Interview w/famous speaker designer-covers common discussions

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The Lipshitz and Vanderkooy paper directly listened to a variety of speaker power responses Including flat power from rising on axis response (scorchingly bright), flat power response with flat axial response (still overly bright) and more normal power responses with flat axial. They found the later were fine and that holes in the power response were fairly benign. Peaks in the power response were undesirable.

Dave, curious, were their listening tests executed in their lab space in Waterloo? Its a concrete bunker very different from most listening spaces, and so I think would greatly inflate the influence of power response, vs a typical domestic living room.

Dave
 
Rather useless? I don't know why an experiment where you directly manipulate the variable of interest would end up be "rather useless" in either mono or stereo.

Well, if you're trying to simply to get a preference rating between these systems, the experiment is fine of course. My comment without further explanation was a little blunt, apologies for that.

However, I like to ask questions :D

However, it is rather easily ignored that there are multiple effects in stereo reproduction which affect the perceived balance. The stereo crosstalk dip together with increased treble due to the 30 degree HRTFs for a 0 degree phantom source dramatically change the perceived balance. Although the crosstalk dip generally leads to dullness, it could well that spectral smoothing of the dip is stronger for speakers with flat power response. In that case, the phantom sources (which are the most important sources, often the main performers!) in the recording might appear too bright due to the HRTF difference between 0 and 30 degrees.

If both effects would occur, then one could hypothesize that not the power response is at fault but the on-axis response. The on-axis response (which is perceptually dominant at higher frequencies) might need to be attenuated towards high frequencies to correct the brightness of phantom sources.

In speakers with increasing directivity, the spectral smoothing of the crosstalk dip might be smoothed only to a lesser degree. Together with the overall reduction in power this might compensate for the HRTF problem of phantom sources.

Of course, it could also be that increasing directivity is indeed the most desirable. I'm just hypothesizing, but I think these questions are relevant. The results of a mono test could very well be different from a stereo test, because the crosstalk dip and HRTF error are eliminated. That's why I was curious about the test method :)
 
Last edited:
Dave, curious, were their listening tests executed in their lab space in Waterloo? Its a concrete bunker very different from most listening spaces, and so I think would greatly inflate the influence of power response, vs a typical domestic living room.

Dave

I'll have to re-read the paper to see if they describe the rooms in detail. They did duplicate the first course of tests in a smaller room and noticed that power effects were somewhat more noticed, presumably due to earlier arrival of initial reflections.

A very worthwhile paper as it directly examines something we constantly debate.

David S.
 
In that case, the phantom sources (which are the most important sources, often the main performers!) in the recording might appear too bright due to the HRTF difference between 0 and 30 degrees.

The reason this argument still does not make any sense is you are listening to recordings manipulated by people listening in the same fashion. You are not hearing a recording straight from a 'perfect' microphone--whatever that is. If the mix engineer stinks, it has nothing to do with HRTF differences between stereo and mono. The better my stereo has become, the better these mixing engineers seem to be.

Dan
 
The reason this argument still does not make any sense is you are listening to recordings manipulated by people listening in the same fashion. You are not hearing a recording straight from a 'perfect' microphone--whatever that is. If the mix engineer stinks, it has nothing to do with HRTF differences between stereo and mono. The better my stereo has become, the better these mixing engineers seem to be.

Dan

Okay, the recording is a factor; it is with every speaker, so that doesn't change my argument. Mixing engineers in a LEDE setting hear a much stronger crosstalk dip than on a typical system in a typical home, so the balance will be different.

If the mix has not been equalized, then there WILL BE a HRTF colouration which will make voices and other instruments in the center sound unnatural to varying degrees. Yes, the mix engineer will have heard it (albeit with a different amount of spectral smoothing), but the colouration is there.

This is of course a non-issue if your opinion is that it's good enough to hear more or less what the mix engineer heard. My opinion is that you should investigate this if you want it to sound as natural as possible with stereo and not just "as mixed".
 
