Reply by Vladimir Vassilevsky●October 15, 20102010-10-15
Ron N. wrote:
> On Oct 12, 5:54 pm, Robert Adams <robert.ad...@analog.com> wrote:
>
>>I am looking for a way to design an audio IIR filterbank. I need to
>>split an audio signal into roughly 1/3 octave bands with 6th-order or
>>8th-order IIR bandpass filters and then sum them back together again
>>with a ripple of +/- 1 dB. I can do a "seat-of-the-pants" design that
>>works well until I approach Nyquist where it kind of falls apart. I
>>have very limited MIPs so I can't do anything very fancy. Any
>>pointers?
>
>
> Has anyone else tried mixing a bank of bandpass filter for
> the middle of the spectrum plus a low pass filter for near
> DC and a high pass filter for near Fs/2?
Isn't it a pretty standard approach? I did that for EQ filters, it was
beneficial for some kinds of responses.
IIRC Dolby-A used a set of basic filters and then added/subtracted their
outputs to get more frequency bands. That allowed saving on the parts,
as the filters were implemented in analog.
Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com
Reply by Ron N.●October 14, 20102010-10-14
On Oct 12, 5:54�pm, Robert Adams <robert.ad...@analog.com> wrote:
> I am looking for a way to design an audio IIR filterbank. I need to
> split an audio signal into roughly 1/3 octave bands with 6th-order or
> 8th-order IIR bandpass filters and then sum them back together again
> with a ripple of +/- 1 dB. I can do a "seat-of-the-pants" design that
> works well until I approach Nyquist where it kind of falls apart. I
> have very limited MIPs so I can't do anything very fancy. Any
> pointers?
Has anyone else tried mixing a bank of bandpass filter for
the middle of the spectrum plus a low pass filter for near
DC and a high pass filter for near Fs/2?
--
rhn A.T nicholson d.0.t C-o-M
Reply by Philippe●October 14, 20102010-10-14
On Oct 14, 1:44�am, Robert Adams <robert.ad...@analog.com> wrote:
> On Oct 13, 7:34�pm, Philippe <kelvin.bitn...@gmail.com> wrote:
>
>
>
>
>
> > On Oct 13, 10:42�am, Robert Adams <robert.ad...@analog.com> wrote:
>
> > > On Oct 13, 4:37�am, Robert Adams <robert.ad...@analog.com> wrote:
>
> > > > On Oct 13, 12:12�am, robert bristow-johnson
>
> > > > <r...@audioimagination.com> wrote:
> > > > > On Oct 13, 12:10�am, robert bristow-johnson
>
> > > > > <r...@audioimagination.com> wrote:
> > > > > > On Oct 12, 11:59�pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>
> > > > > > > P.S. You do know of Linkwitz-Riley filters, do you?
>
> > > > > > you might find both Bob's name and Siegfried Linkwitz listed here:http://www.aes.org/awards/.
>
> > > > > fuck! �i can't believe it. �i *knew* i would see Bob's name on the
> > > > > list because i know a couple of times he got AES awards. �but i cannot
> > > > > believe that Linkwitz's name is not on the list. �unbelievable! �the
> > > > > AES never awarded Linkwitz and Riley a fellowship or, at least, a
> > > > > publication award???!!
>
> > > > > can't figger that one out.
>
> > > > > r b-j
>
> > > > Yes I know about L-R, but I am looking for something with > 30
> > > > bandpass filters, not sure L-R can be extended to that case?
>
> > > > Bob- Hide quoted text -
>
> > > > - Show quoted text -
>
> > > Application is a bit like a graphic EQ but I need to measure the
> > > energy in each band with pretty good selectivity, and my MIPs are so
> > > limited that I wanted to use the same filters for energy estimation as
> > > for the graphic EQ, which means I need to use the "parallel-sum" form
> > > of graphic EQ instead of the "cascade-of-peaking-filters" form.
> > > Unfortunately this means you get all the messy phase effects when each
> > > filter crosses over with its upper and lower neighbors.
