Abstract:
The acoustic assessment of a receiver position in a room is evaluated from the measurement of different acoustic parameters derived from the room impulse response (RIR). The definition of such parameters as well as their measurement procedures are described in the ISO 3382 standard. With regard to the sound source, the loudspeaker should be as omni-directional as possible. Most of the commercial dodecahedron loudspeakers comply with the maximum allowed directional deviations of the source specified in the standard. While this requirement is adequate for the derivation of reverberation times, for the detailed investigation of the RIR’s time structure, however, the directivity of the source becomes more important. Presently it is undetermined whether changes in the sources orientation alter the results of the acoustic parameters to a greater extent than the subjective just noticeable difference (jnd) of the corresponding parameter. In this work, the effect of the source orientation is analysed by the means of a set of measurements carried out with two “omni-directional” sources according to ISO 3382 specifications. The comparison of newly obtained deviations of the measured values of acoustic parameters and both the experimental standard deviation (STDexp) – characterizing the dispersion under repeatability conditions – and the jnds of the corresponding parameters shows that directionality of the source has a significant influence at high frequencies. Several parameters, such as C50, C80, G, and IACCE, show deviations caused by the source orientation that are larger than the respective jnds, while the overall uncertainty of the measurement is insignificant. For acoustic parameters that are calculated with longer integration times, e.g. T30 or EDT, the deviations are lower. The results obtained with the different sources vary on a coincidental basis. The evaluation of the acoustic quality on the basis of the different measurements may yield to vague results even if the same source-receiver position is characterized.