[sdiy] vocoder filters

Magnus Danielson magnus at rubidium.se
Thu Oct 3 14:28:30 CEST 2019


I think you mix up my comments and misses my point.

I know what order of filter and Q of the pole-pairs do.

For speech clarity, you want the group-delay to be flat, and essentially
no ringing. This moves you away from high-Q solutions. Some have used
high-Q solutions to make simple filters with enough slopes, but they
miss out on the properties of the filters and what it does in the
application. For maximum flat group-delay, Bessel-Thompson is the ideal
to look for. The selectivity you get by degree of the filter to have
slopes from that rather than Q.


On 2019-09-04 05:40, Bernard Arthur Hutchins, Jr wrote:
> Magnus –
> I think we would agree that there may be reasons for considering
> bandpass responses of order greater than 2 – but getting a higher Q
> (sharper peak) is not likely to be one of them.  Usually the sharper Q
> is obtainable just by increasing the performance Q of the
> second-order.  It’s not like the case with low-pass where extra poles
> give you a sharper roll-off rate JUST BEYOND the corner (like 24 db
> instead of 12).  With a high-Q BP, the “skirts” begin well after the
> response is well below the peak.
> If you DO want sharper “skirts” (say 12 db on each side of the peak
> instead of the 2^nd -order 6db on each side), then a cascading of
> identical BPs is the ticket. The Q increases in such cases, but
> distressingly slowly- if I recall. So it is probably not even a good
> way to sharpen a BP.
> The real use of (and meanings of) higher-order BP is to achieve a
> flatter passband.  The poles are slightly offsets so that when we move
> away from one, with a declining gain, the next pole takes over.  This
> is so-called “stagger-tuning” - a difficult design.
> But the design is easy if one uses a lowpass-to-bandpass
> transformation for Butterworth or Chebyshev:
>   http://electronotes.netfirms.com/AN336.pdf
> _______________________________________________
> Synth-diy mailing list
> Synth-diy at synth-diy.org
> http://synth-diy.org/mailman/listinfo/synth-diy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20191003/c171705a/attachment.htm>

More information about the Synth-diy mailing list