Bias Analysis of the Coherent Averaging Technique for Loudspeaker Response Measurement

吴鸣,林志斌,邱小军,徐柏龄

DOI: https://doi.org/10.3321/j.issn:0469-5097.2006.01.007

2006-01-01

Abstract:In practical applications such as on-line loudspeaker measurement,it is usually needed to measure the loudspeaker responses in an ordinary room.The most popular techniques include impulse FFT technique,time delay spectrometry technique(TDS) and the maximum-length sequence technique(MLS).But the precision of all these techniques is not high enough.In addition,the frequency response of loudspeaker at low frequencies can not be obtained through these techniques.George and Francois(2003)~([1]) introduced a method at the Eighth Western Pacific Acoustics Conference named coherent averaging technique which selects or rejects contributions to a measurement by choosing some transmission paths to be time invariant and making others-the ones we wish to remove from the measurement-vary.This technique is capable of making measurements of loudspeaker responses in ordinary rooms without the need for a time windowing approach-and hence avoiding the limitations of such an approach-based on coherent averaging of repeated measurements in different parts of the room.Measurement effort and complexity are increased but simple rotation of the transducers appears to be a satisfactory process-especially for medium and high frequencies.However,there exists bias by using this method.In this paper the bias at different frequencies is analyzed when only one reflecting surface exists.It can be seen that the bias at low frequencies is bigger than that at high frequencies.Also,a modified method is introduced in this paper to get the response of loudspeaker with higher precision.