"Comparison by Contrast" method for evaluation of the playback chain

[PaulT]

I've been a fan of this forum, checking in almost daily for over 4 years. I finally found the time (thanks to this brutal Winter)

and the courage to post a few thoughts, and as you can probably tell by the thread title, I'm jumping in with both feet!

I had originally intended to start a discussion on using the differences between audio recordings to evaluate accuracy in the

playback chain. After a few searches on the subject, I found this:

Are You On The Road To... Audio Hell? by Leonard Norwitz and Peter Qvortrup Quite a bit of information, and a far more interesting

read than anything I can ever hope to come up with.

I would like to say thanks to the recording industry professionals, IT guys, software designers, and everyone else, objectivists

and subjectivists alike. Because of your ideas and contributions I keep coming back for more.

 

 

Paul

[Boris75]

I

I would like to say thanks to the recording industry professionals, IT guys, software designers, and everyone else, objectivists

and subjectivists alike. Because of your ideas and contributions I keep coming back for more.

 

Thanks for your message. I would like to join my thanks to yours: I too find that one of the great strengths of CA is that its active membership mixes objectivists and subjectivists, in contrast with other forums which tolerate only one of both species.

 

This diversity brings a rich dialogue that is very valuable (at least to me).

[PaulT]

Hi Boris

Yes I agree. A savvy person will find value in both camps. One can not exist without the other.

[esldude]

Hello PaulT,

 

The article you link to is good. Especially the first half. Not so sure I buy their conclusion on how to proceed, but they are thinking hard and trying.

 

One could make several criticisms. One, as a reducto absurdum, they say the best system will show the most differences between different sources. In general I agree, but imagine a speaker with a very up-tilted response. Top three octaves boosted pretty good. As even fine recordings vary in upper octave energy from how close the mike is, to what they are recording with and other factors this would exaggerate differences. It would show more differences than a more even response. Yet it is less accurate. I have seen amateur recordists use monitor speakers with just this character to zoom in on the treble so to speak. Don't know if good pro recordists do that or not. I think not.

 

I also don't think most audiophiles expect their reference recordings to all sound the same. I pick a few that sound very different for just this reason. They do point out something that happens. Lets say you like a wide soundstage. You might listen and assume the wider soundstage is always better, but even on a recording with such a soundstage components that are inaccurate can exaggerate that. So their criticism of that is correct imo.

 

The idea to listen to what sounds like an hour of recordings before switching components sounds unwieldy. Your aural memory is overstressed verses a quick switch. You will become more comfortable with one vs the other using long auditions. And feel more confident of your evaluation in my experience. Many suggest one needs to live with a component for days to figure out how good it is. I have found one feels great about the decisions using that approach, but it actually is not very discriminating of real differences. And one is prone to be influenced by other things that aren't related to the sound. In some cases like the user interface that is very valid. In evaluating ultimate sound quality it leaves something to be desired.

[PaulT]

Hi elsdude,

 

I'm not sure I understand your speaker example with an up-tilted upper octave response. I think a speaker with this type of

response would impart a type of sameness to most recordings. Some recordings may sound good, but remember we are looking

for more differences or contrast between various recordings.

I find the aspect of a recording that makes it easy for me to discern differences is the recording (for lack of a better word) soundscape. Every recording is unique, just take in to consideration miking and recording techniques, recording studios sound

signatures, concert venues, just to name a few.

The beauty of this method is in its simplicity, it sets the bar low, one need not quantify perceived differences, you need only to

recognize there are more differences between recordings to satisfy the premise:

 

The more accurate system is the one which reproduces more difference, more contrast between various program sources.

 

 

Paul

[bdiament]

I've been a fan of this forum, checking in almost daily for over 4 years. I finally found the time (thanks to this brutal Winter)

and the courage to post a few thoughts, and as you can probably tell by the thread title, I'm jumping in with both feet!

I had originally intended to start a discussion on using the differences between audio recordings to evaluate accuracy in the

playback chain. After a few searches on the subject, I found this:

Are You On The Road To... Audio Hell? by Leonard Norwitz and Peter Qvortrup Quite a bit of information, and a far more interesting

read than anything I can ever hope to come up with.

I would like to say thanks to the recording industry professionals, IT guys, software designers, and everyone else, objectivists

and subjectivists alike. Because of your ideas and contributions I keep coming back for more.

 

 

Paul

 

 

Hi Paul,

 

The idea of what they refer to as "comparison by contrast" is something I mentioned years ago on another forum and have mentioned many times since.

It all comes back however, to two other things I've been saying for a long time and they are:

1. Different folks have different sensitivities to different aspects of sound.

2. What is "good", "better" or "best" depends entirely on what one seeks.

 

Some seek a certain "sound" from their system and pursue this. In that pursuit, I don't believe the idea of differences between recordings would be of use, as what is sought is a certain "character" in playback (which to my mind, is counter to the idea of differences). For my work and for my pleasure listening, what I seek is for the gear to "get out of the way" and allow me access to the recording itself. With this in mind, the criterion I use more than any other is direct comparison against the source. For my recording gear, I compare it against the direct mic feed. For the rest of the system, I compare playback of finished recordings (those we can purchase), with playback of the master used to make them (in those instances where I have access to said master).

 

The realization came long ago that since most enthusiasts do not have access to masters for comparison, there is another criterion that anyone can use. I realized this because every master I've ever heard over the years, sounds quite different from every other master. Different engineers (in the recording, mixing and mastering stages) have their own approaches and even when two take a similar approach, there are going to be differences. Even when the same engineer makes different recordings, particularly using typical studio techniques, there will be differences between those.

