Roman Rumian wrote: ...> Many thanks for your explanation - I read this brochure carefully. > Have you heard about the similar device, or the EQI-LP is the only one ?It seems every harddisk editor and plug-in manufacturer is nowadays offering a linear-phase eq. Google for linear phase eq. I don't know of any other hardware eq though. Regards, Andor
Parametric FIR audio equalizer
Started by ●October 14, 2004
Reply by ●October 18, 20042004-10-18
Reply by ●October 19, 20042004-10-19
Andor wrote:> Roman Rumian wrote: > ... > >>Many thanks for your explanation - I read this brochure carefully. >>Have you heard about the similar device, or the EQI-LP is the only one ? > > > It seems every harddisk editor and plug-in manufacturer is nowadays > offering a linear-phase eq. Google for linear phase eq. I don't know > of any other hardware eq though.And why that is I just cannot figure out. I've got a control point based graphic equalizer that is part of a Mac app that is in progress which can switch between minimum and linear phase realizations. Minimum phase always sounds better to my ears. The stereo imaging is better and there is a crispness that is not matter of HF response but more of "definition". Subtle to be sure, but my ears are not particularly golden, with some industrial damage from my summer job in my college days, and I can hear the difference pretty clearly. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●October 19, 20042004-10-19
Bob Cain wrote:> Andor wrote: > > It seems every harddisk editor and plug-in manufacturer is nowadays > > offering a linear-phase eq. Google for linear phase eq. I don't know > > of any other hardware eq though. > > And why that is I just cannot figure out. I've got a > control point based graphic equalizer that is part of a Mac > app that is in progress which can switch between minimum and > linear phase realizations. Minimum phase always sounds > better to my ears.Well, I guess there is linear-phase eq and then there is linear-phase eq. It seems that some people already mess up the implementation of second order biquad parametric eq. With linear-phase eq, there is even more possibility to screw up. Perhaps you are hearing some short-comings of the implementation. In fact, when I read this> The stereo imaging is better and there > is a crispness that is not matter of HF response but more of > "definition".I would strongly suggest you try some other product (ours? :-). At least, I would consider to try a parametric linear-phase eq. In my opinion, constant Q graphic EQ is ergonomically unsuited for audio work (a debatable matter of taste, I agree). Bob, I know your reservation towards linear-phase eq with regards to pre-ringing. Nobody is saying that linear-phase eq is the solution to every problem. Just sometimes, it seems that it solves the problem at hand better than the standard minimum-phase eq. This depends on what you have to do, and on what kind of material you are working on. From what I gather, our linear-phase eq customers are correcting tracks in a way which is impossible for them with minimum-phase (the EQ1-LP is switchable between linear- and minimum-phase). Some time ago, there was a discussion here whether in fact the optimal phase response was somewhere in between linear and minium (or perhaps even maximum?). I base this assumption on an article by Michael Gerzon: http://www.audiosignal.co.uk/Why%20do%20equalisers%20sound%20different.html What do you think of this? I'm wondering whether it would be useful to have a phase controller in addition to the standard parametric eq controls. The phase controller would go from minimum over linear to maximum phase response. Regards, Andor
Reply by ●October 19, 20042004-10-19
Andor wrote:> [ *SNIP* ]> In my opinion, constant Q graphic EQ is ergonomically unsuited > for audio work (a debatable matter of taste, I agree).>/preamble Admitting that someone(s) will be able to say "I/We told you so" ;) I'll dive into my question. My interest in DSP derives from my interest in improving robustness of restricted forms of speech recognition when subject to restricted forms of acoustic interference. I know just enough of both fields to get me in trouble. /end preamble I wish to manipulate an audio stream (speech), sampled at 44 kHz. Real time is *NOT* required - source is a CD. Is there a filter (FIR/IIR/other) such that given ONE input producing SIX outputs 1. All outputs have the same time delay (measured in sample clocks) 1.a. Specifically number of calculations for each output NOT relevant 2. Q >=10 and easily adjustable 3. center frequencies will vary over ~< 20:1 range 4. available as Scilab (or equiv ) code Bonus Question What should I have been asking for? BTW I've discovered a similarity between "teaching DSP" and "instructing a child in theology" -- answer only question asked ;}
Reply by ●October 19, 20042004-10-19
