Testing audiophile claims and myths

[gregorio]

bigshot: "In music there are certain narrow bands that are very important... especially in the high end- consonants on vocals, high end on cymbals, etc. If there is an imbalance in the wrong place in the midrange, it can obliterate one of those narrow bands and the treble can be greatly affected." That is a myth being made up as I have underlined. Yet you seem to not care because the dude is in our camp. Or else you don't know how egregious of an error it is to say those things. Really this is a fantastical story bigshot told. If we don't correct that, then this is a Kangaroo court designed to chase off the subjectivists regardless of merit.

1. Actually, the first half of what you've quoted is entirely correct/factual. It's correct for several reasons: A. It's something that every recording engineer knows; it's taught at university and experienced routinely in practice and B. It's entirely supported by the science!
For example, there is indeed a "certain narrow band" (approx between 2.5kHz and 3.5kHz) which is very important for the comprehension of speech. Every engineer know this and so too does the science; as discovered by Fletcher/Munson, supported by anatomy, numerous subsequent studies, routinely implemented (noise-shaped dither for example) and routinely called the "critical hearing band". This is just one example and one which you yourself know because I've seen you use the term "critical hearing band"! Why then do you keep quoting/underlining/bolding it and calling it a myth?

2. The second part of your quote is rather speculative and maybe bigshot could have qualified it with "in my experience". It's speculative because he has not provided any evidence beyond a personal anecdote. HOWEVER, "Absence of evidence is NOT evidence of absence", a fallacy upon which you seem to repeatedly rely! Your analogy with "cable believers" is nonsense because it is not an analogous! In the case of "cable believers" we've got (A) an "absence of evidence" AND (B) an overwhelming amount of reliable "evidence of absence". "Myth", "fantastical" and particularly "egregious error" are emotive, potentially insulting and absolute adjectives, their use can ONLY be justified if we have both A and B together! So where is your overwhelming amount of reliable "evidence of absence"? You haven't presented even a single piece of reliable "evidence of absence", let alone an overwhelming amount! Without that, you've created your own "kangaroo court designed to chase off" someone who disagrees with your (apparently fallacy based) opinion, which makes your statement both irrational and hypocritical in the extreme!

I stated clearly a few posts ago that we agree the phenomenon is not masking.

I'm not so sure, I believe it could be masking or maybe more precisely; a phenomena caused by masking (as explained in post #6677).

G

 

[gregorio]

[1] When I was doing mastering of 78s for CD release, and old time transfer guy told me a similar trick. I asked him how to reduce high frequency noise without making the recording sound dull. He said that high frequency noise is important with severely band limited recordings, because it tricks the brain into thinking it's hearing high frequency sounds when it really isn't. He suggested noise reduction right up to the top edge of the recording, then add a low level constant high frequency hiss to the entire track. He explained that the biggest problem with noise is that it's random. Every time it comes in and out with clicks and crackle, you notice it. But if it's constant, you get accustomed to it and it makes it easier for your brain to be tricked into hearing frequencies that don't actually exist in the recording. Dynamic noise reduction tends to punch holes in the quiet parts, providing a sharp contrast. A little hiss smooths it all out. I created my own set of various hiss loops that had just enough random grain to them to sound organic, but not enough variation to attract attention to themselves.

[2] I couldn't avoid seeing his chart. Is he still talking about rooms?

1. No that's not really the same thing at all. What you're describing is absolutely still routine, standard practice in Film/TV, the use of "room tone" to even out (randomise) the differences in the noise which was recorded between different takes for example. Your described trick is somewhat similar to dither in a digital system. The trick I'm describing was essentially the synthesising of a sound which sounds the same (or close enough) as a cymbal (in a mix) and then adding that synthesised sound to the cymbal in order to reinforce it/make it stronger. The reason I quoted that trick is because the synthesising of that sound can be accomplished with (very) band limited white noise, demonstrating the somewhat random, white noise like quality of a cymbal's HF component.

2. Seems to be. I assumed you were just using the term "flat" as music/sound engineers often do, to mean "not explicitely EQ'ed" or relatively flat when talking about acoustics. What you've more recently described is likely to be extremely/extraordinarily flat (with the typical engineer's usage as applied to acoustics) but I'm not convinced it would be perfectly flat in an absolute sense, although probably close enough not to make any material difference. My described experience occurred in a well treated room with professional monitors but it wasn't particularly flat, even as the term is used by engineers. So assuming we are talking about the same perception/phenomena then I don't believe a flat or perfectly flat (in any sense of the word "flat") is a requirement, although a fairly well controlled listening environment might well be.