>
> > > Bob
>
> > the parallel-sum network plus a matching phase in each cell, with pure
> > delay PLUS a fractional (<z-1) with an all pass ?- Hide quoted text -
>
> > - Show quoted text -
>
> I like this idea, thanks. I tried this already with integer delay, but
> it's too coarse near Nyquist, so you do need the allpass fractional
> part as you suggest. I'm tempted to throw all this at the Matlab non-
> linear optimizer, which I have found to be a wonderful crutch when you
> don't really know what you're doing.
Reply by robert bristow-johnson●October 14, 20102010-10-14
On Oct 13, 7:42�pm, Robert Adams <robert.ad...@analog.com> wrote:
>
> Maybe next time we have lunch you can slip the solution under the
> table and I'll pay the tab :)
it seems i can get you to do that anyway. :-)
you'll have to bring your fambly up to VT and i can return the
consideration. it's nice up here now. a little cool (i think i have
to fire up the heat tomorrow night) but the colors are wunnerful.
>
> The problem is that to get the selectivity I need, the parallel
> filters all have to be at least 6th-order.
them's 6th-orders can be split into parallel 2nd-orders also.
but i wouldn't know what to do to get it to add up to a wire if every
slider is set to 0 dB.
r b-j
Reply by Robert Adams●October 13, 20102010-10-13
On Oct 13, 7:34�pm, Philippe <kelvin.bitn...@gmail.com> wrote:
> On Oct 13, 10:42�am, Robert Adams <robert.ad...@analog.com> wrote:
>
>
>
>
>
> > On Oct 13, 4:37�am, Robert Adams <robert.ad...@analog.com> wrote:
>
> > > On Oct 13, 12:12�am, robert bristow-johnson
>
> > > <r...@audioimagination.com> wrote:
> > > > On Oct 13, 12:10�am, robert bristow-johnson
>
> > > > <r...@audioimagination.com> wrote:
> > > > > On Oct 12, 11:59�pm, Vladimir Vassilevsky <nos...@nowhere.com> wrote:
>
> > > > > > P.S. You do know of Linkwitz-Riley filters, do you?
>
> > > > > you might find both Bob's name and Siegfried Linkwitz listed here:http://www.aes.org/awards/.
>
> > > > fuck! �i can't believe it. �i *knew* i would see Bob's name on the
> > > > list because i know a couple of times he got AES awards. �but i cannot
> > > > believe that Linkwitz's name is not on the list. �unbelievable! �the
> > > > AES never awarded Linkwitz and Riley a fellowship or, at least, a
> > > > publication award???!!
>
> > > > can't figger that one out.
>
> > > > r b-j
>
> > > Yes I know about L-R, but I am looking for something with > 30
> > > bandpass filters, not sure L-R can be extended to that case?
>
> > > Bob- Hide quoted text -
>
> > > - Show quoted text -
>
> > Application is a bit like a graphic EQ but I need to measure the
> > energy in each band with pretty good selectivity, and my MIPs are so
> > limited that I wanted to use the same filters for energy estimation as
> > for the graphic EQ, which means I need to use the "parallel-sum" form
> > of graphic EQ instead of the "cascade-of-peaking-filters" form.
> > Unfortunately this means you get all the messy phase effects when each
> > filter crosses over with its upper and lower neighbors.
>
> > Bob
>
> the parallel-sum network plus a matching phase in each cell, with pure
> delay PLUS a fractional (<z-1) with an all pass ?- Hide quoted text -
>
> - Show quoted text -
I like this idea, thanks. I tried this already with integer delay, but
it's too coarse near Nyquist, so you do need the allpass fractional
part as you suggest. I'm tempted to throw all this at the Matlab non-
linear optimizer, which I have found to be a wonderful crutch when you
don't really know what you're doing.
Reply by Robert Adams●October 13, 20102010-10-13
On Oct 13, 12:52=A0pm, robert bristow-johnson
<r...@audioimagination.com> wrote:
> On Oct 13, 4:42=A0am, Robert Adams <robert.ad...@analog.com> wrote:
> ...
