My old article on Loudspeaker Imaging

[rone]

A couple (perhaps unpopular) observations


Without writing a huge essay in response, here are a few observations.

I would think that a true engineer would be more interested in finding out why those reproduction chains (turntable vs turntable vs CD) sounded so different, rather than drawing pseudo-scientific conclusions about it. One _much_ simpler explanation is that the Linn was louder.

It has been mathematically proven that stereo loudspeakers cannot reproduce depth. I have listened to some really high end systems, both analog and digital and none of them had any depth, IMO. Perhaps my ears have lost their "fractal perceptiveness". ;)

Regarding analog having lower level information - this is pretty much taken care of by dynamic range and masking theory. If this low level info were so important MP3's could not sound "anything" like the original. I think the true difficulty in comparing digital vs analog for those not involved in recording is the lack of available "identical" recordings.

If you compare an old vinyl recording of something like Dire straits Telegraph road with a modern CD pressing, you aren't getting the same recording - you are getting a "remastered version" that has been mucked with to suit the modern engineer or modern taste. This is a problem with any comparison - are the CD and LP recordings "identical"?

As part of the Minnesota Audio Society I recently took part in a demonstration of 4 recordings. Ostensibly these were 50's vintage album and 15ips reel to reel versions of an RCA living stereo recording compared to a recent CD and SACD pressing.

To my ears, the SACD and CD sounded the same, and the album and tape had the same balance, but were much brighter than the digital recordings. The tape was played back on a vintage Crown reel to reel and there was a LOT of tape hiss. The difference in spectrum alone was enough to make the album sound preferable to the digital - let alone uncontrolled level differences. The volume knob kept getting jacked up and the digital was played back first - the tape last. Louder is better - a well known psychoacoustic fact.

That's enough of the point by point arguments, although more could be made. I think your essay could benefit from more substantiation of your assertions, and fewer hand waving analogies. I laughed out loud when I got to the "sonic saran wrap" part. While well written from a grammar standpoint, some might prefer more "meat with their potatoes". ;)

[Paul Carmody]

Thanks... I have a few points to bring up


Thanks for that contribution. I really wanted to hear a well-thought-out discussion on imaging. You raise a lot of good points.

I can definitely appreciate your describing detail as "fractal," and I see where you are going with it. However, I'm not sure if I am with you in the digital vs. analog point. (I realize that I'm bringing up an argument that's as much of a slippery slope as interconnect cables, but oh well.)

Anyway, you make the claim that digital has a low-level "floor" that limits what it can record, and that analog does not. However, I have read that analog has the equivalent of about an 80 dB dynamic range, while digital is 90+. My experience with analog mediums has shown this to seem true to my ears. Plus, analog recording DOES have hiss at the noise floor, so even if you say that there is "fractal" information in the single-digit dB range, wouldn't it be eclipsed by hiss?

Anyway, maybe it's not that much an argument worth having, and what we REALLY ought to do is band together and demand to today's record labels: "WE WANT OUR DYNAMIC RANGE BACK!"

By the way, I REALLY like all your points about "smearing," both within speaker cabinets and within a recording. Because yes, most rock and pop recordings nowadays are recorded in different rooms at different times (sometimes in different cities). And all that reverberated content WILL confuse a listener's ears. It really made me think about how I will mix my next album. So thanks!

[Jeff B.]

Re: Thanks... I have a few points to bring up


"Anyway, you make the claim that digital has a low-level "floor" that limits what it can record, and that analog does not. However, I have read that analog has the equivalent of about an 80 dB dynamic range, while digital is 90+. My experience with analog mediums has shown this to seem true to my ears. Plus, analog recording DOES have hiss at the noise floor, so even if you say that there is "fractal" information in the single-digit dB range, wouldn't it be eclipsed by hiss? "

A valid point, and one that I am not inclined to dismiss, because I agree more than I disagree. I really see no reason why digital couldn't out-perform analog in this area. However, at the time I originally wrote this I was still experiencing better image placement from analog stuff than from my digital material. Now days I listen entirely to digital, even though I still have a turntable. I just don't pull those albums out much these days. And as the fractal analogy goes, nothing is purely fractal, except maybe the math. All real things have a limit somewhere. I just used it to help get a picture of what I was referring to.