G

 

[amirm]

1. Actually, the first half of what you've quoted is entirely correct/factual. It's correct for several reasons: A. It's something that every recording engineer knows; it's taught at university and experienced routinely in practice and B. It's entirely supported by the science!
For example, there is indeed a "certain narrow band" (approx between 2.5kHz and 3.5kHz) which is very important for the comprehension of speech. Every engineer know this and so too does the science; as discovered by Fletcher/Munson, supported by anatomy, numerous subsequent studies, routinely implemented (noise-shaped dither for example) and routinely called the "critical hearing band". This is just one example and one which you yourself know because I've seen you use the term "critical hearing band"! Why then do you keep quoting/underlining/bolding it and calling it a myth?

Because I am not taking it out of context of what he said as you are. He was defending the conclusion of what he heard that changing mid-range, changes high-frequencies: "In music there are certain narrow bands that are very important... especially in the high end- consonants on vocals, high end on cymbals, etc. If there is an imbalance in the wrong place in the midrange, it can obliterate one of those narrow bands and the treble can be greatly affected. "

There is nothing in "science" that says if you mess with anything in mid-range frequencies of vocals, "the treble can be greatly affected."

Of course if you mess with content of what is considered vocal frequencies, you can diminish audibility. That wasn't the subject of discussion, nor anything disputed.

It is like me saying because it rains a lot in Seattle, pink elephants like to live there and you arguing that "the first half of what I said is right."

Bottom line is this: a ton of poorly implemented equalizers do not do what their dials says. Changing one band can easily change another set of frequencies. The way to know this is the problem or not is to run a simple sweep in there first, and make sure what is changed is what you think is changed. Only then do you want to go and invent new psychoacoustic effects as Bigshot attempted to do. Since this discussion was specifically about equalizers, this was even more important to know before citing stories of magical things equalizers can do.

 

[bigshot]

Gregorio, the way he was explaining it to me was that a balanced response was important because relatively small imbalances an octave below can cause masking in the octave above. I guess he had found the perfect recording and adjustment to reveal the effect. At the time, he was experimenting with methods of achieving flat output and quick ways to achieve a balanced response. He had just completed a prototype speaker and spent the previous couple of weeks just running tone sweeps and trying different strategies to get to where he wanted to be. The room would change with every venue he installed the speaker in, so he was just concerned with the output of the speaker. A lot of his work is with outdoor arenas so his interest is in creating highly directional sound so it doesn't bother the neighbors... particularly highly directional bass, because that is the biggest offender at outdoor venues in Southern California.

 

[gregorio]

[1] There is nothing in "science" that says if you mess with anything in mid-range frequencies of vocals, "the treble can be greatly affected."
[2] It is like me saying because it rains a lot in Seattle, pink elephants like to live there and you arguing that "the first half of what I said is right."

1. Correct. BUT there is ALSO "nothing in science that says if you mess with anything in the mid-range frequencies" it cannot affect the perception of high frequencies in some cases. So you are still basing your argument entirely on a common fallacy. Absence of evidence is not evidence of absence! Where is your evidence of absence?
2. Not it is NOT even remotely like you saying that, because we have BOTH an absence of evidence for the existence of pink elephants living in Seattle AND overwhelming evidence for the absence of pink elephants living in Seattle!

I accuse you of basing your argument on a common fallacy and of using an analogy which is completely inapplicable because it is not analogous, and how do you respond? You respond by using an argument based on exactly the same common fallacy and a different but equally non-analogous/inapplicable analogy! Impressive!!

Gregorio, the way he was explaining it to me was that a balanced response was important because relatively small imbalances an octave below can cause masking in the octave above.

To the best of my knowledge, that is impossible. Masking only occurs with frequencies which are close to the frequency causing the masking, in musical pitch terms (at a guess) probably somewhere around a semi-tone or so but certainly not an entire octave. This phenomena might be caused by masking as I've described; mid-freqs causing the masking of other mid-freqs which are the lower order harmonics of a cymbal and unmasking those lower order harmonics might allow the brain to extract more tonal information in the HF. In other words, this phenomena might be a consequence of masking but is not masking per se.

G

 

[KeithEmo]

It makes sense to me....

