Paul's Posts — 02 July 2012
By Paul McGowan
I find it valuable to create thought puzzles to try and answer perplexing questions about our art. I thought I might share a few with you and then post about some of the conclusions I’ve drawn from this process in a few days. You may find it interesting.
Let me begin with something we know and are clear about. If you take an electric guitar and plug it into a decent digital audio recorder and then create a bypass switch that allows you to select between the guitar output or the recorder’s output and feed this into an integrated guitar amp/speaker combination – the output of the guitar amp/speaker will be indistinguishable switching between the live guitar and the recorded version of the same if the listener cannot see what’s going on and you’ve taken care to gain and impedance match the interface.
The difference between the direct output of the guitar’s pickup vs. the digitally recorded copy of the pickup’s output – when played back through the amp/speaker combo – cannot, in my experience, be measured or heard.
I have performed this experiment years ago and with reasonably poor digital recording equipment. Using a state of the art A/D and digital recording DAC playback chain of today’s caliber makes this experiment even more interesting.
If this is true then one would have to conclude that in this specific instance, the recording chain isn’t contributing changes in dynamics, tonal qualities, pacing or any other high-end parameter many of us attribute to digital recording – even at Red Book sample rates.
But we all believe this to not be the case. Interesting.
Tomorrow let’s go one step further.
Forward to a friend and help us engage more readers
Related Articles
Share
About Author
(18) Readers Comments
Leave a Reply
You must be logged in to post a comment.
Jackson
I have alway thought high fidelity challenges applied to acoustic music. I have never understood judging HiFi equipment with amplified music.
Bassman23
Agreed. Whether a sine wave or a signal from an electric guitar, electrically generated signals are the easiest to get right. When brought into the acoustic domain, I have heard triangles reproduced with great apparant fidelity from quite low fidelity systems because of the purity of its tone.
The greater the diversity of acoustic signals, the more difficult the challenge of recapturing them – including the capturing of the original acoustic space.
Paul McGowan
Exactly. That’s why we’re starting at the beginning. You guys are too fast for me.
Soundminded
This idea is not hypothetical or it least it wasn’t always. This was the purpose of the tape monitor/source switch found on preamplifiers and preamplifier sections of integrated amplifiers and receivers. The switch acted as a shunt that could be inserted or removed from a processing signal path that included the tape amplifier, recording head, magnetic tape itself, playback head, and tape playback preamplifer. The idea was to set gain, bias, and equalization to get the least effect on the signal. The principle can used to test what disortion audible or measured is added by an element in a ciruit including an interconnect cables, a speaker wire, or even a power cord. Also for preamplifiers and even power amplifiers. To perform many of these tests you have to build a relay arrangement and especially in the case of power cords and power amplifiers you’d better know exactly what you are doing or your experiment can lead to disaster. Advocates of audiophile wires especially don’t like these tests arguing for many reasons they don’t give you valid results. Personally I think their real objection is exactly the opposite, that they do show what they do or don’t do. Any wire that sounds different from a shunt is defective to the degree that it is different because a wire is not intended to be a control (signal altering) element in a circuit. The cheapest $1 Radio Shack 3 foot patch cord passes this test no only with audio signals but with far tougher NTSC video signals having 300 times the bandwidth with flying colors every time I’ve tried it.
“one would have to conclude that in this specific instance, the recording chain isn’t contributing changes in dynamics, tonal qualities, pacing or any other high-end parameter many of us attribute to digital recording – even at Red Book sample rates.”
Not everyone beleives every process contributes audible distortion. Some of us (at least one person, me) believes exactly the opposite. Having studied some clinical psychology, the science of sensory perception back in college I am aware that for every one of our senses there is a threshold for changes to stimulii below which that change is not perceptable. Where that degree of change lies for a given stimulus depends on many factors including the nature of the stimulus, the person being tested, and even varies in the same individual from one time to the next. When large numbers of people are tested under a given set of conditions, the results shown graphically are expected to look like a bell curve, some extremely sensitive to changes, others almost completely insensitive, most of us somewhere in the middle. Audiophiles tend to exaggerate in their own minds their ability to detect small or even significant changes. Often in blind tests they demonstrate that their acuity is no better than a random guess would be. However, as was pointed out yesterday by Paul, the gulf between what technology can deliver at its best in this art and what real music sounds like is so great you don’t need a test this sensitive to detect it.
About the best that can be said for arguing against this type of test is that when enough small incremental inaudible distortions are added together they can add up to enough distortion to become audible. Theoretically this is true. However, consider the dozens, hundreds, or even thousands of transistors, vacuum tubes, wires, capacitors, switches, patch cords and whatever else you get when you buy a commercial recording the signal has been processed through prior to your getting it and then consider how few you will introduce and the argument seems rather silly.
Ivan
If a difference is measureable or audible, a difference exists.
If it’s measurable but not audible, it exists but has no significance.
If even one person can hear it, double blind, at a statistically significant level, then it’s audible–even if neither you nor I can hear it.
But a roomful of audiophiles hearing it while they watch the switch is meaningless.
Soundminded
“If even one person can hear it, double blind, at a statistically significant level, then it’s audible–even if neither you nor I can hear it.”
I’m not sure what that means but if one person in ten thousand can see light that is in the ultraviolet range that is beyond what the other 9999 can see we do not design film, cameras, or TV sets to standards that will include the ability to capture, record, transmit, and reproduce those frequencies. It’s amazing how easily audiophiles can be fooled and how not out of the ordinary their hearing and listening abilities are. This was demonstrated in many different ways and well known decades ago but most of the documented evidence for it hasn’t been published in a very long time.
