288 Recording Studio Design However, the reality is that the ± 2 dB tolerance on the loudspeaker specifi- cation seems to pale into insignificance when one considers that in the process of getting the sound out of a loudspeaker and across the room to the ears of the listener, ± 10 dB would not necessarily be considered to be too bad. This does not apply to the smoothed, one-third octave responses, but to the unsmoothed responses that loudspeaker manufacturers rarely show in their literature. Figure 11.1 shows a pair of response plots for a well-known loud- speaker in an anechoic chamber. In (a) the response is unsmoothed, and in (b) it is as presented in the manufacturer's literature. Compare these plots with Figures 11.2 (a) and (b), which show the loudspeaker in a normal room. Here it can be seen that the difference between the smoothed and unsmoothed plots is gross. To be fair to loudspeaker manufacturers, the in-room responses are not entirely their responsibility, because the rooms are so variable that there is no typical response which would be representative. However, even the ane- choic responses are sometimes smoothed for marketing purposes to an extent that is questionable in terms of providing accurate and honest information. Room reflexions and resonances all contribute to the overall response per- ceived by the listeners, but the additional acoustic loading provided by the room boundaries can cause major response changes at low frequencies. These can be in the order of up to 18 dB in the relative balance of the low and high frequencies between loudspeakers in the centre of a reasonably large room or placed on the floor in a corner of the same room. The variability in the frequency balance in different places in the room can be very great indeed. The situation is not only complicated by the nature of the radiation pattern of the loudspeaker with respect to frequency, but also the reflexion density and decay time within the room. Room resonances, and whether the source is monopole or dipole, also add their complications. Before going into detail about the individual effects, we can perhaps look briefly at them in order to get an impression of how they manifest themselves. They all form characteristic parts of a loudspeaker/room system, as it should be clear by now that the loudspeaker response, alone, does not describe what we will hear in practical circumstances. 11.2.1 Radiation patterns Figure 11.3 shows three polar plots of low frequency loudspeaker radiation. The first pattern (a) is that of a dipole source. This is the typical radiation pattern of a conventional woofer on a simple open baffle board. The effect is also typical of the flat electrostatic loudspeakers with open backs. When the diaphragm is driven forwards, a positive pressure is created in front of the diaphragm, and a negative pressure is created behind it. The lack of enclosure behind the loudspeaker allows the pressure to equalise, simply by travelling around the edges of the baffle. The size of the baffle board on which the loud- speaker is mounted determines the frequency below which the cancellation will be effective. For a listener directly in front of the loudspeaker the perceived frequency response will be flat (assuming a perfect source) down to the frequency where the cancellation begins due to the finite size of the baffle, below which a roll-off will begin, which will finally reach 18 dB/octave. For a listener on the extended line of the baffle, i.e. listening from side-on, the pressures radiated at the front and rear of the loudspeaker diaphragm will