Richard Owlett wrote:> Andor wrote: > >> [ *SNIP* ] > >> In my opinion, constant Q graphic EQ is ergonomically unsuited >> for audio work (a debatable matter of taste, I agree). > >> > > /preamble > Admitting that someone(s) will be able to say "I/We told you so" ;) > I'll dive into my question. > > My interest in DSP derives from my interest in improving robustness of > restricted forms of speech recognition when subject to restricted forms > of acoustic interference. I know just enough of both fields to get me in > trouble. > /end preamble > > I wish to manipulate an audio stream (speech), sampled at 44 kHz. > Real time is *NOT* required - source is a CD. > > Is there a filter (FIR/IIR/other) such that given ONE input producing > SIX outputs > > 1. All outputs have the same time delay (measured in sample clocks) > 1.a. Specifically number of calculations for each output NOT relevant > 2. Q >=10 and easily adjustable > 3. center frequencies will vary over ~< 20:1 range > 4. available as Scilab (or equiv ) code > > Bonus Question > > What should I have been asking for? > > BTW I've discovered a similarity between "teaching DSP" and "instructing > a child in theology" -- answer only question asked ;}If you need a filter with *the* delay, then you need a linear-phase filter, which pretty much (but not absolutely) specifies a symmetric FIR. Getting all the filters with the same delay is then easy; just make them the same length. The other details I leave to you. Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●October 20, 20042004-10-20
Andor wrote:> Perhaps you are hearing some > short-comings of the implementation.That's always possible, of course, but calculations for deriving the filter are all long floating point and the resulting impulse responses, at short floating point precision, pass the tests I can think of to test for linear phase and minimum phase using Matlab as the analytic tool.> I would strongly suggest you try some other product (ours? :-). At > least, I would consider to try a parametric linear-phase eq. In my > opinion, constant Q graphic EQ is ergonomically unsuited for audio > work (a debatable matter of taste, I agree).I agree with you. My goal was a curve based equalizer using various selectable control point algorithms for specifying an arbitrary magnitude function. This is quite different than what I understand a parametric eq to be. The ability to switch between linear and minimum phase realizations was as much for my own edification as for the end product. As I said, and in considerable disagreement with Gerzon as to the audibility of very minor phase changes, I find to difference to be subtle at best and at the edge of my ability to discriminate. Yet, for me, there is a clear subjective preference for the minimum phase that I find very difficult to characterize. It's about as elusive as the idea of listening fatigue which I know is a weak fallback. I must plead guilty also to not putting my discrimination yet to the ABX test.> > Bob, I know your reservation towards linear-phase eq with regards to > pre-ringing.Yes, and Gerzon (from the excellent discussion you link to below) said it well before I began to even think about it, "Everything that is known about the way the ears perceive transients suggests that, all other things being equal, a pre-response (ie before the main impulse) in a filter will have more audible effect than a similar mirror-image post-response after the main impulse. This is not just consistent with Lagadec�s findings on his digital filter, but is also consistent with the Haas Effect, whereby transient sounds tend to be preferentially localised by the transient arriving at the ear first, with later transients (up to about 40ms later, when separate echoes are heard) playing a reduced role. This is also consistent with the physiological effect of forward inhibition or temporal masking, whereby the perception of stimuli tends to suppress or reduce the sensitivity to the perception of stimuli following immediately afterwards."> From > what I gather, our linear-phase eq customers are correcting tracks in > a way which is impossible for them with minimum-phase (the EQ1-LP is > switchable between linear- and minimum-phase).Well, there's no arguing with subjective impression, I have just found my own to be different.> > Some time ago, there was a discussion here whether in fact the optimal > phase response was somewhere in between linear and minium (or perhaps > even maximum?). I base this assumption on an article by Michael > Gerzon: > http://www.audiosignal.co.uk/Why%20do%20equalisers%20sound%20different.html > > What do you think of this?I think he covers all the various considerations in the fashion that can be expected of Gerzon. Few can do it better, AFAIC. I am surprised at his emphasis on the audibility of minute effects. My ears aren't that good. This does correlate well, though, with the interminable discussions I've seen among A/D designers and high end production people as to the effects of the front end filter. They claim audibility of ridiculously small effects near Nyquist (but what may be telling is that none seem willing to put their discrimination ability to the ABX test.) I must disagree with the following observation of Gerzon's based on a good deal of work with measurement based speaker and microphone transformations: "One area of pessimism concerns the viability of using equalisers to compensate for defects in other equipment (microphones, loudspeakers and even multiple stages of bass roll-off in audio electronics). The problem here is that even very tiny residual errors in the frequency and phase responses may turn out to be almost as audible (or in some cases even more so) than the original unequalised errors. Equalisation may improve the tonal accuracy in such cases but it can (and often does) increase the audible colouration." It is always dangerous to disagree with Micheal Gerzon, though, and I well know that. :-) I wasn't aware of that archive of his work and want to thank you for the link. I'm working currently on a method for calculating encoders empirically, based on measurement, for the tetrahedral array he invented and there is much there that I want to read.> I'm wondering whether it would be useful to have a phase controller in > addition to the standard parametric eq controls. The phase controller > would go from minimum over linear to maximum phase response.I think that is a _very_ good idea. The problem I see is that while those three points are well defined, what you might do between them is enormously flexible. Any ideas that you would be comfortable discussing publicly? Thanks, Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●October 20, 20042004-10-20
