Playlist Generator "mplay" now available

I have posted the first "public" release of mplay, the playlist generator I have developed, to my dropbox account where anyone with free or better dropbox account can download.  This is not yet configured for a gui-friendly installation.  You must use Mac or Linux, and be somewhat familiar (first day level) with command line interfaces.  Once installed, however, you can click on icons to use (as I do).  Sample script/icons are included.

The link is

https://www.dropbox.com/s/flgeagiur08z40i/mplay.tar.gz?dl=0

I have added the project to Source Forge under the name makePlaylist, which can be accessed here:

https://sourceforge.net/projects/makeplaylist/

Unipolar vs Dipolar response

Part of upgrading the bedroom system would have to include before and after measurements.  The Before measurements would provide an objective target to set the levels and such.

However the current bedroom system hasn't been much of a focus for many years now, despite having "descended" (or ascended) from my very first audio system, with replacements.  There is nothing in that descent predating 2005 at this point.  That was when I got the Revel M20's.

A few years ago, the living room Tact RCS 2.0 died, so I replaced it with the bedroom one, leaving the bedroom short of the "digital preamp" (source plus level control if nothing more, I never seriously tried the RCS and what little I did I didn't like it, found it frustrating, etc) I had relied on since 2009 or so.  But since at least 2016, if not 2009, the bedroom system was a distant second in getting my constant attentions as the living room.  In fact, it didn't really need the Tact anymore, since I had basically only one "source" I ever used...the TV.  Or, possibly, the turntable/tape/digital recording hub on the side of the room, which has been used in recent years mostly for cassette to digital transcriptions, though at one time was my primary vinyl transcription also.  When doing those operations, I need to be listening to what I'm doing on the bedroom system.  That happens through Sonos, I put the analog inputs to Sonos in (for the Tape node) and listen using the Sonos on the other side of the room.  Any other kind of interconnection would invariably lead to ground loop.  The network cabling the Sonos uses has ground breaks, usually a final length of unshielded cable from the wall panels (all shielded prior to that point).

So, the two essential things by that point were just Sonos and TV.  Which was all handled by the Sonos Zoneplayers themselves, nothing more needed.

Except for the EQ and crossover, and here's where the rub is.  All the historic EQ programming ever since I got my first DCX 2496 and DEQ 2496 wasn't all that great anyway, but work was done, but then lost when those units died, and has now gone through a succession of same units.  I was never good at recording what EQ settings were being used, and once those units died, that information was lost, and I had less time and interest in doing it very well anymore (though, I might have easily done better than before with improved equipment and methods, etc).

So, the system settings had already pretty much devolved to whatever sounded good in the first 30 minutes of setup.  Possibly making the pink noise look good.  But little beyond that.

But, anyway, this time I'm trying to honor past judgements by recording the "before" condition before losing it.  Not merely the settings (the bass crossover set in the DEQ was 71 Hz, which was easily dialed into its replacement miniDSP), but the actual system response.

This turned out more interesting than I had anticipated, because it's now clear that the target response of a unipolar system (as the Revel M20 speakers) and dipolar needs be different.

Here's the M20 with SVS cylinder sub, as measured now but set up pretty quickly a few years ago:

Bedroom Response near Headboard

Wow this is flat and deeply extended.  You might even say it's far better looking than the living room response.  Why shouldn't I equalize/crank the Acoustat High Frequency Control up to get a response more like this?  (The deep bass gets weaker and the midbass stronger as move away from headboard (measuring above it, so this is the best looking response.)

Here's the living room response measurement (I'm back to the long "evolved" upper frequency EQ, because the scratch built version was slightly worse, but further experiments were going to be sufficiently complex that I needed a digital device like DEQ 2496, to be "liberated" from bedroom by the ongoing upgrade the )

Living Room, "Evolved" HF EQ

Utterly different, though generally I think I like the living room system better, and they don't sound as different in HF balance as the measurements suggest.  They sound utterly different as in the Acoustats sounding like there is a life sized concert hall behind them, no fake source of anything anywhere, whereas the bedroom system sounds like everything is being viewed on a 72 inch TV, precise and perhaps even more in focus than real life, but still miniaturized.

I think the drooping response of the electrostats may be required because they light up the room uniformly with high frequencies, while unipoles radiate from a "point source" dome tweeter usually.  The ear is far more sensitive to high frequencies on the side axis that enters the ear canal.  Therefore perhaps what needs to be matched is the ambient measurement.  One way to do this is from the measurement behind the Revel M20's, which looks curiously like the listening position measurement of the Acoustats:

Bedroom behind M20's

Starting the Bedroom Upgrade

 After concluding that I need to liberate the Behringer 2496 DEQ from the master bedroom system so I can use it for EQ fine tuning in the living room, I finally initiated an bedroom audio upgrade project that had been planned over a year ago (which had been in the procrastination stage, after buying the miniDSP OpenDRC 2x2 last year...I had the miniDSP box prominently displayed in the bedroom for the last few months so I'd remember to do this project and thereby get the box out of the way).  But sadly it only got started, now I am waiting on a another new piece of equipment that will take at least a month to acquire (and I worry about whether I'll get it even then, giving supply issues; a couple years ago I could have it next day).

I began by taking a picture of the shelf, cluttered with boxes of gauze, telfa pads, and the box of my new blood pressure monitor. I needed two of these boxes up to September last year for a couple of skin surgeries.  And they served the convenient purpose of blocking the light from the display of the DEQ 2496 to allow me to have a dark bedroom for sleeping.  Even the light(s) of the miniDSP's are too bright, but more easily covered.

