"Things that cannot be measured"

[SIY]

So what do you make of those results?

The old frog in boiling water trope reified.

 

[j_j]

I don't know about all of that. I've done a little recording. Two microphone vs multi mike really is one of those things where you don't need to blind anyone as there is enough difference. I have done that test with people blinded however with level matching (which is a sticky wicket all on its own), and letting people decide which they like. Yes, always preferred the multi-mike version. I don't, but then even blind you always know which is which you are listening to. Now I've never tried to get two versions to sound the same and compare, but then again why would you?

The other test I've done is uncompressed with a little compression. A little compression with a little more compression. A little more compression with even a little more compression and then even a little more vs even more compression. Done this with both two mic and multi mike recordings. Compression was preferred every step of the way by everyone I had listening. The final step was the highest level of compression vs none. Everyone wrinkled their noses at how bad one of them sounded and preferred no compression. My highest level of compression was lower than is your typical squashed commercial recording. No limiting for instance. So what do you make of those results?

I have done EXACTLY that same demonstration time and time again.

Yet it's ignored, and we hear "LOUD IS GOOD LOUD IS GOOD". (as an aesthetic choice, it's fine, but for everything, no) I have been pointing out since the early 1990's exactly just how a little bit louder is "always better" but 10 steps is almost always universally hated, even though each individual step is better.

And the beat goes on.

 

[solderdude]

Almost certainly cognitive bias is the reason you perceive 'great sound' from such recordings.

On headphones the 'stones in my passway' recording sounds great to me. I don't care if its cognitive bias or not and if it is then I am glad that works for me.
It has to be played a bit loud though due to the dynamic range and brings a 'live-feel' for me.
Measuring/seeing 108dB peaks, when I dial down the volume to 102dB peaks it still 'works' for me but below that it looses 'charm' (DR around 13)

Quite a few studio recordings also sound great to me, where compression is used correctly. Don't care if its cognitive bias or not either.

'Simple is best' is a very common and understandable bias, but it should be tested and, in this case, it fails.

Simple certainly is not always best and the idea that 'simple is always best' is a typical audiophile myth been proven to be incorrect in most cases.
Carmen Gomes recordings may be a fail to you but there is taste and preference too so what may be a fail to you or some others may not be so for someone else.

 

[Newman]

I don't know about all of that. I've done a little recording. Two microphone vs multi mike really is one of those things where you don't need to blind anyone as there is enough difference. I have done that test with people blinded however with level matching (which is a sticky wicket all on its own), and letting people decide which they like. Yes, always preferred the multi-mike version. I don't, but then even blind you always know which is which you are listening to. Now I've never tried to get two versions to sound the same and compare, but then again why would you?

It is significantly wrong to assume that, as soon as the difference in two things being compared is easily audible, you don't need to do it blinded anymore. In fact, you need to do it blind just as much as ever.

Also, when conducting a blind test, one absolutely must do it in a way such that the subjects don't know what they are listening to. So, if you think that everybody there will immediately know that they are listening to single mic versus multi mic and will immediately be able to tell which is which, then you need to re-design the test, for example by not telling the subjects what are the two things being compared. Just say for example the gear has been changed (which is true in a way).

The other test I've done is uncompressed with a little compression. A little compression with a little more compression. A little more compression with even a little more compression and then even a little more vs even more compression. Done this with both two mic and multi mike recordings. Compression was preferred every step of the way by everyone I had listening. The final step was the highest level of compression vs none. Everyone wrinkled their noses at how bad one of them sounded and preferred no compression. My highest level of compression was lower than is your typical squashed commercial recording. No limiting for instance. So what do you make of those results?

I don't understand your test methodology well enough to know whether or not I should make anything at all of the results. However, as a general topic, compression is best applied via multiband compression systems, where you can adjust the amount of compression both on each mic feed and on each frequency band of the mixed result. Therefore it is obviously a very sophisticated question, and I don't think a simplified test method tells us much. You do compression well, and it will probably be preferred to not having any, because you did it well: you do it badly and it will be better if you did nothing at all. Self evident.

 

[Blumlein 88]

It is significantly wrong to assume that, as soon as the difference in two things being compared is easily audible, you don't need to do it blinded anymore. In fact, you need to do it blind just as much as ever.

Yes, and it was blinded. But the difference was clearly obvious. Like testing if one can hear the difference between 100 hz and 800 hz. You can test it blind, and get 100% results. In the case of multi vs stereo miking, there is no way to hide the difference. So it became a blind preference test. Everyone preferred multi miking. It was presented as here are two different versions of the same recording, which do you prefer.

Also, when conducting a blind test, one absolutely must do it in a way such that the subjects don't know what they are listening to. So, if you think that everybody there will immediately know that they are listening to single mic versus multi mic and will immediately be able to tell which is which, then you need to re-design the test, for example by not telling the subjects what are the two things being compared. Just say for example the gear has been changed (which is true in a way).


I don't understand your test methodology well enough to know whether or not I should make anything at all of the results. However, as a general topic, compression is best applied via multiband compression systems, where you can adjust the amount of compression both on each mic feed and on each frequency band of the mixed result. Therefore it is obviously a very sophisticated question, and I don't think a simplified test method tells us much. You do compression well, and it will probably be preferred to not having any, because you did it well: you do it badly and it will be better if you did nothing at all. Self evident.

