Hi List,

Recently I was cleaning up some tokens of recorded speech for a
spectrogram display. When I was doing so I noticed several bands of
atmospheric noise from the various fans that were on during recording
(which was not done in a sound-treated room).

My question is, what acoustic difference is there in using a
band-reject filter versus simply some type of parametric equalization?
I noticed that the band-reject filter is quite steep, virtually
eliminating spectral energy around the frequency of cutoff.

Is there any drawback to using other models of frequency selective
attenuation than the band-reject filter? How would you vary the
steepness of a butterbr filter, for example? Of course to make it
steeper you could just put them in series

ie.

a1 butbr ain, kcf, kbw
a2 butbr a1, kcf, kbw
a3 butbr a2, kcf, kbw

..etc

What would you do however to make a single filter less steep? Perhaps
it could be described as "fractional order" filters. Is this type of
operation only possible using biquad or similar?


Thank you for your time and consideration,

David Akbari