[sdiy] Filter ICs

[Magnus Danielson]

From: Paul Perry <pfperry at melbpc.org.au>
Subject: Re: [sdiy] Filter ICs
Date: Fri, 05 Oct 2001 11:34:04 +1000

> At 03:55 PM 4/10/01 -0700, gothgeeksysadmin wrote:
> 
> >I'm thinking they look interesting ( 50db cutoff! ) for perhaps
> >a Vocoder.  
> 
> A sharp cutoff is not necessary (or indeed maybe not even desirable!) for a
> vocoder. Something to worry about, for example, is phase relationships.
> But, looking at it from the angle of an 'instrument', rather than a general
> purpose tool, maybe.

It is a hard balance actually. In one end you want the side-bands to
experience a minimal leakage, so they get quite enougth, in the other
end you don't want too quick changes betwen two bands as the pitch
shifts, this would cause a somewhat unnatural behaivour. Also note
that such pitch-changes is to be expected since they are a natural
event in speech. Spectra-histograms have shown this for a long time now.
It was known during the WWII for instance.

Also, let's not confuse the issue of a sharp cut-off with that of the
slope of the damping. These are different issues. Things gets more
confused as you run resonant filters which all vocoders that I have
seen is doing. With resonant filters this whole phase issue becomes
like a minor issue because the phase variations is all over the place.

Cheers,
Magnus