[davidyohn]

Re: A couple (perhaps unpopular) observations


> Without writing a huge essay in response,
> here are a few observations.

Thank you for your comments. I feel exactly the same way.

[Jeff B.]

Re: A couple (perhaps unpopular) observations


> Without writing a huge essay in response,
> here are a few observations.

> I would think that a true engineer would be
> more interested in finding out why those
> reproduction chains (turntable vs turntable
> vs CD) sounded so different, rather than
> drawing pseudo-scientific conclusions about
> it. One _much_ simpler explanation is that
> the Linn was louder.

> It has been mathematically proven that
> stereo loudspeakers cannot reproduce depth.
> I have listened to some really high end
> systems, both analog and digital and none of
> them had any depth, IMO. Perhaps my ears
> have lost their "fractal
> perceptiveness". ;)

> Regarding analog having lower level
> information - this is pretty much taken care
> of by dynamic range and masking theory. If
> this low level info were so important MP3's
> could not sound "anything" like
> the original. I think the true difficulty in
> comparing digital vs analog for those not
> involved in recording is the lack of
> available "identical" recordings.

> If you compare an old vinyl recording of
> something like Dire straits Telegraph road
> with a modern CD pressing, you aren't
> getting the same recording - you are getting
> a "remastered version" that has
> been mucked with to suit the modern engineer
> or modern taste. This is a problem with any
> comparison - are the CD and LP recordings
> "identical"?

> As part of the Minnesota Audio Society I
> recently took part in a demonstration of 4
> recordings. Ostensibly these were 50's
> vintage album and 15ips reel to reel
> versions of an RCA living stereo recording
> compared to a recent CD and SACD pressing.

> To my ears, the SACD and CD sounded the
> same, and the album and tape had the same
> balance, but were much brighter than the
> digital recordings. The tape was played back
> on a vintage Crown reel to reel and there
> was a LOT of tape hiss. The difference in
> spectrum alone was enough to make the album
> sound preferable to the digital - let alone
> uncontrolled level differences. The volume
> knob kept getting jacked up and the digital
> was played back first - the tape last.
> Louder is better - a well known
> psychoacoustic fact.

> That's enough of the point by point
> arguments, although more could be made. I
> think your essay could benefit from more
> substantiation of your assertions, and fewer
> hand waving analogies. I laughed out loud
> when I got to the "sonic saran
> wrap" part. While well written from a
> grammar standpoint, some might prefer more
> "meat with their potatoes". ;)

Well, I'm glad you liked my grammar. I have seen you get on some guys about this in the past, so I'm feelin' pretty good now.

I think my assertions could stand a little more substantiation as well. I said upfront that much of this was speculation, and some of it could be all wet. Some of it I do not even know exactly how to test objectively. Other parts would not be hard to do. For instance, it would not be that hard to show some correlation between a speaker's sujectively evaluated imaging ability and the cumulative spectral decay of the various panels of its enclosure if this existed, and I think it would be worth doing if someone wanted to.

The only point you make that I would question would be the part about it being mathematically proven that two loudspeakers can not reproduce depth. Are you telling me that in all of the time you have been doing this you have never heard a set-up that produced a depth of image that made your jaw drop? I'm bummed for you, man.

The set-ups and processors that have made an attempt to cancel interaural crosstalk have been credited with some pretty startling demonstrations. I will never forget years ago when I heard a demonstration of Carver's Sonic Holography preamp. The recording used was "Time" from Pink Floyd's Dark Side of the Moon album. It was played over a pair of Dahlquist speakers, and that was all. I was amazed at how these clocks were going off all around me, even behind me. It was really a jaw-dropping experience. These clocks were placed in space and their image was stable enough that you could sit there and point at where each one's sound appeared to be originating from. I don't see how the processor could have created this realistic placement and stability artifically, rather it seems this information was there all along in this recording and the processor was simply able to bring it out.

