Graphical EQ like PEQ4 but with more bands?
Graphical EQ like PEQ4 but with more bands?
I'm currently using PEQ4 to calibrate my sound system, using a measurement mic and pink noise to measure my monitors in my room and applying filters to my signal chain to get a smooth response.
Because PEQ4 only supports 4 bands of EQ I am ending up daisychaining several of these modules together, which is cumbersome.
Is there a stereo graphical EQ that can support more than four biquads? e.g. a "PEQ-n", if you will!
I considered rolling my own in Modular, but the parametric and shelving EQs in modular are only mono, so I would have to make two parallel signal flows for left and right, and it would be a pain keeping them in synch!
Any suggestions?
Because PEQ4 only supports 4 bands of EQ I am ending up daisychaining several of these modules together, which is cumbersome.
Is there a stereo graphical EQ that can support more than four biquads? e.g. a "PEQ-n", if you will!
I considered rolling my own in Modular, but the parametric and shelving EQs in modular are only mono, so I would have to make two parallel signal flows for left and right, and it would be a pain keeping them in synch!
Any suggestions?
Re: Graphical EQ like PEQ4 but with more bands?
fix the room first.
fixing the speakers only leads to sadness. when you polish a turd, it's still a turd.
you could put another peq in series...
fixing the speakers only leads to sadness. when you polish a turd, it's still a turd.
you could put another peq in series...
-
- Posts: 1638
- Joined: Mon Nov 15, 2010 12:57 pm
Re: Graphical EQ like PEQ4 but with more bands?
I second Gary's comment. By using and chaining EQs to get a flat frequency response, you are destroying the phase relationships of those frequencies with each filter added. And phase plays the important part in creating a good sound stage during playback.
You'd be better hanging up a few blankets or foam on the walls to kill those resonant peaks and stray reflections instead.
You'd be better hanging up a few blankets or foam on the walls to kill those resonant peaks and stray reflections instead.
Re: Graphical EQ like PEQ4 but with more bands?
An module to convolve the input with an FIR filter would be a good solution. Does such a thing exist for Scope?
Re: Graphical EQ like PEQ4 but with more bands?
Theres DASGOST ConvolveQ
Re: Graphical EQ like PEQ4 but with more bands?
What about GraphEq?
Isn't this a Stock Device?
Isn't this a Stock Device?
- Attachments
-
- grapheq.JPG (67.4 KiB) Viewed 5348 times
Re: Graphical EQ like PEQ4 but with more bands?
Thanks for the suggestions, all.
I agreed that it's best to treat the room first.
The issue with the fixed-band EQs is that they don't have enough frequency resolution. e.g. GraphEQs has low bands at 20, 25, 31, 40, 50 63 and 80. With a mode at 70Hz there's no way to accurately address it. DASGOST has the same limitation.
With an FIR convolver it would be possible to perform very accurate linear phase EQ.
As a bonuse it would also be able to use impulse responses of real spaces and behave as a reverb!
There are even impulse responses available to download from the web for those who don't want to experiment with their own.
Take a look at http://convolver.sourceforge.net/ which does a similar thing in VST land.
I think this would be a great addition to Scope!
I agreed that it's best to treat the room first.
The issue with the fixed-band EQs is that they don't have enough frequency resolution. e.g. GraphEQs has low bands at 20, 25, 31, 40, 50 63 and 80. With a mode at 70Hz there's no way to accurately address it. DASGOST has the same limitation.
With an FIR convolver it would be possible to perform very accurate linear phase EQ.
As a bonuse it would also be able to use impulse responses of real spaces and behave as a reverb!
There are even impulse responses available to download from the web for those who don't want to experiment with their own.
Take a look at http://convolver.sourceforge.net/ which does a similar thing in VST land.
I think this would be a great addition to Scope!