I then began the task of setting up the new miniDSP plug in.  I first checked out that the "save" feature, it only saves the "slot" you are working in (and likewise, the "load" function only loads the slot you are working on into the miniDSP, until you switch slots anyway).  That's convenient for me now anyway.  So I saved the existing slot 4, but now I realize it was basically pointless as it simply has everything bypassed.  That's my non-crossover program for the panels in the living room.  It should never be used for the woofer or mids, so I can use the slot 4 there for different tests.  This way I never had to save or load anything, since I hadn't figured out (until today) how that worked.  I just had the slots permanently assigned 1,2,3 as tweeter, midrange, and bass, with 4 as bypass or special test--which I changed ad hoc.*

So then I unbypassed the lowpass filter, and set it to the same 71 Hz that apparently subwoofer DEQ is set to.  Apparently I set different frequencies for low and high pass because the settings don't match the actual "acoustic" crossover, which is what counts.  (I'm aiming for an acoustic crossover of 60 Hz.)

It's easy to bypass the FIR (it was done already) and unbypass sides of the crossovers.  But it's irksome that you can't bypass ALL of the PEQ's at once, they have to be bypassed one by one (but the default is they all have amplitude 0 so don't count anyway).  Anyway, some time ago I took the effort to bypass every single PEQ in the plug in.  I didn't check, except note the curve shown is flat.

It turns out I run the DEQ from the optical and the existing miniDSP from the coax output of the Sonos ZP80 which drives it all.  I wasn't daisy chaining them (I think I might have done that once, but changed to this for some reason)  So the new miniDSP will simply use the coax that the DEQ had been using.  I don't have to get some splitter or some such.

And I struggled to find one, but apparently there isn't any "mono summing" feature.  I struggled to find one, and also seemed to fail in some of the other plug-ins I have.  But it's no problem, my subs take two inputs and they can do the summing, I'll just have to have another 15 foot audio coax cable.  I think I can find some such, maybe of lesser quality for now.

But then I realized the rub.  I've been using the analog audio output of the Behringer for the subs because I only have one DAC for the outputs.  I bought two DACs, but I've been using the other DAC for another purpose.**  So I'll need to buy a third DAC.

They're the inexpensive Topping E30.  That should not be a problem, right???

Well Apos shows them out of stock.  I ordered one on Amazon, for delivery in the last week of June.

I wouldn't be totally surprised if that doesn't fall through too (but I hope it doesn't, I'd like to have 3 identical E30's for the bedroom).

Meanwhile I could rig up a Denon 5000 as DAC, and adjust the gain structure so it's noise would not be a problem.

But it's a big heavy monster and a pain to do such things just for temporary.  I'm going to wait until I get the E30, or I'm forced into some other solution by being denied another E30.  Then I could get another kind of DAC, such as a Schitt, and use that for my other purpose, then use two E30s for the two way speaker system (subs and monitors).

*Slots 2 and 3 are used by sub and panel crossovers in the living room.  Slot 1 isn't used in the living room anymore...because I use a 96kHz capable plug in for that (but less useful otherwise).  So the current 1 slot one in the OpenDRC 2x2 plug in was being used for the upper bedroom crossover, for the Revel M20 monitor speakers.  And now slot 4 for the sub in the bedroom.

**I needed to connect the TV audio to the bedroom Sonos ZP80, which can then switch it to the bedroom stereo or the stereo in any other room.  But if I connect the analog TV audio to the Sonos box, the hum will be terrible from a ground loop.  So instead I take the optical coax from the TV.  But Sonos doesn't permit digital audio input, only analog audio input.  So I use the second of my E30's to convert the TV sound from optical digital to analog for Sonos, which then distributes the digital to the two EQ processors via coax and optical.  By using a E30, I'm not only killing the ground loop, I'm using a much higher quality DAC than you could probably hope any TV to have.  And the 2004 era Sonos Analog to Digitial conversion is very nice and good, strangely it's always been easier to make even more perfect ADCs than DACs.  TV sound is the #1 use of the bedroom stereo and has been since my living room stereo took off big time, in 2008, when I got my first electostatic speakers, the Acoustat 1+1's, and the Krell FPB 300 to drive them,   (I did that because of a dental near death experience I had survived in January 2008, after which I decided to "eat dessert first," or actually long delayed by this point, by getting a really cool stereo like I had dreamed about to play with, what I had imagined when I bought this home 30 years ago...it was my #1 priority for home layout...it had to be good for having electrostats in the living room, and good stereo in every other major room too, which I now see may be less important than I figured, but always nice to have good TV sound too.)  Since then I don't bother to do serious listening in the bedroom, despite having a fine system which is derived only by improvement from my historic "best" system going back to childhood.  For background listening, I just play the living room system and hear it well enough anywhere in the house to be ok for background (which you don't want to be too immersive).  TV sound is not unimportant, and it needs good bass.  The bedroom TV is a 55" Samsung 850.  It may not be as good as the latest greatest, but I wasn't even dreaming about such nice TV's as a kid.

Thinking about EQ

 In the last few days I've been re-examining my upper midrange EQ.  So far, it seems as if something like I've dialed in for years, my extended version of a "Linkwitz-Gundry" dip, is needed, though I'm not exactly sure what form it should take, what process should be used to optimize it, and how it should be implemented.