As for the Linn. It wasn't a matter of being louder. It was a matter of localization of sounds having this depth (that's mathematically impossible to have. Oops, forgot that part ;-) compared to the Rega's "flatter" sound stage presentation.

Now, when you were laughing at my Saran wrap analogy I hope you were enjoying it and not laughing AT me, because that might hurt my feelings, Ron ;-) Anyway, sorry I left out the meat, glad I gave you some potatoes

Jeff

[Guest]

Re: Yes, thanks JeffB, now I need a cigarette!!


That was an incredible read... except for a few points I didn't follow, that was a very understandable and a far greater detailed opinion of what many of us "old school" hifi fans have believed for quite some time.

I can only wish my music collection was analog, my listening room much larger/acoustically neutral, and my wallet able to add more active channels to my matched tone/time/phase "dream system".........

Thank you very much for the post/info... it's going in my storage files!!

GC

[kene]

Re: A couple (perhaps unpopular) observations


> One _much_ simpler explanation is that
> the Linn was louder.

These kinds of hand-waving conclusions are way overplayed, IMO. The majority of stereo systems sound better and more accurate turned down, as far as my ears are concerned. Better yet, turned OFF. Basically, I don't think my ears like distortion, which almost always increases as volume goes up. My *opinion* is that these "louder = better" results primarily reflect the people they are studying, who may not know any better...even when they're audiophiles.

> It has been mathematically proven that
> stereo loudspeakers cannot reproduce depth.

This is a nice-sounding argument but is completely irrelevant - your ears don't actually hear depth, either. They interpret the presence of additional acoustic cues to secondarily derive an idea of how far away things are. This involves the presence of reflections, their amplitudes relative to the direct sound and time delay as well as changes in tonal balance. It's just like when you walk into a room with a lot of reflections and you can tell the room is acoustically "hot". If you record a person talking in such a room and play it over speakers with good fidelity you will still hear how "hot" the recording room is. The better the fidelity, the easier it is to hear. Depth operates using all the same cues - you just have to notice the fact that you can hear it; there's nothing magical about it at all. The better your system resolution, the more realistic the reproduction and the more convincing the presentation. Pretty simple.

Matter of fact, there's plenty of recording equipment out there (designed by engineers, I suspect) that are employed to artificially recreate the sensation of depth over instruments, and some artists use such equipment in really interesting ways. I just mentioned in a nearby thread one very easy example you can hear for yourself: NIN's "Closer" features an odd violin-like instrument that moves around in a circle, most obviously right at the end of the song. You may want to fast-forward to this part of the song if you don't like vularity, though. Close your eyes and turn off the lights, it should be pretty hard to miss even on a lot of less-ambitious systems assuming you've setup your room and system appropriately, which I would imagine to be the case.

> Regarding analog having lower level
> information - this is pretty much taken care
> of by dynamic range and masking theory. If
> this low level info were so important MP3's
> could not sound "anything" like
> the original.

Uh, low-rez MP3's DON'T sound anything like the original. Haven't gotten to hear any "good vinyl" systems though, so no comment there.:-)

[damkor]

Magic elves in the fractals?


"Unveiling lower level imaging information requires removing the multilayered veil that hangs over the sound in the first place."

AH! You used the word "veil"!

"The key is to progressively strip away these veils..."

You mean resolve the source as precisely as possible?

"For instance, by nature analog recordings are fractal in structure, but digital sources are not."

I see what you mean, but that doesn't matter. The end output of a digital or analog source is the fractal reality produced to a certain level of resolution. Beyond that level, the analog output adds its own additional fractals (distortion) and the digital output does not.