I'll have to re-read the paper to see if they describe the rooms in detail. They did duplicate the first course of tests in a smaller room and noticed that power effects were somewhat more noticed, presumably due to earlier arrival of initial reflections.

A very worthwhile paper as it directly examines something we constantly debate.

David S.

Thanks, please do. I'm surprised I missed this paper.

I was there just a bit later ('88) and they tended to listen in the TA office space, which was a big ol conrete room on campus.
 
Now that I'm thinking about it, one could argue that varying the power response in a stereo setup perceptually changes two variables. The first would of course be the spectrum of room reflections, the second would be the amount of spectral smoothing to perceptually fill the dip in the direct sound. The ear/brain would compensate in varying degrees for this "constant" flaw in the direct sound.
 
Don't forget that the constant flaw changes--starting to sound like DeCartes I know. The mix is typically checked in numerous places, EQed several ways and times(not to mention the million other ways it is manipulated-you'd be shocked), by no means is LEDE de facto standard(not even popular from what I've seen), and finalized by a mastering engineer. The HRTF argument just makes no actual sense. I haven't seen it defended well yet. You'd think it would be easy to defend well if it were correct. It just seems so dead before it gets started. If it were a mono recording, not mixed/mastered, made with a perfect mic(?), I could see the problem. It could then make sense and we should listen in mono. There would be no need or desire for stereo. It would be counterproductive.

I've never heard any 2 systems with the same tonal balance, spaciousness, and/or image. It's just not going to happen unless they are literally the same. One of the tricks often done in mixing is to bring some of the recordings you are familiar with, listen to them, get a sense for the balance that system has and do you mix. Then go through a serious of double and triple checks in many different ways and locations. If you are extremely familiar and comfortable with the system you are mixing on, you can cut the time(and times) of doing this by degree. The recording process if far more complicated then the HiFi/audiophile(not an insult) community seems to realize. I own several 500 page books on the subject. The lack of standards are always going to plague the enthusiast though they may not be aware, but if you build a system by the current evidence I'm sure you can pretty much set it, and forget it. I have. Dr. Toole's book really is most of what you need to make very well educated decisions. I'm sure are other helpful ones. The HRTF guys seem to have a desire to remove the recording process from their thoughts. That seems to be the fundamental flaw I see in their thinking.

We do not live in an 'ideal' world.

Dan
 
"We do not live in an 'ideal' world"

Jeez, Louise, you're probably right! You are suggesting a more holistic approach? There's some sense in that. Nevertheless one-variable scientific research does have advantages over the holistic approach. Therefore I agree with a_tewinkel in that if your aim is to correlate power-response with timbre, you should (at least initially) do it in mono. Stereo has its own defects that affect timbre, like the stereo cross-talk dip (see Toole p. 151).

However, I haven't read the paper concerned, so I'll hold my judgement, at least for now.
 
I want to do some binaural measurements to see what the 'dip' looks like with a head in there. Has anyone seen such a measurement?

Dan

It's in section 9.1.3 in Floyd Toole's "Sound Reproduction". His reference is this article, at least for the last of the three measurements, which was done in a reflective room.

The dip is 10 dB at 1.9 kHz in an anechoic setting, compared to a centre loudspeaker. Also, the frequency response rises above that with about 3 dB compared to a centre speaker.

In the reflective room, the difference is about 4 dB at 1.9 kHz and level rises to +3 dB for higher frequencies, with a mild dip at 6 kHz (0 dB).

In my view those are numbers not to be ignored.
 
