The unit under test and a small precision lab microphone whose response
is well known in the frequency range of interest, are placed close
together in the sound field of the speaker at the desired test distance.
Either the two mics or the loudspeaker are moved around during the
measurement, while the mics remain fixed with respect to each other. 
The output of the unit under test, compared to that of the reference
mic, is averaged in the time domain and transformed into the frequency
domain using an FFT.  Time averaging reduces the effect of external
noises as they are not synchronous with the excitation signal, and the
spatial averaging that results from varying the relative position of the
loudspeaker and mics reduces the effects of any reflections in the
measurement environment.  The FFT output, which contains amplitude and
phase information, is normalized for the shadowing and diffraction
effects of the two mics on each other and for the known response
variations of the reference mic.  According to the makeup of the
excitation signal and the analysis of the microphone output, most
performance parameters can be measured with great accuracy, without an
anechoic environment.  Conventional spectrum analysis, with the mic in
an isolation chamber, is used to determine noise performance.  
^ ^ ^ ^ 
Here's testing microphones transcribed of a talk at an annual AES
meeting.  
The full transcript is on the Web at the Josephson Microphones site!
* POW 1.2  On Trial * Deja News all over again.
--- Alexi/Mail 2.02b (#10000)
---------------