An Efficient DSP Implantation of Wavelet Audio Coding for Digital Communication

Digital audio offers increased sound quality and greater signal processing flexibility than its analog counterpart. The key enabling technology for digital audio wireless products is audio compression due to channel bandwidth constraints. However, current audio coding schemes can hardly achieve ultra-low delay of encoding and decoding for live productions. This paper presents an efficient DSP implantation of wavelet audio coding using very short block processing to meet the very low delay requirement. The audio signal decomposition and reconstruction is performed by a two dimensional (2D) spatial-frequency processing of lifting wavelet transform to fully exploits the correlation for better compression performance. The lifting wavelet with boundary effects minimized is developed with lifting coefficient optimally quantized and implanted by fixed-point arithmetic applying only bit shifting and addition operations to replace multiplications and divisions, thus minimizing the computational complexity for real-time applications. A modified 2D embedded SPIHT algorithm with more bits used to encode the wavelet coefficients and transmitting fewer bits in the sorting pass, is implemented in fixed-point computation. Experimental results demonstrate that the proposed coder is efficient and has low complexity with less memory requirements for digital communication.