Last edited:
Cool! Never searched for the paper, but I remember the part in the book. Don't recall him saying they did dummy head measurements though. I must be getting old. I thought he said he wished someone had... Looks like they got it covered. It's not surprising that 1 source for a mono signal would be more intelligible and a good reason to advocate for a center channel. More so with the stereo background of a similar nature when they keep it in stereo either way. Mixers use the panning to improve intelligibility all the time. The biggest reason to pan is to make room for everything and keep it intelligible--not just adding spaciousness. Intelligibility is typically job one--effects are further down the hierarchy. Of course that's not always the case. Just scope out some death metal recordings. Unintelligible in any playback format. Personally I love my center channel, but don't use it for mixing stereo-sort of defeating the purpose. I'll use it to check mono compatibility, but I don't even do that anymore. No one uses mono for serious listening now(I think) even though I did just a few years ago. I'm an oddball and the only person I knew who did at the time or now. How useful that test is for a mix designed to be intelligible in stereo is certainly questionable. Anyone decent at mixing makes sure the vocal is absolutely intelligible. The vocal is really the song!! This assumes a reasonable quality system of course--nothing fancy, but some Bose may not be good enough. That's the cardinal rule unless some other effect is desired or the vocal stinks. Of course, a big unintended suck out cannot be a good thing and it has to be detrimental to some degree, but I have never once had an intelligibility issue with any decent mix. My old delta blues recordings have intelligibility issues regardless of mono or stereo playback. Nothing to do with the mix, the dip, or stereo/monophonics, but the recording quality and singer intelligibility often leave something to be desired. That's also part of the charm! Those things are pushing 100 years old now. The tech really wasn't all that then regardless of what they old timers say. They can never go back. If they could, I'd bet they'd come back to the future ASAP as far as sound quality is concerned.

If something is mixed in stereo and playback is with a center channel of some processing or summing--then you can have an HRTF issue if it isn't corrected. It won't likely make it more 'correct' or more like the original performance. It will make it tonally different and possibly (perhaps even probably I don't know) more intelligible.

If any of you are having any real intelligibility issues, it's not likely to be attributed to stereo reproduction. I can't recall ever having intelligibility issues related to stereophonic sound. I mean not once--not even close. If anyone knows a stereo recording that only sounds intelligible with a center channel, please let me know. I want to check it out and see if there can be a stereophonic problem. I bet it's not.

Dan
 
Last edited:
I've never experienced a recording that was not intelligible without a centre channel. I have experienced that intelligibility can be rather poor on speakers with relatively strongly increasing DI, especially with medium to large fullrange units, compared to well-behaved speakers.

Keyser and I recently did a comparison of pink noise and voice in mono and stereo on his speakers. One of us would be listening, while the other would quickly switch and adjust the volume setting.

His speakers are highly directional and we had added absorber panels of 4" fibreglass to eliminate the remaining reflections < 20 ms. Especially on "dry" tracks I had the idea that the crosstalk dip was very audible in this setting (by the way, in an anechoic setting the dip is quite wide). Hence the experiment.

On pink noise the difference was quite noticeable. The voice track we used contained a lot of reverb, which seemed to reduce the dullness although there still was a noticable tonal difference.

By the way, it is very easy to precisely locate the sweet spot this way; it's the place where the sound is dullest. Toole also recommends this experiment.

However, my own room is much more reverberant and my speakers are less directional (dipole). In this setting the dip is less audible. If the dip is sufficiently smoothed by reflections, the approx. 3 dB rise at high frequencies might become more audible and cause some excessive brightness, depending on the material of course. It is rather difficult to determine what exactly happens of course...
 
Last edited:
A_Tewinkel, I bet that had was partly or largely to blame for my dislike of my own narrow pattern speakers close up. 3-4 meters away and beyond, they were great. Close up they had a lot of issues. Fortunately my listening distance is beyond 3 meters. With my current system using more wide dispersion speakers, I haven't really noticed any issue at all. Nothing to be concerned with anyway.

