论文部分内容阅读
本文提出了一种新的时一频分解方法──自适应旋转投影分解法（ＡＯＰ法）．在表征信号空间的线性调频高斯信号集上，我们针对原始信号自适应地搜索出一组与信号匹配最好的基函数序列．以此用尽可能少的基函数来重构信号子空间．根据分解系数，得到信号的时－频能量分布．由于调频高斯信号时频会聚性能极佳，又能灵活高效地匹配各类信号，该算法不论从分辨率、效率还是描述能力等方面都具有良好的性能．将它用于语音压缩也取得了很好的结果．
This paper presents a new time-frequency decomposition method ─ ─ Adaptive Rotate Projection Decomposition (AOP). On the chirp signal set, which represents the signal space, we adaptively search a set of basis function sequences that match the signal adaptively to the original signal. In order to use as little as possible basis to reconstruct the signal subspace. According to the decomposition coefficient, the time-frequency energy distribution of the signal is obtained. Due to the excellent convergence of frequency-modulated Gaussian signals in time-frequency range and the flexibility to efficiently match various signals, the algorithm has good performance in terms of resolution, efficiency and descriptive power. Using it for voice compression also yielded good results.