“If it’s measurable but not audible, it exists but has no significance.”
At least not until enough of such distortions pile up to add up to an audible distortion. Then the panic sets in and audiophiles go out to kill every source of distortion real or imagined they can find no matter how insignificant. That’s where a lot of manufacturers and salesmen make their profit, on fear and lack of knowledge. The giveaway is usually that these people talk in generalities without hard numbers, data, or context. So they show you two graphs and one is clearly better than the other. However, what they don’t show you or tell you is that the worse of the two is far below the threshold of what you can hear so if you’re like many audiophiles you buy it anyway. And then when you get it home you’ll swear you hear the improvement. Noboby likes to admit even to themsevles that they’ve been duped, outsmarted.
petewilson
Suppose Paul’s claim is completely true – that is, that you really cannot tell the difference with ear, mind or equipment between the direct input and the A/D-D/A path. But that you can tell when you listen to ordinary recorded music that there’s something not there.
Then that strongly suggests that microphones have an effect on what they ‘hear’.
Not a surprise. They’re mechanical, and have non-linearities at high volume, and resonances, and…
It would be interesting to find an equivalent test for loudspeakers, which have a harder job to do (they have to move lots of air, and so mechanical issues arise)
And it’s actually what was being said earlier – the electronics are pretty good these days. But the trnasducers are an issue.
– P
Soundminded
The sensitivity of the test is subject to certain variables such as the program material being used. If for example the recorder cannot properly reproduce some frequencies that are not present in the test material that won’t show up in this test. Suppose the recorder cannot reproduce very low bass frequencies but there are none in the guitar music, you won’t find that out from this test. It is difficult to devise a fair test that will tell us everything we want to know. Some people feel pink noise as a test signal is very revealing.
That being said, in this test, only those components being tested by being shunted out are subject to being determined to introduce audible changes or not. The test does not speak to the adequacy of the sound system as a whole to perform its intended function. For example if the speakers cannot reproduce very low frequencies either then whether or not the recorder can do that makes little difference in that sound system. It might be a very different case in another sound system.
Steve Parry
Guitar speakers radiating in a large space are not as sensitive as a home system. Many people enjoy the sound of their car stereo…and may hate to admit it!
The crazy thing is we expect the same sound at home as we hear live. Pretty preposterous, and quite a wall to scale. This is a magic act. Figure out the music you like, the aspects of sound that matter to you, and choose wisely! This is why there are so many products in the market.
Soundminded
The analogy between a magic act and a successful recreation of music from a recording is a good one. The first thing a successful magician must do is determine what the visual field would be from the audience’s perspective of the illusion he is trying to create. Then he must invent a way to create that same visual field by different means. Therefore we see the visual field that a woman has been sawed in half or in thirds and different parts of her body are being wheeled around a performing stage although she seems perfectly fine. We know that can’t be what’s actually happening. So far those who want to create an audible illusion of music haven’t even figured out the first part of the problem, small wonder they couldn’t possiby have solved the second part. That it has not been done doesn’t demonstrate that it can’t be done. The real mental puzzle we should be considering is the one which explains why.
Paul McGowan
Yes, but how the hell do they do that trick anyway?
Soundminded
If I told ya, I’d have to kill ya
Actually if you really want to know, there are books in the library that explain exactly how that and all other standard magic trick illusions are created. You can probably also find them on the internet. However, I advise against reading it. Once you know how they are done, that takes all the fun out of seeing it from then on. By contrast, knowing how the audible illusion is created doesn’t diminish the enjoyment of it one bit. When well executed, it never fails to please. It’s also a difficult illusion to create.
Paul McGowan
I am sure there are and I really only ever wanted to know how that one was done. I don’t want access to all of them because as you suggest it removes all the fun.
Soundminded
Here’s a start to understanding the illusion. There are clearly variants of it, different ways to present it. Learn it, practice it, and eventually you’ll be sawing women in half with the best of them I’m sure. Do you actually want to saw women in half painlessly?
Do enough of it and eventually there will be twice as many of them 
Soundminded
Oops, I left out the link for the saw trick;
http://en.wikipedia.org/wiki/Sawing_a_woman_in_half
There are many others sites as well. I hope your spouse is an understanding assistant as you practice this “art.”
oliver T. Finch
Using the sound of an amplified guitar as a standard for judging fidelity does not seem right. The sound is too simple,lacking in complex overtones compromised further by less then stellar amplification. Reproducing such mediocre sound with perfect fidelity digitally does not prove high fidelity.The sound of an amplified guitar is bright to begin with and so is the sound of digital. Regards.
Paul McGowan
Actually there are very complex overtones – but that wasn’t the point. Hang in there and I’ll help explain why this is an ok idea to roll around in your head. I am not trying to convince you to listen to guitars! Digital is not bright – by itself – we’ll cover that topic one of these days as well.
adam75
I am a bit late getting back to you on this one. It seems to me that there is nothing really surprising in the result an A/D-D/A can be shown to be transparent to the extent of several iterations. After all, if done properly all it will add will be some dither noise at a very low level -even at 16 bit resolution.
The implication of all of this is that, as you show on the next few postings, the real losses in the hifi system are at the transducer chain and the room acoustics. You will be aware though that many people including Ethan Winer and AVI, argue that this means that amps and D/A converters are essentially transparent, and that money should only be spent on speakers and (in EW’s case) room acoustics.
In any event I am not sure i understand how this reflection is consistent with the suggestion that changes the software on a server (where the data is either way streamed bit perfect via ethernet) could audibly affect the output of the dac.
How could the PWD be in a worse position than the dac in the A/D D/A chain?