But in terms of the original shape that was really there in the source material, the digital source should have higher resolution, right? In other words, the analog system's ability to output fine fractal details is really distortion that we don't, hopefully, notice.

To say that the additional levels of detail from the analog source, the chaos, which is simply low level distortion, add to imaging would be magic elf talk.

[damkor]

Really? You can't hear depth?


Ron E.: > It has been mathematically proven that
> stereo loudspeakers cannot reproduce depth.

Ken: "This is a nice-sounding argument but is completely irrelevant - your ears don't actually hear depth, either. "

Thanks for clearing that up for me. You don't even need an especially good music system to hear vocals seeming to appear in front of you, and drums in back. And then you ask the engineer and he says that's what he was going for, and you see the band live and that's the way they set up! And that's the way it sounds. I mean: Come on! The sound system is succesfully re-creating the illusion(?) of depth in the sound image. Depth that was there when the band played in the recording studio. Is that OK?

Or should I actually be striving to perfect my sound system so that it can no longer re-create this sound image, since it is an illusion? Back to mono? Ridiculous!

"Yeah, actually, the "appearance" of those objects in space to the right, left, in front and and behind the speakers? That may sound like realistic 3-dimensional imaging, but it's actually an illusion, so I go with a single Bose 301 now."

BTW, the objects that appear on the computer monitor in front of you and on your TV screen are only illusions, so you should turn them off so as not to be taken in by the distortion of reality!

[kenperkins]

Re: Really? You can't hear depth?


Kinda makes you wonder why Opus 3 has had the "Depth of Image" test record in their catalog for over 25 years.

[dbe]

some rambling observations...


of your observations. An indepth discussion would take hours of face time.

observations.

The differences are all defined by microdynamics or the lack thereof. Microdynamics is a phenomena that is further defined by 2 major (and some minor )players. One is the noise floor of the system and the other is the faithful reproduction of small signals. On the surface these would seem to be the same, but in a digital/analog comparison they are not. Nyquist frequency limits the small signal component while providing a very low theoretical noise floor in a digital recording. When you zoom in on a waveform you will see that there is a lot of small deviations imprinted upon the musical signal present in an analog recording/ and the absence of these deviations in a reconstructed digital signal of the same section of music. These small deviations are the "fractal" issues that Jeff was talking about. They are supersonic harmonics of the parent signal that a digital sampler cannot reproduce due to the sampling rate limitations, but they are also the timbral components that give analog its' characteristic sound. These cues are perceived but not heard in a very good analog system with extended HF response. The cilia in our ears move at frequencies in the 80-100 KHz range and send chemical messengers to the brain telling it something. Perception is reality in audio as in politics. In a Red Book recording virtually everything above 20 KHz is thrown away by filters that have their own characteristic sounds. You should sit in a mastering session and listen to the differences in digital filters, someday. There are distinct signatures unique to different order filters as well as different types of dither that use up the least significant bit or two by creating noise that is an emulation of the furry analog signal with the supersonic components. Dither tries to fool the brain into thinking that there is consonant HF data in a digital recording where there is none. There are some forms of dither that are very effective. Damkor is wrong about the digital signal being less distorted than the analog signal in this regard.

Depth is an illusion in a stereo recording (itself an illusion of binaural) that is created by microdynamic contrasts in the source recording. My motivation in audio is one of reducing noise in every aspect of the recording/reproduction chain. Recording is the home of GIGO. A good recording begins with the right mic on the right instrument in the right place in a very quiet room feeding a very low noise signal chain. Unfortunately many studios are not "low noise" in any aspect. One recording console that I modified years ago started out with a S/N ratio of 68dB with all of the buses and faders at unity gain. This console was an AMEK Einstein designed by Rupert Neve (a great designer). Good design, crappy components, a victim of the accountants at Harmon International. Anyway, by designing a new very low noise high speed power supply, changing opamps and replacing (or eliminating some of) the 32 (count 'em, 32) coupling capacitors in the signal path of each channel the console and a few other tweaks, we improved the S/N ratio to 82dB. These changes (in conjunction with others too numerous to go into here) made a profound difference in the recording quality of the console. This enabled the mixing engineers to place, psychoacoustically, instruments farther back in the mix using delay, reverb and amplitude without noise (that is generally white in nature) masking the effect of depth. These effects are of very low amplitude and just get lost in noise...