 

I came to see that the systems that "get out of the way" the best are the ones that reveal these differences the most plainly. Any coloration in the system will be applied to everything passing through the system and thus will in the end, diminish the apparent differences between recordings, as they will all exhibit this character upon playback. (Some examples that come to mind are a few recent, well reviewed DACs that, to my ears, put a treble "zing" on everything played through them. The reviewers rave about "detail" but I can remember when such behavior was referred to as "distortion". ;-})

 

The degree to which a component is truly neutral, truly transparent, is the degree to which it will reveal the greatest differences between recordings.

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

[PaulT]

Barry,

 

Thanks for your usual clear and concise response. I've been aware of the "comparison by contrast" method for some time (possibly through one of your earlier references). Its only been about the last year or so where I've put the method to use. Previously I've used what is described as the "comparison by reference" method. And the reference I am most familiar with is the human voice. So I would concentrate on getting the vocals right and hope everything else would fall into place. With this approach you can never be sure which is the more accurate and which is the colored component. Its possible that one component can be overcompensating for the deficiencies of another. I can see now this is not the way to go.

Some of the changes I've made to maximize differences between recordings include server and playback software, and cables. I stated in an earlier post the aspects of the recording I focus on are the spatial cues. The best way to describe the improvement would be to say recordings sound more like there are human beings behind the instruments (that is to say playing the instruments) as opposed to just hearing the instruments in space.

[bdiament]

Barry,

 

Thanks for your usual clear and concise response. I've been aware of the "comparison by contrast" method for some time (possibly through one of your earlier references). Its only been about the last year or so where I've put the method to use. Previously I've used what is described as the "comparison by reference" method. And the reference I am most familiar with is the human voice. So I would concentrate on getting the vocals right and hope everything else would fall into place. With this approach you can never be sure which is the more accurate and which is the colored component. Its possible that one component can be overcompensating for the deficiencies of another. I can see now this is not the way to go.

Some of the changes I've made to maximize differences between recordings include server and playback software, and cables. I stated in an earlier post the aspects of the recording I focus on are the spatial cues. The best way to describe the improvement would be to say recordings sound more like there are human beings behind the instruments (that is to say playing the instruments) as opposed to just hearing the instruments in space.

 

 

Hi Paul,

 

As tangential thought:

Not for comparison against other components or assessment of a system but there is one place I find using voice as a reference to be very useful. That is when setting playback level for maximum realism.

Long ago, the engineer Peter Walker said words to the effect that every recording has one "correct" (or perhaps "optimal") playback level.

 

My experience has convinced me he was correct as I've found the volume control can act as a realism "focus" control, where adjustment can pass to either side of maximum realism from a given recording. Too low or too loud and good as a recording might sound, it isn't as convincing as it might be at the "just right" level. For me, that is easiest to find by focusing on the vocals.

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

[Don Hills]

...

Long ago, the engineer Peter Walker said words to the effect that every recording has one "correct" (or perhaps "optimal") playback level. ...

 

I find that applies in many cases. Even for genres where the music is assembled by the mix engineer rather than recorded as ensemble, there is often a level that sounds right. This is often fairly loud, probably reflecting the level that the mixing and mastering was done at. This presents a problem when listening at a significantly different level. The classic "loudness" control was an attempt to partly compensate for this, but it's a non-trivial exercise to adjust the EQ to obtain a "natural" sound at different levels. To do it well requires some significant dynamic DSP processing. To my knowledge, only Audyssey have made a serious attempt to address this.

[bdiament]

I find that applies in many cases. Even for genres where the music is assembled by the mix engineer rather than recorded as ensemble, there is often a level that sounds right. This is often fairly loud, probably reflecting the level that the mixing and mastering was done at. This presents a problem when listening at a significantly different level. The classic "loudness" control was an attempt to partly compensate for this, but it's a non-trivial exercise to adjust the EQ to obtain a "natural" sound at different levels. To do it well requires some significant dynamic DSP processing. To my knowledge, only Audyssey have made a serious attempt to address this.

 

Hi Don,

 

I think this is very recording dependent and perhaps to a degree, listener dependent as well.

 

I find that in the overwhelming majority of cases, at least for my ears, if I set the lead vocal to play back at a level that is convincingly like what I'd expect it to be were I present at the performance, everything else falls into place and the overall is not very loud at all. Even with most "big" rock recordings, the vocals in typical studio productions tend to be pushed forward, so adjusting them to a realistic level will still keep the rest of the band down, relatively speaking.

 

I believe the results, either way, are unrelated to the level at which the recording is monitored during any stage of production. What the monitor level *can* do is influence the EQ the engineer might choose. But then, with the monitoring in most studios I've seen, there are much more pressing issues than where the level control is set. =8-0

 

Then again, as the saying goes, "Your mileage may vary." ;-}

 

Best regards,

Barry

Soundkeeper Recordings

The Soundkeeper | Audio, Music, Recording, Playback

Barry Diament Audio

[Don Hills]

Thanks for the view from your perspective, Barry.

 

From the "receiving" end, I've found that the level at which it sounds right also has to do with the reproduction system. I do much of my listening via headphones. For some time I used a set that were acknowledged to be slightly recessed in the vocal range. I found myself listening at lower than previous levels. This was the opposite to what I would expect - that I would boost the level until the vocals sounded "right". Instead, I found myself adjusting to get the overall balance correct.

[amir57bs]

good article , 

I recommend to test CD vs LP the same Album. for example CD album of "on every street" vs LP album of "on every street" (Dire Straits).

more contrast means less filter.

 

I should add the most important thing in this approach is loudspeaker quality.

if you have low sensitivity loudspeaker then the contrast is very low.