Probably part of the reason why near field monitors have a broad pattern. I've listened to all these speakers for some time in both mono and stereo barring the 1030A(though I've listened to it both ways just not for a ton of time): audio blog: Review of Polar graphs
Much more so in stereo however. The narrow patterned ones have given me more fits with set up for sure. In an 'ideal' world where the recording is king(and I'd love for it to be), they seem like the wiser choice. Perhaps the only. I love those Mackie monitors-designed to be mid field. Just stay away from them and they really expose what's going on. Which to me means they have great intelligibility in spite of their narrow pattern. Knowing how to use your speakers is so critical.

We need reflection I think, lots of them--and well within the Haas interval just not too early. Where that point is I don't exactly know. Mr Danley might disagree, but I'm not a fan of outdoor shows. Seems the reflections you get there are just too late and destroy intelligibility and provide no envelopment. I do have friends that talk about the sound as a good thing outdoors. Beats me. Maybe b/c the bass or probably mid bass has a quick decay.

Narrow pattern speakers in a dead room do seem like a bad idea to me. Most living rooms here is the States have a comfortable quantity of reflections when furnished I think. Just my experience.

Oh, and I'm with you now on the mono for PR/tonal change. It may not make a real difference on the test, but it may and having less things to question sure is better IMO.

Dan
 
We need reflection I think, lots of them--and well within the Haas interval just not too early. Where that point is I don't exactly know.

[...]

Narrow pattern speakers in a dead room do seem like a bad idea to me. Most living rooms here is the States have a comfortable quantity of reflections when furnished I think. Just my experience.

I agree. Looking at some of the detection thresholds in Toole and the image taken from Barron, I'd say 8-10 ms more or less reflection free would be desirable, but we've debated this long enough in some other thread :D

By the way, haven't mentioned the article this threat started with yet but it was nice to read. The link between correlation and colouration of reflections is interesting.
 
Dan wrote;
“Mr Danley might disagree, but I'm not a fan of outdoor shows. Seems the reflections you get there are just too late and destroy intelligibility and provide no envelopment.”

I am not sure I disagree but I do think this topic generally is spanning several separate issues and treating them as one.
These would be something like;
The recording process, how the stereo image is created, what is it actually supposed to sound like? Since most recordings are entirely contrived spatially speaking what is the reference?
Here, the home experimenter can use a measurement microphone and sound card to make recordings which a quality unattainable at the semi-pro level in the “good old days”. You can only make mono recordings though because no mic placement of a pair captures a real stereo image and a mono signal is the ultimate torture test for a stereo system.

The speaker, does it shout out it’s physical position in depth? Generally, the larger the number of drivers, the more easily you can hear how far away it is (with your eyes closed).
That happens because what reaches each ear, even with just one speaker, is different enough that your ear/brain can tell you where the sound is coming from in depth. A speaker like a quad esl 63 or SH-50 playing a voice often sounds like the sound is “somewhere” behind the speaker, in the case of the SH-50, it sounds like it’s coming from somewhere in front of you even with your head inside the horn.
Both of these radiate a portion of a sphere as if the sound originated at a single pint over a broad band. What reaches your ears (with one speaker) is much closer to being the same and so you can easily tell direction but not distance when your eyes are closed.

So what? Aren’t you supposed to hear your speakers, like the neighbors child who imagines she has a good voice on stage at a choir concert?

It wouldn’t matter at all until you have a stereo system, now, you do not want what I call “source identity” when producing a mono signal because it (the source on each side) competes with the image your trying to create with what is ideally identical signals in each ear.
If you have a speaker that radiates like a simple source, you have something your ear interprets as happening inexplicably, something it doesn’t recognize but does recognize the aural q’s within the recording (or not).
It is those little things that can convey the feeling of being outside or in a cave.
In large scale sound the speaker systems have even more self interference than home speakers and so are even farther from being a simple source. If even a small wind blows, the radiation is moved around slightly and the variation in the radiation is clearly audible. Do the same stadium with one of our large Synergy horns which radiate as a single source and even in a 30mph crosswind the sound is hardly effected.


