SOUND SOURCE CHARACTERISTICS RECOGNIZING METHOD AND EQUIPMENT

摘要

PROBLEM TO BE SOLVED: To enable grasping a physical sound source signal from mathematically independent signals, by rearranging singular values operated from a mutual spectrum matrix constituted of an input signal which is the nearest to a selected sound source position, from measurement signals of a multi-input uni-output system. ;SOLUTION: Sound source signals u₁-u₃ whose waveforms have no correlation are output from speakers SP1-SP3, and input signals x₁-x₆ are measured with microphones M1-M6 for input measurement. The measured signals are once input in a frequency analyzer FA, and the time region signals are converted to frequency region signals, which are operated and processed with a computer COM. That is, singular values are calculated by applying singular value resolution to a mutual spectrum matrix constituted of a specified number of signals selected as the sound source signal which is the nearest to the sound source signal. The singular values are rearranged by using a rearrangement matrix, and physical characteristics such as the magnitude of the sound source signal and the frequency are grasped, thereby recognizing sound source characteristics of noise at an observation point.;COPYRIGHT: (C)1999,JPO