On Tue, 19 Oct 2004 20:54:00 -0700, Bob Cain wrote:>> I'm wondering whether it would be useful to have a phase controller in >> addition to the standard parametric eq controls. The phase controller >> would go from minimum over linear to maximum phase response. > > I think that is a _very_ good idea. The problem I see is > that while those three points are well defined, what you > might do between them is enormously flexible. Any ideas > that you would be comfortable discussing publicly?Many years ago ('95-ish, from memory) David McGrath and I presented a paper at an AES conference on "optimal phase" FIR EQ synthesis. I think. It's certainly been a feature of the EQ software for Huron systems for about that long. The user can choose minimum or linear phase, or dial-in a desired group delay. I don't think that we ever bothered to allow dialing-in group delays that would result in maximum-phase filters though. I'll see if I can dig up a paper reference for you. The papers used to be on-line on the web site, but I think that they've fallen of in one of the re-designs. The implementation is a simple generalization of the Hilbert transform process that one usually uses to get a minimum-phase impulse response. http://www.lake.com.au/driver.asp?page=main/products/huron/modular+software You might also be interested in Lake's "Contour" product line, which includes some fairly fancy arbitrary-curve FIR EQ features, and a neat GUI to drive it. http://www.proaudio.lake.com/ [I do work for Lake, but was not invloved directly with these particular products, so I'm afraid that I can't tell much more about them.] -- Andrew (also A.Reilly at lake.com)
Reply by ●October 20, 20042004-10-20
On 2004-10-18 10:48:39 +0200, an2or@mailcircuit.com (Andor) said:> Roman Rumian wrote: > ... >> Many thanks for your explanation - I read this brochure carefully. >> Have you heard about the similar device, or the EQI-LP is the only one ? > > It seems every harddisk editor and plug-in manufacturer is nowadays > offering a linear-phase eq. Google for linear phase eq. I don't know > of any other hardware eq though. > > Regards, > AndorWell, yes, and to what end. If you ask me, it's because a linear phase EQ is easy to design (from a non-DSP developer's perspective who thinks poles and zeros are nationalities) - "FFT and multiply" is fairly easy to understand. Not forgetting that today processor power is no longer an issue, so FFT away and be done with it. Linear phase EQs also have the desirable property of sounding good wrt. marketing (not necessarily wrt. the sonic results). The "linear" buzzword is always good for sales! :-) If you really compare a good non-linear phase EQ (all analog EQs I know are of this type) to any of the recent linear phase EQ you'll soon realize that linearity doesn't make it sound good. In fact, most linear phase EQs sound horrible. I for one would rather spend my money on stuff that sounds good, be it linear, non-linear or else. In a way, this reminds me of the discussions I've seen in the days when the CD was something new: is "DDD" better than "AAD"? Discuss! :-) And in the end, the final result is presented to us through a highly non-linear device anyway: our very ears! -- Stephan M. Bernsee http://www.dspdimension.com
Reply by ●October 20, 20042004-10-20
On 2004-10-16 06:51:05 +0200, PaavoJumppanen@iname.com (Paavo Jumppanen) said:> > Thanks Rick, > > We had a bit of an infamous start about a year ago being branded snake > oil merchants.I must admit I too was about to dig out my popular Furutech CD Demagnetizer link again when I first saw the web site... but not! Apparently this is an EQ of some sort. Even though the MP3s don't really help much (for some tracks, the "before" version sounded better than the processed version to me [too much midrange], but I guess that's just the normal exaggeration on the demo files to demonstrate the effect) I bet it can be used to help correcting problems. There's a lot more to a good mix than just the proper overall spectral balance (stereo placement of instruments/stereo width, proper application of reverb, good adjustment of the dynamics etc.) so I wouldn't quite agree with all of the claims made on your web site, but it's a start and probably a nice tool for people who just want to make some quick and easy to set up adjustments. -- Stephan M. Bernsee http://www.dspdimension.com
Reply by ●October 20, 20042004-10-20
Jerry Avins wrote:> Roman Rumian wrote: > >>Hello, >> >>maybe my question is stupid, but does such a device/algorithm exist in >>practice ? > > I know what a parametric amplifier is, but not a parametric equalizer. > There is such a thing, but I'm not familiar with its specs. Matlab code: > http://www.mathworks.com/matlabcentral/fileexchange/loadFile.do?objectId=1963 > > How would one change the equalization parameters in an FIR?A time domain method is to window the the impulse response of an IIR parametric equalizer. With enough taps, the FIR can approximate many but not all frequency responses of the IIR. In particular, the FIR will need to be quite long for low filter frequencies or high Q's, and combining those two would probably increase the FIR length too much.