>
> > Application is a bit like a graphic EQ but I need to measure the
> > energy in each band with pretty good selectivity, and my MIPs are so
> > limited that I wanted to use the same filters for energy estimation as
> > for the graphic EQ, which means I need to use the "parallel-sum" form
> > of graphic EQ instead of the "cascade-of-peaking-filters" form.
> > Unfortunately this means you get all the messy phase effects when each
> > filter crosses over with its upper and lower neighbors.
>
> Bob, i might have mentioned this to you one of those times we were
> having lunch at the Green Papaya, but an idea for a paper that i had
> but have done little about is an automated way of turning that
> "cascade-of-peaking-filters" form into an equivalent parallel-sum
> form:
>
> a graphic EQ with 3 sliders per octave, and a pretty constant 1/3
> octave bandwidth on each peaking filter. =A0but then apply
> (automagically) the Heaviside partial fraction expansion to this thing
> and come out with a bunch of parallel 2nd-order filters (each with 2
> coefs in the numerator) and one feedforward path of the dry input. =A0i
> sorta think this would be a good thing for an EE grad student to do.
>
> the reason i liked it this way is because the double-precision word
> that results from every parallel filter need not be truncated, but
> summed in a big accumulator with that result truncated (and noise-
> shaped). =A0being parallel, it might be cleaner than 25 peaking-EQs in
> series, each quantizing its output. =A0but being the same transfer
> function as the 25 peaking-EQs in a row, it wouldn't have the phase
> cancellation problem of the side skirts of adjacent bands typical in
> the parallel context.
>
> r b-j
Thanks RBJ
Maybe next time we have lunch you can slip the solution under the
table and I'll pay the tab :)
The problem is that to get the selectivity I need, the parallel
filters all have to be at least 6th-order.
Bob
Reply by Robert Adams●October 13, 20102010-10-13
On Oct 13, 2:08=A0pm, dbd <d...@ieee.org> wrote:
> On Oct 12, 5:54 pm, Robert Adams <robert.ad...@analog.com> wrote:
>
> > ...
> > I can do a "seat-of-the-pants" design that
> > works well until I approach Nyquist where it kind of falls apart. I
> > have very limited MIPs so I can't do anything very fancy. Any
> > pointers?
>
> > Bob
>
> The off-the-wall answer is "Don't go near Nyquist." One way to do that
> is to upsample the signal only for those filters that are too close to
> Nyquist for your design technique. Upsample by inserting alternate
> zeros, design the filters as you are now, but at half the center
> frequency and bandwidth. You may have those designs already. After
> each filter, desample to the original sample frequency by dropping
> alternate samples. The capricious resampling works because of the
> narrow filters that you application consists of remove aliasing
> artifacts.
>
> Downside: more processing
> Upside: uses design techniques you already have
> YMMV
>
> Dale B. Dalrymple
Dale
Thanks, I did think about this but I'm out of MIPs, unfortunately.
Reply by Philippe●October 13, 20102010-10-13
On Oct 13, 10:42=A0am, Robert Adams <robert.ad...@analog.com> wrote:
> On Oct 13, 4:37=A0am, Robert Adams <robert.ad...@analog.com> wrote:
>
>
>
>
>
> > On Oct 13, 12:12=A0am, robert bristow-johnson
>
> > <r...@audioimagination.com> wrote:
> > > On Oct 13, 12:10=A0am, robert bristow-johnson
>
> > > <r...@audioimagination.com> wrote:
> > > > On Oct 12, 11:59=A0pm, Vladimir Vassilevsky <nos...@nowhere.com> wr=
ote:
>
> > > > > P.S. You do know of Linkwitz-Riley filters, do you?
>
> > > > you might find both Bob's name and Siegfried Linkwitz listed here:h=
ttp://www.aes.org/awards/.
