MPEG音频编码的高效算法研究及单片DSP实现

ＩＳＯ标准推荐ＭＰＥＧ－１音频编码算法要实现起来运算复杂，为便于单片ＤＳＰ实现，算法可以优化。本文简要分析了该算法的各个模块并估计了它们的运算复杂性，然后对子带滤波和心理声学模型作了进一步的探讨，提出了基于逆余弦变换（ＩＤＣＴ）的子带滤波的高效实现算法和不用ＦＦＴ或只用部分ＦＦＴ的快速心理声学掩蔽计算模型。最后用单片ＴＭＳ３２０Ｃ３１＠４０ＭＨｚ实现了ＭＰＥＧ－１层２编码算法。

Efficient implementation of MPEG-1 layer 2 audio encoder based on single DSP

Implementation of the MPEG-1 audio encode algorithm as recommended by ISO standard1] is computational complex. In order to implement it with a single DSP, the algorithm procedure must be rearranged or simplified. This paper briefly ana- lyzes different parts of the algorithm and estimates their complexity. Then the sub-band analysis/poly-phase filtering and psycho-acoustic model are investigated. A more efficient implementation of sub-band filtering based on fast inverse cosine transform and a simplified psycho-acoustic model without FFT or with partial FFT is introduced. Finally, this paper presents a completed real-time MPEG-1 layer 2 audio encoder/recorder based on a single DSP TMS320C31 running at 40MHz.