Virtually all masking effects depend on both the amplitude and frequency of both tones involved.
(Usually we say that a certain tone of a certain frequency will mask another tone of a certain frequency if it is a certain number of dB louder than the tone it's masking.)
Therefore, it seems perfectly obvious and logical that anything that affects the relative amplitudes of the two tones - for example a frequency response that is far from flat - will alter the masking effect.
(The masking effect is based on the relative amplitudes of the tones IN THE ROOM; if we alter that then we alter the masking effect.)

The same argument is very relevant for all of the lively discussions about the audibility of lossy encoding.
If I have a file that has been lossy encoded, and part of the lossy encoding process involved omitting a certain harmonic "because it was masked by a 500 Hz tone at -32 dB SPL"......
Then it is clearly going to affect the results if your system has a +5 dB bump at 500 Hz, while mine has a -5 dB dip at 500 Hz.
The "masking tone" will be at a different amplitude; so how well it masks some other nearby tone will also be different.
So, for ALL lossy perceptual coding, we should be specifying both the "flatness" and the "absolute amplitude - in dB SPL" at which the test was conducted.... because it WILL (and should be expected to) alter the results.

In fact, since the frequency response of our ears varies with amplitude, a given masking effect will vary depending on the dB SPL the test is conducted at.
And any valid discussion of masking effects REQUIRES a statement of the db SPL the test was conducted at, and either specific in-room SPL measurements of both sounds, or a statement of how flat the test system is.
(This might even explain, at least in part, why people with different ears, and different test systems, seem to report different results with various "lossy perceptual CODECs".... simply because the test conditions are NOT the same.)

Gregorio, the way he was explaining it to me was that a balanced response was important because relatively small imbalances an octave below can cause masking in the octave above. I guess he had found the perfect recording and adjustment to reveal the effect. At the time, he was experimenting with methods of achieving flat output and quick ways to achieve a balanced response. He had just completed a prototype speaker and spent the previous couple of weeks just running tone sweeps and trying different strategies to get to where he wanted to be. The room would change with every venue he installed the speaker in, so he was just concerned with the output of the speaker. A lot of his work is with outdoor arenas so his interest is in creating highly directional sound so it doesn't bother the neighbors... particularly highly directional bass, because that is the biggest offender at outdoor venues in Southern California.

 

[upstateguy]

Hey, No One Puts Baby In The Corner !

 

[bigshot]

I'm not exactly sure what bands he was adjusting, I just know they were remote from each other, and they sounded like they were a couple of octaves apart. I tried to google to find more specific info, but a lot of this stuff is over my head, and a lot of it is studies on guinea pigs and goldfish!

It might have something to do with "psychophysical tuning curves"... I don't know exactly what all this means, but this chart seems like it might be illustrating something like what I heard.

The masking frequency was somewhere in the low treble range (2kHz?) and the frequency being masked was in the treble range ssss around the cymbal (6kHz to 8kHz?). If this chart actually does represent what I was hearing, the masking band would have been there at the low point, and the masked frequency would have been right around this further two dots.

I remember somewhere reading that masking roughly followed octaves, but I can't find that right now. I'll see if I can google that up. Like I say, he was talking about masking and how it related to the balance of frequency response.

 

[amirm]

1. Correct. BUT there is ALSO "nothing in science that says if you mess with anything in the mid-range frequencies" it cannot affect the perception of high frequencies in some cases. So you are still basing your argument entirely on a common fallacy. Absence of evidence is not evidence of absence! Where is your evidence of absence?

That is a repeat of high-school debating tactic that was made earlier. My answer remains the same: pink elephants don't exist even though there is no research on their existence.

Here is a list of topics where there is zero research on them yet we say they can't be true:

1. USB cables sound different.

2. Power cables sound different.

4. Footers under electronics make them sound different.

5. Putting little dots on the walls makes the room sound better.

5. Fancy outlets and power cables make the sound better.

On and on. So be careful you don't give away the farm in an empty attempt to defend one of our brothers. The other camp can run with these things and bury us in our own hole!

That is beside the fact that a much simpler, backed by science/engineering explanation for what Bigshot heard: that the EQ was not properly designed and changing the mid-range changed the high-frequencies. To ignore the obvious and search for the unknown especially when Bigshot did not even bother once to verify the accuracy of what he thought was happening, is reckless and oozes biased thinking.

 

[amirm]

I'm not exactly sure what bands he was adjusting, I just know they were remote from each other, and they sounded like they were a couple of octaves apart. I tried to google to find more specific info, but a lot of this stuff is over my head, and a lot of it is studies on guinea pigs and goldfish!

Yet we were told to google this for ourselves....

 

[amirm]

It makes sense to me....