>
> > > fuck! =A0i can't believe it. =A0i *knew* i would see Bob's name on th=
e
> > > list because i know a couple of times he got AES awards. =A0but i can=
not
> > > believe that Linkwitz's name is not on the list. =A0unbelievable! =A0=
the
> > > AES never awarded Linkwitz and Riley a fellowship or, at least, a
> > > publication award???!!
>
> > > can't figger that one out.
>
> > > r b-j
>
> > Yes I know about L-R, but I am looking for something with > 30
> > bandpass filters, not sure L-R can be extended to that case?
>
> > Bob- Hide quoted text -
>
> > - Show quoted text -
>
> Application is a bit like a graphic EQ but I need to measure the
> energy in each band with pretty good selectivity, and my MIPs are so
> limited that I wanted to use the same filters for energy estimation as
> for the graphic EQ, which means I need to use the "parallel-sum" form
> of graphic EQ instead of the "cascade-of-peaking-filters" form.
> Unfortunately this means you get all the messy phase effects when each
> filter crosses over with its upper and lower neighbors.
>
> Bob
the parallel-sum network plus a matching phase in each cell, with pure
delay PLUS a fractional (<z-1) with an all pass ?
Reply by Philippe●October 13, 20102010-10-13
On Oct 13, 10:42=A0am, Robert Adams <robert.ad...@analog.com> wrote:
> On Oct 13, 4:37=A0am, Robert Adams <robert.ad...@analog.com> wrote:
>
>
>
>
>
> > On Oct 13, 12:12=A0am, robert bristow-johnson
>
> > <r...@audioimagination.com> wrote:
> > > On Oct 13, 12:10=A0am, robert bristow-johnson
>
> > > <r...@audioimagination.com> wrote:
> > > > On Oct 12, 11:59=A0pm, Vladimir Vassilevsky <nos...@nowhere.com> wr=
ote:
>
> > > > > P.S. You do know of Linkwitz-Riley filters, do you?
>
> > > > you might find both Bob's name and Siegfried Linkwitz listed here:h=
ttp://www.aes.org/awards/.
>
> > > fuck! =A0i can't believe it. =A0i *knew* i would see Bob's name on th=
e
> > > list because i know a couple of times he got AES awards. =A0but i can=
not
> > > believe that Linkwitz's name is not on the list. =A0unbelievable! =A0=
the
> > > AES never awarded Linkwitz and Riley a fellowship or, at least, a
> > > publication award???!!
>
> > > can't figger that one out.
>
> > > r b-j
>
> > Yes I know about L-R, but I am looking for something with > 30
> > bandpass filters, not sure L-R can be extended to that case?
>
> > Bob- Hide quoted text -
>
> > - Show quoted text -
>
> Application is a bit like a graphic EQ but I need to measure the
> energy in each band with pretty good selectivity, and my MIPs are so
> limited that I wanted to use the same filters for energy estimation as
> for the graphic EQ, which means I need to use the "parallel-sum" form
> of graphic EQ instead of the "cascade-of-peaking-filters" form.
> Unfortunately this means you get all the messy phase effects when each
> filter crosses over with its upper and lower neighbors.
>
> Bob
ahah
do yourself a favor, buy yourself an intel one and run linux on top.
Reply by dbd●October 13, 20102010-10-13
On Oct 12, 5:54 pm, Robert Adams <robert.ad...@analog.com> wrote:
> ...
> I can do a "seat-of-the-pants" design that
> works well until I approach Nyquist where it kind of falls apart. I
> have very limited MIPs so I can't do anything very fancy. Any
> pointers?
>
> Bob
The off-the-wall answer is "Don't go near Nyquist." One way to do that
is to upsample the signal only for those filters that are too close to
Nyquist for your design technique. Upsample by inserting alternate
zeros, design the filters as you are now, but at half the center
frequency and bandwidth. You may have those designs already. After
each filter, desample to the original sample frequency by dropping
alternate samples. The capricious resampling works because of the
narrow filters that you application consists of remove aliasing
artifacts.
Downside: more processing
Upside: uses design techniques you already have
YMMV
Dale B. Dalrymple