A perceptually motivated multiple-F0 estimation method
A perceptually motivated multiple-F0 estimation method
复制标题
DOI：
10.1109/aspaa.2005.1540227
复制
发表时间：
2005-11
期刊：
影响因子：
--
通讯作者：
Anssi Klapuri
中科院分区：
文献类型：
--
作者：
Anssi Klapuri
本文描述了一种估计多个同时出现的乐音的基频的方法。该方法包括一个计算模型的人类听觉周边，其次是一个新的周期性分析机制。通过从混合物中消除每个检测到的声音并通过重复对残差的估计来实现多个基频的估计。该方法的计算负荷保持合理，因为外围听力模型（即，最难的部分）只需要计算一次。该方法是相对简单的实现和优于两个国家的最先进的参考方法在模拟。
This paper describes a method for estimating of the fundamental frequencies (FO) of several concurrent musical sounds. The method consists of a computational model of the human auditory periphery, followed by a novel periodicity analysis mechanism. Estimation of multiple fundamental frequencies is achieved by cancelling each detected sound from the mixture and by repeating the estimation for the residual. Computational load of the method remains reasonable since the peripheral hearing model (i.e., the hardest part) needs to be computed only once. The method is relatively straightforward to implement and outperformed two state-of-the-art reference methods in simulations.