a new era wherein the benefits of CDMA can be directly applied to voice transmission. In particular, the resistance to multipath and graceful degradation of performance (voice quality) in the presence of multiple access interference are the significant benefits of CDMA. Spectral efficiency is a key issue in mobile communications. The use of vocoders reduces the average bit transmission rate for each user, thereby reducing the bandwidth necessary to transmit the speech of each user. Qualcomm-Code Excited Linear Prediction (QCELP) algorithm provides a way to implement a variable rate vocoder which could reduce the average bit rate by a factor of two. This thesis describes a software implementation of the QCELP algorithm and the integration of the vocoder with the CDMA transmission and reception segments. A subjective performance analysis of the vocoder, using a Mean Opinion Score (MOS) test on the speech quality under various channel conditions including 1-Ray and 2-Ray Rayleigh fading environments is performed. Objective quality measures such as Signal-to-Noise Ratio (SNR), Segmental SNR (SSNR) and the Coherence Function (CF) are also investigated for comparison with the subjective analysis results. / Master of Science