With several other venues on my project list, time is tight. Only the most essential measurements can be taken, and topping this list is overall system response.
First, both of the outside loudspeakers in each group are angled outward about 45 degrees in order to broaden and smooth the horizontal coverage. Then I place a flat-response omnidirectional mic at ear level in the seating area and feed pink noise through the system, one channel at a time.
The mic signal is routed to a third-octave filter, which is swept slowly and continuously over the entire frequency range, and then plotted on a graph. This process is repeated at a total of eight mic locations throughout the seating area. (Left channel results can be seen in Figure 2.)
Close examination shows that each trace is a different color, and each is the measured frequency response at one of the mic locations.
Following an individual trace clearly shows places where the level varies up and down over a range of 5 dB or 10 dB in less than an octave. This is a very ragged response and represents what a listener at that location would hear.
Such variations are quite audible – and don’t sound good!
Figure 2. Third-octave frequency response of the left channel, measured in eight audience locations. (click to enlarge)
Further, looking at a particular frequency shows that the level of the different traces varies, typically by 5 dB or 10 dB as well.
This is variation in level at that given pitch from one audience location to another. Also not good.
Already, there is a lesson: with such strong response variations from one location to another, no one location can even come close to providing an accurate indication of the overall system response.
Experience and a few studies indicate that six samples, well spaced, are an absolute minimum.
More is needed in cases like our example, where there is a strong variation from one location to another. It’s likely to be the culprit of poor equalization results.