But there are many ways of creating this dip.  The actual center frequencies could be anywhere within a wide range of possibilities, from 2kHz-12kHz.  There has to be more than one Parametric EQ (PEQ) or Graphic EQ slider in use.  It has to start falling around 2-3kHz and keep attenuating until at least 7kHz if not 12kHz or 16kHz.  The resulting response curve should be fairly smooth and more or less monotonically falling at higher frequencies.  Those are about the only things I think I know for sure.  There are infinite combinations of PEQ's that would meet that general sketch.  The smooth and monotonic response criteria are about the hardest to achieve.  I'm not sure if the monotonic part actually matters at high enough frequencies (above about 12kHz).  I've long used a supertweeter that peaks around 22kHz and thought that was dandy, however perhaps the supertweeter doesn't even make any difference at all (I haven't bothered to test in double blind experiments...I suspect it would not be easy to prove it as I believe nobody else has!...the widely known "proof" of the audibility of ultrasonics is currently from brain wave experiments...and even those results are uncertain if not speculative...but I do ultrasonics because I can and it always seems to make an important difference in sighted testing and it's nice to believe I'm ahead of the curve in this way).

But at audible frequencies, roughness and lack of monotonicity often seem to be especially audible, about as much as the actual response levels involved.  So I think it's reasonable to try to minimize those even lacking proof that they are, specifically, a problem.

I have to be very clear in saying that something like Real Time Analyzer response curves, like the many pictures included in the previous post, are not the truth in and of themselves.  This is true regardless of resolution, accuracy, repeatability, or other qualitative factors...but those factors might make the RTA even less true.

It's like taking a picture of a crowd.  Every picture is going to be different.  It could happen that even every pixel in every picture will be different, and yet it's the same crowd.  Meanwhile, the information in any one picture isn't very complete.  Some people, for example, might be turned away from the camera and you don't see their faces at all, yet their faces may appear in other pictures.  Either way you know everyone has a face, etc, even if you can't see it.

So it is with RTA's and in fact any other kind of measurement.  It's true of sighted and blind listening tests as well.  They are all incomplete.  Make that very incomplete.  There are deeper truths that no one measurement, no one listening test, etc, no matter how good, will capture.  And most measurements we make are very very incomplete.  As with a snapshot of a crowd.

We know, for example, that along with the frequency response variation shown in an RTA there is also time delay variation, sometimes called phase, or more correctly Group Delay, and this is NOT shown in the RTA.

You can get the Group Delay from some kind of comparison measurements, where you are able to compare the input and output signals.  The measurement could be made with simple impulses, "maximum" length sequences, or even gated noise, so long as an input to output comparison is being made, and typically this is done with Fast Fourier Transforms (FFT) or something similar.  I have several programs that do that and they show both the frequency and the phase response.

That's just one tiny way in which an RTA is incomplete, and it may not even be that important anyway, for one of several reasons:

1) It appears we are far less sensitive to variations in time delay than to frequency amplitude response. 

2) We don't even have any simple ways to describe it.  We could, hypothetically describe distortions in the depth or other aspects of imaging that might result from it (though in some cases it might not even be relevant, imaging depth being more dependent on inherent reverb applied to some voice...and reproducing that might be mostly a matter of low noise and distortion, not linear phase reproduction per se).

3) If we are to fix the frequency response, in nearly all cases we will also fix the group delay response.  This is because most physical phenomena are minimum phase, which means they can be inverted to an equal and opposite minimum phase process which perfectly cancels it.

I'm not entirely clear that all room response related variation is minimum phase, but at least one audio reviewer and Mathematician REG has said.  It does in fact seem you can model the room system with linear equations.

Non-minimum phase processes are made possible by things that are not linear...like discrete sampling.  Thus, we can have non-minimum phase processes in digital audio, essentially we can model or approximate anything.  We can make the phase and/or amplitude change as we wish and not only in minimum phase ways.

Now, imagine a series of non-interacting frequency response deviations a system might have.  You could imagine them as a list of unplanned/unwanted parametric EQ's (UEQ).  For each UEQ, we create an equal and opposite PEQ, and all these deviations are corrected.

But there are many issues here.  For one thing we don't really know the parameters of each UEQ.  We don't even know how many UEQ's there are!  Judging by the complexity of response curves, which get more complex the more well you can measure them, I'd guess the answer is way more than we can easily count.

So, probably even in the best of cases, were are only correcting to a simplification of the actual problem.  But imagine the difference between a system we could perfectly correct, and one in which we didn't quite perfectly correct, but approximate.

Imagine the true UEQ has a full octave bandwidth at 1kHz.  But what we have, instead, is a 1/3 octave graphic equalizer that gives us 800, 900, 1000, 1200, and 1500 Hz sliders.  We can push the sliders up to approximate the inverse of the full octave bandwidth UEQ, but it will never perfectly match (though Behringer had a trademarked method which was supposed to remove the response ripples from using sliders similar to this, something like "True Response").  Instead, compared with the the original, there will be tiny ripples in the amplitude responses at the transitions, and ripples in the phase response as well.

Should we go ahead with the equalization anyway?  I believe the answer would be yes, even if we can't (and in fact we never can) perfectly correct the input, our approximation helps, and it may in fact help in both the frequency response AND the group delay domains, restoring a better approximation of the original.

I'm a bit shaky, however, about the ripples.  If the ripples we make with our imperfect corrections approximation are too big, then the answer might be no.

So I can see the wiggles in the frequency response RTA, to at least 1/6 octave accuracy, to get some idea if my alterations are making things better or worse (but each measurement is fairly adhoc, I have no perfect positioning mechanism, there are various kinds of home noises included, random junk in the living room, etc etc).  But I can't see the changes in the ripples in the phase responses.  But see the points made above, I think it's a second order concern at best.  I need merely keep to keep the resulting frequency response fairly smooth, and most likely the group delay ripples won't be too big either.  And we especially have to avoid jaggedness or lack of monotonicity.

So this is justifying a kind of "correct to the measurements" approach I've long been abhorring, as compared to "correct to the model."  The problem is, with anything other than the bass response, I don't really have a very good model yet.  These are simple "response peaks" caused by room modes, though they may be caused or influenced by room modes.  They could also be affected by the geometries and materials used in the Acoustat speakers, including their compound transformer system.

