An Audio Compression Method Based on Wavelet Packet Decomposition, Ordering, and Polynomial Approximation of Expressive Coefficients

This paper presents a method for audio compression that makes use of wavelet packet decomposition (WPD), ordering, and polynomial approximation of expressive coefficients. The proposed encoder orders the WPD coefficients for a suitable polynomial approximation found by multiple linear regression. In addition, the method makes use of an adaptive system of rate versus distortion control (ASRDC) to maintain a nearly constant bit-rate in the audio output. A discussion on the flexibility of the encoder to different settings is presented. Lastly, experimental results of the proposed system and those of well-known codecs are compared. The results indicate that the proposed system achieves competitive performance concerning SNR, particularly for some instrumental audios. In objective measures of perceived audio quality, the current implementation of the proposed codec has negative results compared to other codecs. This method provides an efficient way to exploit the key strengths of polynomial approximation to audio compression and can be easily adapted to other data types.