Avoidance of noise-related errors in the analysis of room impulse responses

Room impulse responses can be analysed to yield several room acoustical parameters, such as reverberation time, early decay time, clarity index and strength factor. Despite the considerable noise immunity of modern impulse response measurement techniques (such as swept sinusoids), it is common for room impulse responses to have sufficient noise to have some effect on the derived room acoustical parameters. The effect of noise is typically seen by a change in the slope of the late part of the decay function (derived from Schroeder’s reverse integration method), that can be detected using a series of short term correlation coefficients over the course of the decay function. By describing a room acoustical parameter error with a correlation coefficient value, a mathematical model can be established to correct errors from a calculated correlation coefficient over the 20 or 30 dB decay range used to calculate reverberation time. Such a technique is ideally suited for room soundfields that exhibit exponential decay, and needs to be used carefully in real room level decay functions. The method is evaluated using real room impulse responses with artificially introduced noise at various levels.