Virtually all masking effects depend on both the amplitude and frequency of both tones involved.
(Usually we say that a certain tone of a certain frequency will mask another tone of a certain frequency if it is a certain number of dB louder than the tone it's masking.)
Therefore, it seems perfectly obvious and logical that anything that affects the relative amplitudes of the two tones - for example a frequency response that is far from flat - will alter the masking effect.
(The masking effect is based on the relative amplitudes of the tones IN THE ROOM; if we alter that then we alter the masking effect.)

Masking is not the same as modifying the character and tone of frequencies as Bigshot said happened in the EQ experiment.

Even if it did, the effect in mid-frequencies has no chance of reaching the high-frequencies as he mentioned. If you read my article on perceptual effects of room reflections you see that we can approximate the auditory filter bandwidth using a measure called ERB. In doing so we get this kind of graph:

At 2500 Hz, the ERB is just a few hundred hertz. No way it is going to reach out > 10,000 Hz and do anything there.

Again, the simplest explanation here and a common problem in EQs is that they don't do what their dials say. Both analog and digital versions can easily bleed outside of the indicated band.

 

[castleofargh]

...
To the best of my knowledge, that is impossible. Masking only occurs with frequencies which are close to the frequency causing the masking, in musical pitch terms (at a guess) probably somewhere around a semi-tone or so but certainly not an entire octave. This phenomena might be caused by masking as I've described; mid-freqs causing the masking of other mid-freqs which are the lower order harmonics of a cymbal and unmasking those lower order harmonics might allow the brain to extract more tonal information in the HF. In other words, this phenomena might be a consequence of masking but is not masking per se.

G

I'm also thinking of something like that. or indeed an EQ spreading a little too far like Amirm suggested, although I'd like to think that an audio engineer doing such a demo would know his tools. but without much data, it's all conjecture anyway.

 

[KeithEmo]

I just Googled "pink elephant" ........ and I got back hundreds of thousands of hits. Some of them looked to be Photoshopped, but a lot of them looked quite authentic to me. They included quite a few pink cement elephants, lots of pink stuffed elephants, several that looked to be plastic, and a massive number of pink cartoon elephants. There were even several live pink elephants - and, although most of them looked to be Photoshopped, several actually looked to me like a real live elephant who had an interaction with a can of spray paint.

And, even stranger, here's a news item that purports to include a picture of a real live naturally occurring pink elephant.......

https://www.catersnews.com/stories/...arade-meet-the-adorable-pink-albino-elephant/



The sound made by cymbals covers a very wide range of frequencies.... so it makes perfect sense to me that any sound that masks specific frequencies may make PART of the range covered by the cymbals less prominent, and so alter the overall tonal balance. I've also read many accounts of mixing engineers doing what seem to be odd things to manipulate the way cymbals sound (for example, mixing in some band-limited white noise to help "keep the cymbals from getting buried in the mix"). Therefore, to me, that puts cymbals firmly into the category of "things that sometimes exhibit unexpected interactions with other components of the recording". Therefore, while I haven't specifically heard of the example BigShot described, I wouldn't automatically assume that it doesn't exist either. Perhaps, when the primary frequencies of the cymbal are partly masked, our minds tend to interpret the harmonics as noise, and so "not notice the cymbals"..... I don't know if it's true - but it's certainly not unreasonable - and I sure don't know that it ISN'T true.

Perhaps, if we asked NICELY, someone might even come up with a sample recording to prove that it does.....

Yet we were told to google this for ourselves....

 

[amirm]

The sound made by cymbals covers a very wide range of frequencies.... so it makes perfect sense to me that any sound that masks specific frequencies may make PART of the range covered by the cymbals less prominent, and so alter the overall tonal balance.

Let's review again what was said:

I have a friend who is a sound engineer who gave me a vivid example of it once. He stood to block what he was doing on the equalizer and asked me to listen and tell him what I heard. I listened and I heard the high end of the cymbals going in and out. Muffled, sharp, muffled, sharp. I told him the treble was going in and out. He turned away from the equalizer and showed me the band he was adjusting. It was a midrange frequency, and the adjustment he was making in the midrange was quite small and almost inaudible.

So an almost inaudible change to mid-range was causing the *high-end* of the cymbals to cut in and out.

You still think you are considering his example in your mind?

He is trying to say a tiny change in a completely independent spectrum of music has an effect on another. Nothing in masking works this way. Let's not keep imagining scenarios for what is likely caused by other obvious causes.

 

[bigshot]