Speech Compression using Multi-Resolution Hybrid Wavelet using DCT and Walsh Transforms
Rekha Vig, Sachin Singh Chauhan · Procedia Computer Science · 2018
Speech signals need to be compressed in order to be stored and transmitted efficiently. Wavelets have shown excellent results in compression of all types of signals. In this paper, a hybrid wavelet has been used for speech compression. This wavelet decomposes the speech signal into multiple resolution components and discards those with low energy in order to compress speech. This hybrid wavelet has been constructed using two transforms, DCT and Walsh. Various levels of compression can be achieved by varying the percentage energy to be retained, which acts as a threshold. Values of PSNR and MSE have been observed by varying this threshold from 99% to 99.9%. As compared to the existing techniques, hybrid wavelet is seen to perform better as significant improvement in PSNR is observed for similar bit rates.