Comparing audio compression using wavelets with other audio compression schemes

Abstract

Speech compression is the technology of converting human speech into an efficient encoded representation that can be decoded to produce a close approximation of the original signal. In this paper, we propose a new algorithm which compresses speech signals using a wavelet compression technique. The performance of this method is compared against the following representative coding and compression schemes: adaptive differential pulse code modulation (ADPCM) which reduces the transmitted data by a factor of two; linear predictive coding (LPC) with compression ratio of more than twelve to one; linear predictive coding algorithm using the United States Department of Defense Standard 1015 with compression ratio of 26:1; Global System Mobile (GSM) algorithm which reduces the transmitted data by a factor of five. The following parameters are compared: (i) quality of the reconstructed signal after decoding; (ii) compression ratios. (iii) signal to noise ratio (SNR); (iv) peak signal to noise ratio (PSNR); (v) normalized root mean square error (NRMSE).