Now a couple of days ago, I was saying how wonderful that I could boost two dips around 1kHz (at 1013Hz and 857 Hz specifically) for a better sounding midrange, and how that was better than using 1/3 octave sliders (actually, it can't very well be done with sliders...maybe...now I'm not sure I did it very well with two PEQ's because I keeps seeing that 900 Hz dip in some if not all RTA's, which just tells you how variable RTA's can be, sometimes 900 Hz is even higher than 1000 Hz, etc.  But I'm now thinking my magic didn't fix the problem that well).

I'm about ready to re-do the whole thing in graphic EQ and see how that works.  But I've come to appreciate the digital accuracy and control of the DEQ 2496 as compared with an analog Graphic EQ.  I need to liberate one DEQ from the bedroom, where it has long been slated to be replaced by a new miniDSP which has been in my inventory about a year without being set up yet.  I need that to do more convenient tests, I've been straining my body getting down on the floor to adjust the DEQ currently in line for the panels (and repeating every minute or two for readjusting and then back up again for the measuring).

A fairly reasonable strategy would be to do most correction using the 1/3 octave graphic, then fix remaining irregularities with the PEQ which can do very tiny corrections down to 1/10 octave.  Though I'm still tempted to tame the big 4.5 kHz rise with a one or more octave PEQ, just for starters.

Now about the Linkwitz-Gudry dip I remain uncertain.  But also it seems to me that a planar radiator is radiating A LOT more highs into the room than a traditional point source tweeter, which has very reduced dispersion at high frequencies, aimed right at the listener for best effect.  As a result, more of the highs from are indeed reflected from the room into the side if the head and straight down the ear canal, much more than in a real venue or room with point source tweeter.  So I think it's very reasonable to believe planar speakers require a falling high frequency response.  Possibly point source tweeters do also.

I need to do sweeps on my other systems to get a clearer idea too.

*** Update

I forgot to make a key point that's been in my mind.  There is, it turns out, only one way to determine the frequency components that make up an unwanted resonance, and that is by actually cancelling it out.  The proof is in the pudding, so to speak.  Measurements alone don't get deep enough into the mix, or so it has always seemed to me, besides the measurements themselves can vary quite a bit depending on technological choices and random factors.

FFT measurements are limited by numerous factors, including the size of windows, the windowing function, etc, etc.  Then, what we get in RTA compounds this construction further with averaging.

I've never seen anything like FFT magically pulling out node frequencies that need to be cancelled.  FFT analysis systems seems particularly weak in the bass, especially if they are trying to do a full spectrum analysis, the bass gets relatively few points to begin with.

Meanwhile, I've seen very little discussion of how program like REW compute required equalizer settings, or even how well it does that.  People seem to think it just works, but in my experience, my time honored methods, including slowly sweeping with an oscillator, seem better.  (Slow sweeping with an oscillator is very revealing in the bass, anyway, above the bass it often seems useless, unless there is some little thing physically rattling.  Above the bass, RTAs and FFTs work better than oscillator sweeping, it seems.)

It has long been known by experts that simple pulse-type transients (such as the on/off "dirac") don't really give enough information to see much.  An FFT of a single pulse looks very revealing, but there may be a lot buried under its noise floor.  So for decades now different stimuli have been used, including "maximum length sequences" (which may have been maximum length decades ago) which generally are some kind of fast sweep, and even canned truncated bits of pink noise.

Real music may contain long bass tones, long enough to do many reflections and build up a nodal response, whereas diracs do not.  So using an oscillator you can show 40dB peaks and nulls where an FFT spectrum of the bass looks like a nice roll of flubber.

The higher frequency resonances seem to be something like nodal

Finally getting around to the Midrange EQ

I recall now that measurements over the past 12 years (since I've had a serious living room audio system featuring Acoustat electrostatic speakers, for the the 1+1's and now since 2019 the 2+2's) have fairly consistently showed a dip (or the start of the general High Frequency rolloff) right around 1kHz.  I had often thought to myself "that's not good" but also, almost always, "I'll do something about that later when I have time for serious listening.  Midrange EQ requires serious listening, not mere measurements.  Just tossing in some EQ by measurement is as likely to make things bad as good.  Especially starting from a well regarded speaker like the Acoustats, which pretty well determines the frequency response at 1kHz since room nodes are not important there."

Starting Response, Uncorrelated

Starting Response, Correlated

Notice that of all places, there is a slight depression right at 1kHz, where we are the most sensitive!

 Now, in principle I could crank up the tweeter control on the Acoustat...but then the part above 1kHz would get raised even more, where it begins to sound edgy.  So you have to pick a spot like this in the HF attenuator, a compromise.  (I haven't actually technically explored the HF attenuator control much, I'm afraid of being able to re-create the current setting without an objective test, which I haven't yet verified.  It was chosen with great care in listening as exactly what I describe, a compromise.  Which happens to be, for the 2+2's, right at the center of the control range, as you might expect it to be.)

Most of my parametric EQualization's (PEQ's) are simply notches for notching out room modes.  The technique is not unlike what a computer would do.  I sweep up and down with an oscillator to find the worst node resonance, try to cancel it out exactly with a PEQ notch by setting the Center Frequency and Q, and move on to find the next remaining worst node resonance, until everything is generally smoothed out, and/or I've run out of my 10 available PEQ's.