See J_J's response above regarding compression. In this case, they were the same recordings processed differently. That is all the listeners knew. Not what processing, and they were only asked to pick a preference. Done this more than once with more than one group of listeners. A little more compression is always preferred to a little less. Then at some point quite a bit is not preferred to none. I think this is how studio people will often defend to the death that louder mixes and masterings are better. They do shoot outs for themselves or with the band. Every small bit louder always wins. But do they try only a little vs a lot? I'm sure some do, and I think many do not.

In one case I was doing simple across the board compression. In some others I was doing multi-band on each track then a slight bit more on the end. A little more is always preferred.

 

[j_j]

I don't understand your test methodology well enough to know whether or not I should make anything at all of the results. However, as a general topic, compression is best applied via multiband compression systems, where you can adjust the amount of compression both on each mic feed and on each frequency band of the mixed result. Therefore it is obviously a very sophisticated question, and I don't think a simplified test method tells us much. You do compression well, and it will probably be preferred to not having any, because you did it well: you do it badly and it will be better if you did nothing at all. Self evident.

Really. Please explain just what missing controls or confounding issues are present, then?

The test is simple, uncomplicated, has been done in two different ways, using either "just level" or "adding more and more compression", and the results hold perfectly well.

What you should make of the results is simple: If you turn it up a bit, you're likely to think it sounds better. If you turn it up a bit more, and compare it to the one that's turned up once, again, the new one will sound better. The problem comes after several steps of this sort when you compare the final to the original, and lo and behold, now the original sounds better.

You can not assume that pair-rankings in a subjective test of an effect are transitive. It's that simple. It's trivially easy to fool yourself.

 

[Newman]

Really. Please explain just what missing controls or confounding issues are present, then?

The test is simple, uncomplicated, has been done in two different ways, using either "just level" or "adding more and more compression", and the results hold perfectly well.

What you should make of the results is simple: If you turn it up a bit, you're likely to think it sounds better. If you turn it up a bit more, and compare it to the one that's turned up once, again, the new one will sound better. The problem comes after several steps of this sort when you compare the final to the original, and lo and behold, now the original sounds better.

You can not assume that pair-rankings in a subjective test of an effect are transitive. It's that simple. It's trivially easy to fool yourself.

I was talking to Blumlein and I don't know what it is that you took exception to. He described a test and I simply said I don't know too much about that test and all its methodological details, so I can't comment. What is the problem? Are you asking me to assume that the test was conducted perfectly, in the manner of a competent professional audio researcher? If so, why are you asking me to assume that?

 

[Newman]

See J_J's response above regarding compression. In this case, they were the same recordings processed differently. That is all the listeners knew. Not what processing, and they were only asked to pick a preference. Done this more than once with more than one group of listeners. A little more compression is always preferred to a little less. Then at some point quite a bit is not preferred to none. I think this is how studio people will often defend to the death that louder mixes and masterings are better. They do shoot outs for themselves or with the band. Every small bit louder always wins. But do they try only a little vs a lot? I'm sure some do, and I think many do not.

In one case I was doing simple across the board compression. In some others I was doing multi-band on each track then a slight bit more on the end. A little more is always preferred.

I don't take issue with any of this, and it doesn't seem to be at odds with anything I wrote, either. Are we violently agreeing?

I was making a few comments about single microphone versus multiple microphone recording, and you responded to that and then jumped across and brought up the subject of dynamic compression, which I had not been discussing. I'm not sure if it was related, or if you simply wanted to bring up a second topic? But I certainly was not discussing dynamic compression in any way shape or form, so I don't know why I am suddenly in a debate about it.

 

[Blumlein 88]

I don't take issue with any of this, and it doesn't seem to be at odds with anything I wrote, either. Are we violently agreeing?

I was making a few comments about single microphone versus multiple microphone recording, and you responded to that and then jumped across and brought up the subject of dynamic compression, which I had not been discussing. I'm not sure if it was related, or if you simply wanted to bring up a second topic? But I certainly was not discussing dynamic compression in any way shape or form, so I don't know why I am suddenly in a debate about it.

You were implying the test wasn't blind or at least said I cannot assume something doesn't need to be blind. I know that. While I thought it was clear, it was blind, but was one of those differences where the difference was enough blinding didn't matter. This in regard to stereo vs multi-miking. You also said subjects should not know what they were comparing that they shouldn't be told it was stereo vs multi-miking. They weren't. I thought that much was clear.

There are such differences. I wouldn't bother to test some tracks with a roll off starting at 250 hz on the low end with a full range track on full range speakers. No need to test for that.

Then yes, I switched to compression levels. You said compression done well will likely be preferred. My point was more compression in small steps done well or not is always preferred. Until a whole lot (too much?) is compared to very little or none, and suddenly it isn't preferred. So many commercial recordings are compressed way beyond the pale, and I bet plenty of them ended up that way in small steps with each louder step being preferred.

 

[Newman]

You were implying the test wasn't blind or at least said I cannot assume something doesn't need to be blind. I know that. While I thought it was clear, it was blind,

OK, then we are violently agreeing. What I was actually saying is that I didn't have enough detail to assume it was perfect. Hey: you got a result that confirms my claim, so it would have been really easy for me to 'overlook' that. I was just being honest.

Can we please move on?

 

[Blumlein 88]

OK, then we are violently agreeing. What I was actually saying is that I didn't have enough detail to assume it was perfect. Hey: you got a result that confirms my claim, so it would have been really easy for me to 'overlook' that. I was just being honest.

Can we please move on?

Sure no problem.