Considerarion of practical algorithms for digital signal processing based on varlable rate sampling

基于可变速率采样的数字信号处理实用算法的思考

基本信息

批准号：
14550356
负责人：
TADOKORO Yoshiaki
金额：
$ 2.3万
依托单位：
Toyohashi University of Technology
依托单位国家：
日本
项目类别：
Grant-in-Aid for Scientific Research (C)
财政年份：
2002
资助国家：
日本
起止时间：
2002 至 2003
项目状态：
已结题

项目摘要

We developed some algorithms for new digital signal processing based on variable-rate sampling by the research project, grant-in-aid for scientific research of 2000-2001. In the research project of 2002-2003, we reconsidered these algoritluns for the practical use. We can treat these algorithms by mainly additions and subtractions and so we can implement these algorithms by simple circuits. We present some results obtained in this research.1.connection method between filters having different sampling rates.We improved the connection method between filters from the oversampling method to the method using the first-order interpolation. As a result, we could reduce the amount of operations to one tenth for the frequency estimation processing.2.speed-up of frequency estimation based on synchronous addition and subtraction processing.We proposed a parallel synchronous addition and subtraction processing, and we could reduce the estimation time to about one third and improve the tracking characteristic.3.improvement of multi ate DFT(MR-DFT) and comb Fourier transform for noises.We improved the ability of the MR DFT for noises using mean operation of two MR DFT having different phases. For the comb Fourier transform using notch-type comb filters, we proposed other new Fourier transform (resonator Fourier transform) using resonator-type comb filters, and we applied the resonator Fourier transform into a signal receiver (DTMF receiver) and a transcription system for musical sounds including a percussion sound.4.transcription systems for real musical instrument sounds and sung sounds.For real musical sounds having frequency changes and noises, we developed two transcription systems noticing minimum or maximum outputs in the parallel connected notch or resonator type comb filters, respectively. We also proposed a transcription system using adaptive comb filters. These transcription systems can have good ability of the pitch estimation.

我们根据研究项目的可变速率采样（授予2000 - 2001年的科学研究）开发了一些新的数字信号处理算法。在2002 - 2003年的研究项目中，我们重新考虑了这些算法供实际使用。我们可以主要通过添加和减法来处理这些算法，因此我们可以通过简单的电路实现这些算法。我们在这项研究中介绍了一些结果。1。在具有不同采样率的过滤器之间进行连接方法。我们使用一阶插值从过度采样方法到该方法之间的过滤器之间的连接方法提高了连接方法。结果，我们可以将操作量减少到频率估计处理中的十分之一。2。基于同步添加和减法处理的频率估计速度估计。我们提出了平行的同步添加和减法处理，我们可以将估计时间降低到大约三分之一的估计时间，并提高跟踪的特征，并改善了对多f fter（MR MR MR）的转换（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT），MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）（MR-DFT）。 DFT用于使用两个具有不同阶段的MR DFT的平均操作。对于使用Notch Type梳子过滤器的Comb Fourier变换，我们建议使用谐振器型梳子过滤器将其他新的傅立叶变换（谐振器傅立叶变换）提出在平行连接的Notch或谐振型梳子过滤器中注意到最小或最大输出。我们还使用自适应梳子过滤器提出了转录系统。这些转录系统可以具有良好的音高估计能力。