I have generally avoided EQ in the 125-17000 Hz "midrange" signal sent to my Acoustats.  It has a few resonances (one around 515 Hz for example) that I've sometimes notched out, but then later undone the notches because I felt that perhaps it was obscuring the midrange, which always has sounded slightly distant and opaque in the first place, and I worried (but didn't actually prove) that the notches might be making that worse (perhaps they weren't I'm thinking now).  But I never felt strongly enough about that to actually do the serious listening to adjust the midrange EQ around 1kHz by ear.  The "raw" Acoustats are good enough, I sometimes opined.  I don't want to mess with something I'm more likely to make worse than better, and where it's important.

Another issue is that in the conventional wisdom, you should use EQ to make cuts only.  You should not try to use EQ to boost depressions, and especially not notches (where EQ generally doesn't help at all and such high levels of EQ cause other problems).  Well, actually, lots of people use a little EQ boost here or there, and sometimes a lot of EQ boost.  But I wasn't willing to try without some kind of serious listening experiment, which I was never much inspired to do, basically because things generally sound so very good (especially compared to all the various systems I've had before AND nearly everything I hear elsewhere with very few exceptions.)

Years ago I purchased a separate EQ unit to playing with midrange equalization from the listening position, a Monoprice 615031, a 31 band analog Graphic EQ featuring both balanced and unbalanced inputs and excellently low noise and distortion.  Previously and about up to then I had a odd collection analog graphic and parametric equalizers for EQ experiments I intended to do but never got around to.  Back in the 1980's I had a Technics 31 band (I actually played with that briefly, but found I couldn't change a certain characteristing "dryness" which had been on my mind then, so I prematurely sold the Technics then over the years acquired a bunch of funky not-quite-replacements, several ADC's, and an previously overused Soundcraftsmen which was always noisy and useless but I hoped to someday refurb.  I think I still might have had an ADC when I decided to get the Monoprice to replace it.

The Monoprice 615031 is simply in a higher shelf of performance than all the others I've mentioned.  Integrated circuits nowadays are nearly as good as the best possible discrete circuits such as might be in instrumentation or preamps designed by John Curl.  Integrated circuits have gotten so good since around 2004 they blow away discrete transistor circuits in almost all other gear, especially older gear, nowadays.  Especially in stuff made by Emotiva, for example, which tends to use the LM4562, one of the best analog chips (other goodies are OPA211 and AD797, with the latter being somewhat hard to stabilize).  And of course the likes of Mark Levinson, who fills their gear with the latest and greatest, though it's funny how even with all the incredibly cool parts and circuit board materials and all their specs don't even match Emotiva until you get to the very top and stratospherically priced units.  I don't trust Monoprice as much as Emotiva (especially after one of my Monoprice HDMI extender units failed in 2 years, though that has not been unusual for most makers of HDMI extenders, don't get me started on that.)

But this fine graphic EQ simply sat in storage until I had my first attempt at listening to Purple Rain by Prince, an album I bought a few years ago but had not played yet.  I was looking at my record collection trying to decide on what to play, and Purple Rain was an obvious choice simply because I had it but had not yet bothered to listen to it (many new albums sit in the collection for years before being played the first time...the album buying process has often seemed almost entirely disconnected from the album playing process).

Immediately the midrange equalization came back to mind because, after having removed my upper midrange Gundry-Linkwitz inspired depression a couple weeks previously without much apparent difference, I now needed to dial it back in just to make this album listenable.  Without that EQ depression, which was made up of 3 different PEQ's actually (because I added one on top of the earlier one(s) to make more extended and flatter in the periphery) the album was so harsh sounding it was pure torture to listen to. 

So I dialed back in the Gundry-Linkwitz inspired depression (I have been lowering the 2-10khz range) and Purple Rain was far more tolerable, but it was still very obviously sounding thin and edgy.  I thought to myself, "Prince mixed this on high end equipment, he would have mixed it just right to NOT sound thin and edgy."  This is the "test case" I need, I said to myself, to see if I can set the midrange EQ to make it sound better.

This album is a great test case precisely because it is NOT a "audio purist" kind of record, say recorded in real time to a high resolution or DSD recorder using a single stereo ribbon microphone, etc.  Something like that will sound "good" on nearly any system.  Instead Purple Rain is a very complex mixture of recordings combined on very low noise and distortion equipment mixed and eq'd to a very precise endpoint.  If that endpoint is high end enough to be revealing, but not EQ'd just right, it will sound terrible.

As I often argue, inverting much audiophile conventional wisdom, you need the very best fidelity for reproducing the lowest fidelity.  The lowest fidelity is the most challenging to reproduce to sound even good, unless reproduced by such comparably low fidelity it doesn't even matter.

In this case, I'm not saying the Purple Rain is low fidelity, it's just ridiculously overdubbed and complex, and that makes it very revealing of tiny issues in reproduction equipment of high resolution generally.

(BTW I started from the LP I bought years ago, and migrated to the Qobuz high resolution version I like somewhat better.  I need to do more LP adjustments too.)

So, on Saturday May 14 I hooked up the Monoprice graphic equalizer.  Actually by that time I was thinking other ideas I'd long had in mind.  Such as that perhaps a deliberate peak at 1kHz might add more "acoustic" sound to the reproduction, curiously the LS3/5A's used woofers with a huge 1kHz peak (which is also common in other drivers) but they did equalize it out mostly.

So I started by dialing in a 1kHz peak.  It sounded horrible at +15dB, but became unnoticeable around +3dB.  Listening at +2dB it clearly sounded better...less electronic and edgy.  It's funny how raising the midrange seems to clear up the high end more than the midrange itself.  And improves the bass too.

I then took a look at measurements (which at that time I'd basically forgotten there was a 1kHz depression).  And strangely enough, my dialed in adjustment to make it sound better coincided with flatter response, essentially filling in the depression.

