A novel inverse quantization algorithm based on Taylor series for digital audio codecs

We propose a novel algorithm for inverse quantization in some audio codec systems, such as MP3, AAC, etc. The optimized algorithm not only reduces the memory size and increases the accuracy of coefficients, but also simplifies the calculations. Compared with other reported algorithms based on float-point calculation, the novel method significantly improved both the accuracy and calculation complexity. We also show that the performance of the system is improved tremendously when the new algorithm is implemented under fixed-point calculation.