Compresation of High Frequency Power on Mpeg/Audio Decoding

Abstract

In Mpeg/Audio coding, an audio signal is effectively compressed as compared with an original audio signal. The technique of compression is the lossy compression by using human's perceptual characteristic. Since human's perceptual characteristic at high frequency bands is relatively weaker than other frequency bands, and this characteristic is efficiently used in compression. In other words, SNR (Signal to Noise Ratio) for decoded audio signal in high frequency bands is worse than in other frequency bands.
In this paper, we propose our method which improve SNR for high frequency component by what we propose, when compressed audio signal is in decoding. In order to realize the purpose, we will add compensating process to Mpeg/Audio decoding process. This process calculated high frequency component which is fell off for high effective compress rate in encoding process, in frequency domain. The result that is calculated compensation of high frequency is synthesized with the result of ordinary Mpeg/Audio decoding, so that synthesized audio signal improve SNR for high frequency component, without bad influence to other frequency components (i.e. making other frequency components to be worse).