So I tried more EQ's to fill it in even better.  But changes in pink noise response that initially looked better in uncorrelated pink noise than uncorrelated, and didn't necessarily sound better.

After a late night of fidding, I ended up with a pretty good sounding adjustment that was fairly small and simple and didn't make things worse in either correlated or uncorrelated pink noise.

+2dB at 1kHz

+1dB in the 1.25 and 1.6kHz bands

I dialed an approximation of that to the Behringer PEQ for the Acoustats (which does the Gundry-Linkwitz inspired dip) and it sounded just as good done that way as done in the separate Monoprice EQ, and when I added the Monoprice EQ on top of it (doing the EQ twice over, making it +4dB at 1kHz) it sounded worse.

Note here that I have no idea why these adjustments are needed.  I have no model, so I'm not following my "Adjust to the model, not the measurements," rule.  But adjustments like those can be made if listening does, and continues, to justify them.

Now I need to do a serious listening test and reconstruction of my Gundry-Linkwitz inspired dip, which was mainly configured ad hoc in response to pink noise measurements over time.

Present results suggest that perhaps you should use small boosts, if necessary for flatter response, at least below 2kHz.  Above 2kHz some sort of depression or roll-off seems desirable in the response measured omnidirectionally.

Final Response, Uncorrelated

Final Response, Correlated

Update: Re-doing the EQ on Sunday May 15

Somehow the new midrange was leaving me unsatisfied listening to KPAC on FM radio the next morning.  Rather than smoother midrange, I was feeling that there was a lumpier midrange with extra degrees of freedom.  It also bugged me that the final "simplified" graphic eq adjustment had created two peaks, one at 1kHz and one 1400 kHz.  Meanwhile there was a +1dB adjustment at 1400khz.  That was probably wrong.

Though possibly convenient sometimes, I just don't like graphic equalization.  It seems as though you are adding and subtracting resonance at frequencies that are arbitrary compared to the real phenomena you are trying to correct.  Plus rather than simply adding or subtracting resonance in the width required, all the resonances are exactly the same, 1/3 octave at present.  That just doesn't seem right to me.

I like parametric EQ, where I can set the frequency of the resonance or anti-resonance precisely and set the Q or bandwidth as well.

But first I thought I'd use my Kron-Hite oscillator to sweep the area to see what the real phenomenon I am trying to correct look like.  Nothing can give you a better feel, I have believed, than slowly sweeping with an oscillator.  I used no EQ for these tests.*

But strangely, right here at the center of the midrange, it wasn't much use.  What I was hearing sweeping from 100-1000 Hz and 1000-10000 Hz (sadly the Kron-Hite doesn't provide any easy way of sweeping around 1000 Hz because of its choice of frequency ranges, and in the regards, General Radio oscillators are often nicer with their 200-2000 Hz range, though what would be best of all would be some way of controlling the ranges themselves) was one tight little resonance after another, just going around and around and around from peak to trough every 3% of frequency change or so, but rarely changing in amplitude very much.  Ignoring the tiny peaks and troughs, which are probably caused by room reflections, the general response was very very smooth, it was hard to hear any changes at all.  Or in other words, the tiny peaks and troughs seemed to be nearly all the interesting stuff that was going on.  But I did seem to hear that a particular resonance just above 1kHz, 1015 Hz to be exact (measured with my Fluke 8060) was slightly less resonant than the resonances around it.

So I decided to start there (actually 1013 Hz was the closest available frequency I could select) with a Parametric EQ having bandwidth of 1/3 octave and +2dB boost.  It seemed to do a good job fixing the notch around 1kHz without creating a new peak above it.  But this seemed to highlight another trough below 1kHz, somewhere around 850 Hz.

I possibly didn't try hard enough to eliminate that lower trough with a wider bandwidth for the 1013 Hz PEQ but I still don't think it would have worked right.  Instead I added a second PEQ, this time starting around 830 Hz and moving it up until the response looked right.  I settled on 877 Hz, with bandwidth of 1/6 octave and height of 1dB.  Ultimately I widened the bandwidth of this PEQ to 1/3 octave as well.

Correlated PN, 1013Hz at +2dB

This looked far better I thought than the Graphic EQ adjustment.  And replaying Purple Rain again, I thought it sounded way better, though I was goaded to raise the PEQ at 1013 Hz a bit more to +3dB, and that made this album solid as well as magical (and far away from the unequalized sound which had been harsh, thin, and barely tolerable).  With Graphic EQ I had never been able to get both 1Khz and the band below it flattened without causing the periphery to rise up like mountains.  The problem is that the required correction frequencies do not like up with what's available on the EQ (and perhaps the bandwidth as well, but this time I ended up using 1/3 octave bandwidths).  (Note the periphery does still rise up a bit, but no more than it does without any EQ.)

* Method Notes

I do things in ways that would probably disturb most audiophiles.  But with purpose...generally attempting to do correct level matched comparisons (though, technically I skipped that a bit during these tests...but it would have no effect on the measured responses).

When I used the Monoprice Graphic EQ for rough adjustment test purposes, I had the Monoprice directly feeding my rarely used "A" amplifier, the Aragon, and I was using my ABX Switch to select between this equalized signal and the unequalized version fed through my Hafler 9300.  Both amps were getting their signal from the same Emotiva Stealth DC-1 dac, but I was feeding its unbalanced output directly to the Hafler, and the balanced output to the Monoprice and from that to the Aragon.

The level of the Monoprice/Aragon signal was then matched to the Hafler by adjusting the EQ level controls of the Monoprice (which infuriatingly are only engaged when the EQ is active) and switching back and forth until they sounded the same not only in level but in quality.  I did this precise level matching before adjusting any of the equalizer band controls--they were all set to flat.