Re: Graphical EQ like PEQ4 but with more bands?
i aggree on the fir but
what you show is fast convolution like convolveQ_ a fft filter which is a special form of a fir_
not useable for realtime just for mastering
with 65536 taps you have a latency of 65536 samples + the host latency_
but good marketing vibes
what you show is fast convolution like convolveQ_ a fft filter which is a special form of a fir_
not useable for realtime just for mastering
with 65536 taps you have a latency of 65536 samples + the host latency_
but good marketing vibes
Re: Graphical EQ like PEQ4 but with more bands?
I've not tried convolveQ_ so cannot speak as to its latency.
My understanding of FIRs is that each sample is passed through with no latency (see h[0] path below)
Additionally, each sample is then delayed by one sample per coefficient, convolved with that coefficient, and then mixed with the output.
As such, there shouldn't be any latency!
My understanding of FIRs is that each sample is passed through with no latency (see h[0] path below)
Additionally, each sample is then delayed by one sample per coefficient, convolved with that coefficient, and then mixed with the output.
As such, there shouldn't be any latency!
Re: Graphical EQ like PEQ4 but with more bands?
yes but that is NOT what you pointed to in the link__
this is a FFT convolution__ marked the way you might think it is a true fir_
there is NO WAY to process /multiply add/ 65536 taps even at 44k1Hz in stereo on ANY cpu avail________
maybe 2048 in stereo at the fastest cpus avail;
but then nothing else will run on that pc_
AND you will have dropouts_
as for the true fir df1 this is also only valid depending on the coeffs you use_
and sorry but the pic is not completly correct;
it should be z-1 z-2 to z-n
if z is the inputsample and -n the timeindex;
good vibes
this is a FFT convolution__ marked the way you might think it is a true fir_
there is NO WAY to process /multiply add/ 65536 taps even at 44k1Hz in stereo on ANY cpu avail________
maybe 2048 in stereo at the fastest cpus avail;
but then nothing else will run on that pc_
AND you will have dropouts_
as for the true fir df1 this is also only valid depending on the coeffs you use_
and sorry but the pic is not completly correct;
it should be z-1 z-2 to z-n
if z is the inputsample and -n the timeindex;
good vibes
Re: Graphical EQ like PEQ4 but with more bands?
Is there any way to implement a true FIR filter on scope? 1024 or even 512 taps would be sufficient I think.
I have the tools to generate coefficients to any tap length, and could just truncate them down to 1024 or whatever is feasible to run on Scope.
I have not developed for Scope so I do not know if this is possible with the SDK.
I have the tools to generate coefficients to any tap length, and could just truncate them down to 1024 or whatever is feasible to run on Scope.
I have not developed for Scope so I do not know if this is possible with the SDK.
Re: Graphical EQ like PEQ4 but with more bands?
basically it is possible with feedback and adds;
thisway you may get 128 taps on 1 old dsp;
but to get out the max of it you better code your own fir in assembler;
the dsps are made for this;
readout of sample AND coeff AND multiply AND add in 1 circle;
we have a 4096 prototype running on 1 new dsp with headroom at 44k1Hz
good vibes
thisway you may get 128 taps on 1 old dsp;
but to get out the max of it you better code your own fir in assembler;
the dsps are made for this;
readout of sample AND coeff AND multiply AND add in 1 circle;
we have a 4096 prototype running on 1 new dsp with headroom at 44k1Hz
good vibes
- Attachments
-
- FireQ_01.JPG (49.14 KiB) Viewed 5309 times
Re: Graphical EQ like PEQ4 but with more bands?
Thanks for the insights, tgstgs.
Looking forward to seeing the finished FireQ!
Looking forward to seeing the finished FireQ!