That is the same type of thing that goes on in the comparison of the Rega and the Linn Sondek. It is ALL about noise. The Sondek is very low noise design whereas the Rega is not. This is the kind of noise that you don't really notice until it isn't there. Noise masks microdynamics. Microdynamics are what define systems. Microdynamic cues are the qualifyers of engaging musical reproduction. Microdynamics are the elements that give rise to "feelings" in audio. (see the link below). A low noise system will resolve data that a high noise system will not. This is where the whole "snake oil" realm lies. Obsessive compulsive behavior takes over at this point and the objectivist / subjectivist battle is on, full boogie. It is the old 80/20 concept. 80% of a systems’ performance is bought by 20% of the dollars. Otaining that last 20% of potential is where 80% of the money in an audiophools’ pocket goes.
These ramblings only begin to scratch the surface of soundfield, the place where “imaging” lies.

Here is a link to a basic paper on perception of sound that has a wealth of internal links that further illuminate the subject:

<A HREF="http://www.pubmedcentral.nih.gov/art...i?artid=400748">http://www.pubmedcentral.nih.gov/art...i?artid=400748</A>

Good discussion ;-)

Dave

[alann]

Re: Thanks for your comments and the link *NM*

[alann]

Re: Thanks for reposting your article *NM*

[Steve H.]

Re: My old article on Loudspeaker Imaging


There are definitely some good thoughts in this thread, how about archiving it all on <A HREF="http://www.audiodiycentral.com">www.audiodiycentral.com</A>?

Steve

[dbe]

Re: Magic elves in the fractals?


> "Unveiling lower level imaging
> information requires removing the
> multilayered veil that hangs over the sound
> in the first place."

> AH! You used the word "veil"!

> "The key is to progressively strip away
> these veils..."

> You mean resolve the source as precisely as
> possible?

As precisely as multiple systems will allow: different systems have inherently different non-linearities. Picking nits, I know, but...

> "For instance, by nature analog
> recordings are fractal in structure, but
> digital sources are not."

> I see what you mean, but that doesn't
> matter. The end output of a digital or
> analog source is the fractal reality
> produced to a certain level of resolution.
> Beyond that level, the analog output adds
> its own additional fractals (distortion) and
> the digital output does not.

Actually, it does matter in that analog is an infinte number of time dependant samples and digital is a finite number. Analog "fractals". as postulated by Jeff, are elements of the harmonic series that expend into supersonic realm of perception. Quantization limits the ultimate bandwidth of the digital system to an upper frequency limit. Analog is capable of 80 kHz in tape (with a good headstack) and on vinyl with some systems and timbral information as explained (briefly) in my post above. Tape hiss, clicks and pops are non-consonant to the encoded signal and are differentiated as separate from the musical system by the wonderful error correction system called the brain. Psychoacoustics is a fascinating thing. Digital information is limited by the Nyquist frequency AS noted above. Over-samplers do some interpolation of high harmonics but this interpolation is a quess, not hard data.

> But in terms of the original shape that was
> really there in the source material, the
> digital source should have higher
> resolution, right? In other words, the
> analog system's ability to output fine
> fractal details is really distortion that we
> don't, hopefully, notice.

Again, see above. Think of music like a prime cut of steak. Analog keeps all of the marbling that gives steak its' flavor. Digital is just expensive hamburger.

> To say that the additional levels of detail
> from the analog source, the chaos, which is
> simply low level distortion, add to imaging
> would be magic elf talk.

Low level signal in analog is not chaos. Dither in digital, being random or semi-random, is chaos by definition.

Dave