Bravo T., Maury C.

This paper shows practical results on the experimental synthesis of an acoustic diffuse field over a test partition separating a source room from a receiving room. The objective is to validate a novel approach in order to reduce the variability of sound transmission measurements in the low-frequency domain. The approach is based on the generation of an acoustic diffuse field with a near-field array of sixteen loudspeakers suitably driven and located in the source room side of a transmission suite. The sources drive signals are optimised to reconstruct the same spatial correlation structure than an ideal diffuse field over a sufficiently dense grid of microphones that are distributed a short distance away from the test panel surface. The technique enables to compensate for the modal behaviour of the source room which is a major limitation when sound transmission measurements have to be performed using a small sized reverberation chamber. Experimental results show the efficiency of the approach to provide a measure of the sound reduction index that only depends on the properties of the panel itself over a broad frequency range, especially below the source room Schroeder frequency. Empirical criteria are proposed on the number of partially correlated sources required to achieve an accurate reconstruction of the assumed diffuse field statistics over the test panel. This is analysed in relation with the influence of the degree of diffusivity upon the sound reduction index.