Hi all,
      I've been attempting to run a simulation of Fletcher's band-
widening experiment where the threshold amplitude of a pure tone is 
measured as a function of the a bandwidth of a bandpassed noise 
masker centered at the pure tone frequency (2000Hz in this case). 
It's essential for the experiment that noise power density is held 
constant for different bandwidths of noise masker. I have been using 
the Butterworth filter opcode to filter  'rand' generated white 
noise, and in this basic form the max amp in a segment (as indicated 
by CSound) increases with bandwith. I've then tried the approach 
of using 'balance' to match the filtered noise samples to a reference 
RMS value, and/or multiplying the instr 'out' signal by a fraction 
that causes all max amp values for each bandwidth to be approximately 
equal. I need some more info. or reasurrance that this 
approach is indeed producing a more or less constant power density 
across frequencies in the noise bands. When I look at the wave files 
with a digital frequency analyser (Wave SE) the noise power density 
seems to decrease with bandwidth, eventhough all the max amp values 
are now the same. Is this just an artifact of the processing in the 
frequency analyser, have I got decreasing noise power density, or 
both ? 
Any help appreciated......   
                Thanks in anticipation,
                
                         Pete Kearton