Uncertaintiy discussion of spatial sound field changes in auditoria
- Research Area:
- Auditorische Szenenanalyse
Master Thesis of Kliesch, David
In the scope of this work it has been shown that it is possible to evaluate a model function which relates distances between microphone pairs to a chosen room acoustic metric. With the help of the framework defined by the Guide to the Expression of Uncertainty in Measurements the distances between measurement positions as well as the differences in the examined metric have been equipped with a determined uncertainty. A program has been written which uses large sets of room impulse responses which are affiliated with a known uncertainty as input. From each impulse response the underlying uncertain position is taken and a chosen uncertain room acoustic parameter evaluated. Calculations of the acoustic metrics are refined with a compensation energy to account for the lass of signa l energy for the late room impulse response in the noise floor while also taking the respective uncertaint y of the compens tion into consideration. By evaluating all possible distances of microphone positions and the corresponding differences in metrics while also accounting for the uncertainty of both parameters said model functions can be generated. With the use of these mode l functions it is possible to establish borders for the resolution of measurements in auditoria. Using the just noticeable difference found by Cox et al.  of ldB results in a measurement resolution of only a few centimetres to up to lm. The JND found by Höhne et al  of 3dB is only in very rare cases for low frequency bands surpassed. This result is in line with the fin ding from de Vries  where fluctuations over higher distances rarely pass the 3dB mark as well. Further on model functions have been generated for octave band filtered impulse responses as basis. The model functions can be separated in a strongly increasing and a more or less constant part if any only with a small slope. It has been shown that the crosspoint between the two parts of the new model functions is dependent on the respective center frequency when using octave band filtered impulse responses. Several sets of impulse responses have been generated in order to validate certain state ments. In the first part of the result chapter the repeatability of measurements has been discussed with the conclusion that the examined model functions show only minor differences. In the second part the distance to the sound source has been analysed. For lower frequencies the model functions vary greatly but tend to move together as the frequencies get higher. Lastly in the third part the influence of the room properties absorption and reflection have been discussed. The more absorbent the walls of the studied room become the lower becomes the slope of the slowly rising part of the model function resulting in smaller differences in the clarity metric.