The measured impulse response is often used as a true reference of a real room impulse response and geometrical acoustic simulations are considered to be only a crude representation of it. However, both approaches have their own challenges and limitations. Geometrical acoustic models do not include wave phenomena, such as interferences and diffraction, as they simplify sound propagation by rays. The advantages of acoustic simulations with such models include a perfectly omnidirectional and impulsive sound source, no distortion problems, full control of the background noise, and a well-defined onset time of the impulse response. On the other hand, impulse response measurements include wave phenomena, but they do have their own weaknesses, which may cause significant errors in the derivation of the ISO-3382 room acoustic parameters. Due to the presence of background noise in the measured impulse response it is difficult to evaluate which part is valid. In addition, the sound source used for measurements often has strong lobes at high frequencies and cannot produce an ideal Dirac function. For this reason and due to distortion products by the octave-band filtering process detection of the arrival time of the direct sound from a measured impulse response is of questionable accuracy.

In this paper simulated and measured parameters are compared in a number of well documented cases and the various sources of errors are discussed. It is concluded that doing room acoustic measurements correctly may be more difficult than it appears at first glance.