I suggested the outdoor test for a reason too. One cannot make an argument about the desirability of room reflections unless one has listened to the same speakers where there are no room reflections, outdoors (if you can). I suggested having a bbq and making an afternoon of it listening to your most familiar recordings. I have done this so many times (and had to go back to the same speakers indoors) that for me the conclusion is inescapable. You preserve FAR more of what is on the recording the less room reflections, especially close ones from side walls you have.
What do side wall reflections look like?

Sit in your listening room, empty all the furniture (mentally) and imagine your side walls are full size mirrors. You look left and right and wave at the reflections of your R and L speakers in each mirror. What you hear in the reflected sound from the side walls is a lot like it came from those images in the mirrors. Now imagine a mirror on the floor and ceiling and back wall. When you turn your head, how many images of the R&L speakers can you see?
Each one of those reflections represents a snippit of the original single, altered in spectrum and each arriving at it’s own time according to it's path length. A single impulsive signal fed to the speaker, arrives spread out from the loudest/closest arrival first to the weakest last reflected signal some time later.
A typical 3 way hifi speaker might look great at a meter but if it’s + - 10 or 20dB at the listening position as a result of all the reflected sound, how can that be thought of as anything but more or less inevitable given the speakers being used? If it also screams out “here I am” when it is supposed to be producing a stereo image, no wonder the mono phantom image is wide or even three images.
Best,
Tom
 
It's in section 9.1.3 in Floyd Toole's "Sound Reproduction". His reference is this article, at least for the last of the three measurements, which was done in a reflective room.

The dip is 10 dB at 1.9 kHz in an anechoic setting, compared to a centre loudspeaker. Also, the frequency response rises above that with about 3 dB compared to a centre speaker.

In the reflective room, the difference is about 4 dB at 1.9 kHz and level rises to +3 dB for higher frequencies, with a mild dip at 6 kHz (0 dB).

In my view those are numbers not to be ignored.

I think a more valid comparison is the tonal error between center and phantom created by +/- 30 deg.

Here's my analysis on the subject:

http://www.diyaudio.com/forums/multi-way/172806-flat-not-correct-stereo-system-19.html#post2363393
 
Tom, I don't think anyone can logically disagree with what you say, but we have been shown to 'learn' the room signature above the bass and to have a poor ability to hear the time domain issues in the bass. Maybe not perfect, but what is? The room adds it's signature that can be modified and no one I know wants to totally remove. I've never heard an outdoor system equal what I have in my house as far as sound enjoyment is concerned. I'm actually planning an outdoor system right now to enhance my BBQing experience. It would be cool to hear my current system outdoors to get a better idea of the room's impact.

DDF, is there any recording made where this HRTF issue exists? I figure you must have a library full of it b/c you keep referring to your theory. Do you know of even one available for purchase? I don't, but I'd love to hear one or at least know how to get one. I have an eclectic palette, so don't worry about what I might like.

FWIW, even the cheap mixing boards have EQs built in. Nice ones are very complicated full of all sort of effects and several routes to add more in several ways. There can be walls of stacked hardware effects and software plug-ins that would probably blow your mind. I have several hundred plug-ins and I'm no pro or even a gear obsessed amateur. They just came with the software. Dozens more available for free and far more available for a price. No pro recording outfit I know of is stocked with worse equipment than you can buy in the store for under $100 and I've never heard of an engineer who won't use their gear. They actually buy all that stuff to use it. Many have favorite vocal mics that cost around $10,000 that aren't even designed to be flat(lets ignore pattern for simplicity sake)! I have a locker full of mics, none of them measure flat. Not even my measuring mic, designed to be flat, is flat! Had to get it calibrated. Anyway, there are some free softwares for the iPod that are simple. Heck, some of them only do mono! No effects, no panning, just level adjust. Oh wait, that means no HRTF issues when played back on one speaker. Hmmm... should have done that a long time ago... he he Wait till I tell the pros how easy this is to get the tone right!

Thanks in advance for letting me know,

Dan
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.