Audiophiles might look at the fact that I was using two different amplifiers, not to mention all the different cables, and the equalizer circuitry, etc.  But the truth is that good amplifiers, when level matched, sound exactly the same.  And quality line circuitry as well as decent cables are perfectly neutral as well.  This has been proven over and over again by endless tests by objectivist audiophiles, a long running $1000 contest, etc.  And by my own tests with Hafler, Krell, and Aragon amplifiers--when working properly and level matched they all sound the same!  Subjectivist audiophiles don't believe it from their own experience because they almost never do proper level matching, nor do they ever do instantaneous A/B switching made possible with an ABX Switch and so have to rely on their highly inventive audio "memories" (which are really more like stories you tell yourself).

This time I didn't even bother to do voltage measurements, I simply level matched the different amplifiers by ear.  After the first pass, the level sounded the same, but the one amplifier or the other sounded different in some way, such as MORE highs or MORE bass, etc.  When one side has qualitatively more of any one thing or another, its level is turned down very slightly or the other side is turned up.  Eventually you get to the point where indeed the amplifiers sound absolutely identical.  Then you can add in a dollop of EQ, and the difference that makes is fairly obvious.

Now the place were I did not do level matching is this.  When I turned up the EQ by +2dB at 1013 Hz, for example, I did not attempt to level match THAT overall to the unequalized condition.  To Be Perfectly Fair, in evaluating which sounded better, I'd have to do that, say using C weighted full spectrum pink noise and trying to match the levels by ear or C weighted SPL measurement.  One problem here is that since the frequency contours are different, there never might never be a place where they would sound qualitatively the same, I could at best only match the "rough" level, which would probably not be much different than no matching at all (since the levels are the same everywhere except the 1/3 octave around 1013 Hz).

Since I know what my alterations are doing, I think this is still pretty fair.  Where this sort of open ended unmatched testing would be a problem is when you don't know what physical effect your change is making, and you are making lots and lots of changes like that.  The tendency will be to prefer changes that slightly increase the level, so after many such changes the resulting levels will be significantly higher than before at the same volume level setting, but then you might also be setting the volume control down over time as well to compensate, with the possibility that all your changes have done nothing, or nothing but change the level.

I also hooked in the oscillator through the Aragon for convenience.  I put the oscillator on top of the Monoprice EQ, and hooked the cable going to the Aragon to the Oscillator output.  That was a fine way to do sweeping (of only one channel btw) but then I could not at all tell the effects of different PEQ's in the Behringer DEQ 2496 (which feeds the Emotiva DAC) on the sweeping signal.  I had to evaluate PEQ changes by measuring pink noise, and by ear.

Update May 16, taking a look at that "Linkwitz Gundry Dip"

Now, what has also been on my mind for a long long time.  I've fallen into the habit of making a depression in the 2-12kHz region because, well, various reasons.

Linkwitz and Gundry recommended depressions above 2kHz

Linkwitz rationalized that domestic rooms have a lot more high frequency reflectivity than large auditoriums.  High frequencies can enter the ear not just from the front but from the side where there is a more direct path to the eardrum.  Resulting in a sensation of excess highs.

I seem to recall these dips being recommended as 2-4kHz or perhaps 2-6kHz but not roughly 2-12kHz as I am doing.

But I was also faced with some unevenness in the highs that was hard to iron out, most infuriatingly right at 6kHz and even 12kHz which are characteristic "metallic resonance" frequencies.

Anyway from rather ad hoc measurements and tests I ended up hacking away at the 2-12kHz response rather significantly.  Perhaps too much.

Some of that hacking may have been inspired by the relative lack of midrange around 1000 Hz, which I have just now fixed.  So the EQ in the 2-12kHz region now needs to be entirely reconsidered, perhaps even from scratch.

But first I needed to re-measure the system response since I had not done so after adding an extra dB to the 1013 Hz boost, making it 3dB, which really seemed to bring things alive (without being too much).

Well, now I believed I should up the 873 Hz to 3dB as well, as now a trough appeared at "900 Hz."  That didn't seem to erase the trough completely.  I believe I need to actually find the null point I am trying to correct, since I see now that the 1/6 octave depression was at 900 Hz and not 800 Hz as I had believed.

I did some sweeping and found the actual low null nearer to 853 so moved the boost point there.

Meanwhile I did zero out all three of of the high frequency EQ cuts, and I am casually listening.  It does sound a bit on the bright side (despite almost appearing flat 20-12000 Hz).

I tried listening to Supertramp with only the lower frequency cut out, then 2 cuts outs, then all 3 more or less as before.  With each additional EQ cut it sounded better, high frequency fuzz was being removed, revealing voices and percussion better.

Sweeping revealed that the bottom cut at 2729 was already as close as available to a peak at 2736 Hz.  The other two cuts were a bit off from measured peaks, so I moved them closest to the exact peaks similarly.  The 1/2 octave wide and full octave width for the top cut seem psychoacoustically justified.  Frequencies from 6kHz on up sound especially irritating, up to about 12kHz where my sensitivity is beginning to fall.

I'm thinking the Linkwitz-Gundry principle applies all the way up to 12kHz or so.

At first brush, all the upper EQ's seem justified.

For years I dismissed HP's snide comment about the Acoustat 2+2's.  He coined the term "credit card coloration" in reviewing them.  Well, indeed, there seems to be modest peaks at 2736 Hz and a few other high frequencies.  It appears to really enjoy them you need parametric EQ.

(Or at least for me in my listening room.)