Re: Graphical EQ like PEQ4 but with more bands?
it's all moot.
you can't eq a three dimensional space and have the curve be proper for all listening positions(each centimeter) anyway.
fix the room.
is it audio that you care about or computers? if it's audio, then handle it the correct way. eqs are used in the real world, but a good control room is eq'd in the design of the room. eqs are a last resort, or a bandaid. use computers for what they do best. fixing the room isn't one of those things.
or that's what my opinion is anyway...
i'm certainly nor discouraging anyone from making such an eq however, and you can be sure that the GHOST products are done by someone who cares about his work and KNOWS what he is doing and talking about.
anyway, the reason i'm so adamant is that i spent years trying to use eqs to fix my room. i used good stuff, too. i used hardware that had come from real working environments, White, Klark Technic, Urei, and Orban, but i could never hear the speakers properly. sure, i could measure the room response in a couple of places and get a sweetspot where i had dependable flat response, but i couldn't move much. also, the real problems weren't fixable via eq. even with excellent speaks and a measured "flat" response, the image was smeared and murky. there's more to hearing the system than perfectly flat eq of the system.
once i put up a few diffusers and a couple of bass traps things got a lot better. i removed the eqs and lo and behold, i was able to make good judgements based on what i could hear even though the rooms respose was far from flat. suddenly, i had an idea of what was really going on! the point of the rant is that it's better to fix the room a little and NOT eq the system at all, than to eq the room with the finest eq, but leave the room as it is.
you can't eq a three dimensional space and have the curve be proper for all listening positions(each centimeter) anyway.
fix the room.
is it audio that you care about or computers? if it's audio, then handle it the correct way. eqs are used in the real world, but a good control room is eq'd in the design of the room. eqs are a last resort, or a bandaid. use computers for what they do best. fixing the room isn't one of those things.
or that's what my opinion is anyway...
i'm certainly nor discouraging anyone from making such an eq however, and you can be sure that the GHOST products are done by someone who cares about his work and KNOWS what he is doing and talking about.
anyway, the reason i'm so adamant is that i spent years trying to use eqs to fix my room. i used good stuff, too. i used hardware that had come from real working environments, White, Klark Technic, Urei, and Orban, but i could never hear the speakers properly. sure, i could measure the room response in a couple of places and get a sweetspot where i had dependable flat response, but i couldn't move much. also, the real problems weren't fixable via eq. even with excellent speaks and a measured "flat" response, the image was smeared and murky. there's more to hearing the system than perfectly flat eq of the system.
once i put up a few diffusers and a couple of bass traps things got a lot better. i removed the eqs and lo and behold, i was able to make good judgements based on what i could hear even though the rooms respose was far from flat. suddenly, i had an idea of what was really going on! the point of the rant is that it's better to fix the room a little and NOT eq the system at all, than to eq the room with the finest eq, but leave the room as it is.
Re: Graphical EQ like PEQ4 but with more bands?
Would this be easier on the 2146X series Sharcs, w/ built in hardware accelerators for FIR, FFT & IIR filtering ?tgstgs wrote:i aggree on the fir but what you show is fast convolution like convolveQ_ a fft filter which is a special form of a fir_
not useable for realtime just for mastering with 65536 taps you have a latency of 65536 samples + the host latency_ but good marketing vibes
Re: Graphical EQ like PEQ4 but with more bands?
sorry i dont get the question;
4096taps on 1 new xite dsp = not possible with a dualcore cpu_
good vibes
4096taps on 1 new xite dsp = not possible with a dualcore cpu_
good vibes
Re: Graphical EQ like PEQ4 but with more bands?
tgstgs wrote: but good marketing vibes
Re: Graphical EQ like PEQ4 but with more bands?
1 new xite dsp = 21369 Sharc.tgstgs wrote:sorry i dont get the question;
4096taps on 1 new xite dsp = not possible with a dualcore cpu_
good vibes
I was asking if it would be easier on 2146X Sharc.
Re: Graphical EQ like PEQ4 but with more bands?
--
sorry there is no yes or no answer in my opinion_
it depends on more than just dsp specs_
as soon as there is a board avail and i have realized a filter for it i tell you_
good vibes
sorry there is no yes or no answer in my opinion_
it depends on more than just dsp specs_
as soon as there is a board avail and i have realized a filter for it i tell you_
good vibes