A novel compression algorithm using DWT

In this paper, a novel algorithm based on Discrete Wavelet Transform has been proposed for good compression ratios. In proposed algorithm, the audio samples of the signal are first decomposed into N equal frames and decomposed signal is processed to remove nonzero coefficients which significantly reduce the size of audio signal. The signal is then thresholded and psychoacoustically modeled to obtain the tonal and non-tonal components. Further, quantization is performed on the masked signal. In the end, the signal is reconstructed and various parameters are studied. In this paper, best compression ratio is achieved with db2 wavelet at level 4 and signals are analyzed on the basis of mathematical parameters viz. SNR, distortion. This has been implemented on MATLAB 7.1 R2009 on an Intel i-5 processor based machine.