Uncorrelated, Full EQ

Correlated, Full EQ

Update May 17

Sometime after making the last measurement, possibly while listening fairly seriously to the Beatles White Album, I realized there was no deep bass on the right side.  In general, deep bass seemed pretty anemic (though it's not terribly in abundance either, in The White Album).  In due course I figured out that the right subwoofer had gotten unplugged.  The IEC pulled out a bit, and I hadn't noticed.  I decided to order a new audiophile cable featuring connectors in solid plastic (though it doesn't appear to be name brand solid plastic, like Marinco for example).  We'll see about those later.

Meanwhile on Tuesday afternoon I knew I first needed to repleat the full spectrum graph that ended yesterday's post, but with both subwoofers operating as they should.  That yielded this:

Both Subs Working

So there's a pretty healthy deep bass, at least between 80 and 31.5 Hz, still nevertheless fitting into my concept of "electrostatic" bass (it's implemented with big subwoofers but thanks to careful nulling of room nodes, it is supposed to sound as if it were very good electrostatic bass).  I ought to be able to fix the weakness about 70 or so since there's lots of EQ there, but not up higher than 125 since that's where the panels take over and I'd have to use boost (again).  Perhaps I shouldn't be so scared of boost now that I'm using right in the center of the midrange.  Below 31 Hz it's damned hard to fix, despite the woofers working very hard, there's massive cancellation in the front of my room where (putting image width and therefore instrumental separation at the top of the reproduction goals) I have my listening seat.  Well it mostly works for me.  Headbangers can sit in the back of the room where they get super exaggerated bass, deep bass included:

To make the base at the serious listening position higher, I'd most likely have to make the bass in the back of the room higher still, and it's already at the wall rattling point.  Except with bass absorption boxes, that would likely have to be about 1/3 the size of the room to have significant effect.  Or a line of active absorbers which cost $1600 apiece.  Or multiple new bass woofers hooked up in swarm.  Or a 2 dimensional FIR approximation of bass moved to the listening position, which might be super-sensitive to where you are listening.

So anyway, that's for the future, I'm still investigating the midrange and upper midrange today.

So let's look at the response with and without the "Gundry-Linkwitz" inspired upper midrange EQ, on and off (I'm going to repeat the graph at the top of this update simply to have both graphs together).

With Upper Midrange EQ

Without Upper Midrange EQ

There are lots of curious things about this.  Starting from the fact that the unequalized version has an apparent peak around 4500 Hz, and that peak is vaporized by the EQ, but the EQ doesn't have a cut at 4500, it has cuts at 2700 and 5300.

The unequalized version looks like it ought to be OK, the rounded peak around 4500 being mostly dwarfed by a gradual downturn that looks like it ought to sound fairly soft.  But in fact it sounds bright and harsh as hell, and not even primarily because of the apparent peak at 4500.

Speaking of which, comparing the EQ'd version with the starting point of this now very extended post, it's not clear my slight readjustments of the EQ nulls to local peaks measured with sweeping worked out for the best.  It appears my earlier adjustments (which were made and remade over many years, primarily to make the RTA look good) were more effective at creating a smoother HF rolloff.  I will have to investigate that later.  Meanwhile, I tried nulling out the 4500 bulge with a corresponding octave wide cut at 4500, yielding this:

This looks like it might sound the best of all, it's a mostly equal and gradual slope from above 2k to 20k, but it's barely changed from the unequalized version in sounding bright and harsh.  At least I guessed this because the sound of the pink noise had barely changed from the No Upper Midrange EQ (NUMEQ) condition.  A little bit better than NUMEQ, but it's clear it still needs more EQ.  I've labeled NUMEQ as "Boost" in my DEQ memories, because it still has the +3dB boosts at 858 and 1013 Hz, both 1/3 octave wide, whereas I've labeled the currently fully EQ'd version "BG" for "Boost plus Gundry Dip EQ."

Perhaps it needs at least one more EQ point to deal with the little bulge roughly 6-7kHz, which is right at the center of the "metallic" sound.

So I added an additional notch just below 7kHz, yielding this (I'm not 100% certain this is right photo but it seems like it has to be)

Well this didn't work out as expected.  Instead of cleanly notching out 7kHz, it's created a new peak above 8kHz and a new dip at 5kHz.  It has a similar rolly polly look as "Boost."  And, curiously, it sounds about equally well balanced as Boost...and maybe better, as a very slight upper nasality has been removed, but it also has too much "brilliance" around 10kHz.  I'm calling this "P2" because it uses two parametric EQ's in the upper midrange.

And I'm calling this next one P3 because it uses 3 parametric EQ's.  This time I went back to correct the resonance right in the middle of the usual Gundry Dip area, thinking that to be more important than the excess brilliance around 10khz.

Now I have a response that looks very similar to BG, but it's constructed entirely differently!  Rather than cuts at roughly 2700, 5300, and 9900 it has cuts at 2700, 4500, and 7000.

Sonically I think P2 sounds like it has more potential than BG, but it's not actually better just different.  And P3 is beginning to sound worse, maybe.

This is all very confusing, but there are several lessons that we know and others we can guess:

1) There is more than one combination of PEQ's that yield roughly the same RTA response.

2) When you whack down one peak, others you may not have been aware of appear.  Also, new dips may appear.

3) We might think that the fewer PEQ's the better.  But this may not be the case.  What we really want are PEQ's that somehow precisely invert "resonances" or combinations of "resonances."  This might mean more PEQ's than at first appear necessary.  But it's hard or impossible to tell where the true resonances are.  With a sine wave sweep, it constantly goes up and down just a tiny bit more here or there.  Without looking at the RTA, I would have never guessed there was a NUMEQ apparent "resonance" at 4500 Hz.  That apparent resonance might actually be the combination of other resonances.

4) Sonically it appears best when the high end above 2kHz is gradually falling.  Rises above this gradually falling floor are undesirable.