Testing audiophile claims and myths

[amirm]

Yes, it's unlikely it was flat, but as the Genelec graphs show, not one of them is really "flat" so it's a question of degree.

I want to make sure this important point about the measurements I showed in the Genelec paper is not lost:

As you see, they used 1/3 octave smoothing. That is the wrong-headed system used to calibrate movie soundtrack recording studios to Dolby standards. It completely smoothes out peaks and valleys in low frequencies, making the response far more flat than it is. For bass region, smoothing should be limited to 1/12 octave.

This was the subject of a paper I wrote for Widescreen Review Magazine on this very topic. See https://audiosciencereview.com/forum/index.php?threads/target-room-response-and-cinema-x-curve.10/
And the deeper dive on "X-curve" here: https://audiosciencereview.com/forum/index.php?threads/validity-of-x-curve-for-cinema-sound.204/

 

[amirm]

I remembered a very simplified video and google was nice enough to show it to me immediately. watching it again, it's even more simplified than I remembered and despite him mentioning it, those really interested in this should go look for a paper or 2 on the subject. but as a starting point, it gets the main ideas across very clearly IMO.

You are undeselling that video. That is most excellent and best overview of frequency masking I have seen. Definitely must watch.

I should not that there is another kind of masking: temporal masking. That is responsible for such things as IEC weighting of WoW and Flutter in the analog days:

As we see there is a peak our hearing at around 4 Hz. Inverting this, we get 0.25 seconds. This is one of the reasons you want your room reverberation time to not be much above 0.25 as it can affect speech intelligibility.

On the topic of EQ, if you have room resonances, the peaks can cause time domain ringing. This ringing can obscure other sounds well above bass frequencies. A parametric EQ applied to these bass frequencies can indeed cause the higher frequencies to sound better even though they remain unchanged by the EQ. Reducing those peaks reduces rigning and with it, the temporal masking.

 

[castleofargh]

maybe I'm indeed underselling it. it's just that it would be bad to have people assume those straight lines are a fair representation of the masking area, as they discount how in reality as it always is with human hearing, nothing is that linear. and also there is the issue of how the masking area tends to impact the frequencies above the tone more than the frequencies below it. and indeed there is also temporal masking to consider even more so when we're talking about room and speakers.
but the way he explains masking is very clear indeed, which is probably why I remembered it so vividly.

 

[pinnahertz]

Yet he is insisting in post after post that to hear the same effect, the room must be "flat."

And you are insisting, in post after oost, that it can't be. Which of you knows the room and was there?

That he doesn't know there is no such thing as a "perfectly flat" means is a crime in its own, no?

There are many definitions of "flat". Remember when +/- 3dB was "flat" for electronic devices? And yet +/- .05dB is not actually perfectly flat. "Flat" is a subjective term. "Perfectly" is an adjective on a subjective term. That makes the entire term still subjective.

Bottom line is what I said: he heard and EQ where if you modified the mid-range, the highs changed. This is easily characteristic of poorly designed equalizers. It could have low Q, distortion, etc.

Perhaps, but you don't know what he heard.

It certainly is NOT the type of EQ we encourage people to use. The best and main use of EQ in acoustics is to reduce resonances in low frequencies. The modal bandwidth there can be as low as half a hertz! As such, you need an ultra-narrow, high-Q parametric EQ to correct those. No way, no how you want to use the poorly made EQ bigshot is talking about.

I agree with the application of high Q parametric to mediate modal response, but it's not a cure of course. I'm still not convinced you know what he's talking about.

What to make of the fact that he has never tried to duplicate that results? Not in his system or for us here? It is all OK to say this is a scientific "fact" and we should just accept it??? Is that how it works?

I accept his comment as what it is: anecdotal. It's not scientific, and doubtful it was intended to be.

If we are going to accept any and all anecdotal stories, what sets us apart from subjectivists? You are picking on any and all people on the other side but want to give a pass to Bigshot because he is on our camp? Where does that leave our credibility?

Our (and your) credibility is secure. Our posts (usually) deal with substantiated scientific fact. I'm not challenging that. And one anecdotal post more or less doesn't change scientific fact. However, as a scientist, we should also recognize where lack of data inhibits valid conclusion, and where drawing conclusion without sufficient data is just bad science. In this instance we lack data, and it doesn't further the scientific cause to presume then conclude. Nothing I've suggested validates the anecdote, but your attempts to invalidate it without complete data make scientists look egotistical and silly.

Bottom line is that you seem to be defending him just because. That is not helpful. What is helpful is to explain in plain language that what he describe has no relevance, includes serious mistake about room conditions even in best rooms, and has nothing whatsoever to do with "auditory masking."

[
You haven't helped, though. I'm defending someone's right to post an anecdote. You can point out where data is missing, and point out the possibility that the observations were in error, but when you proclaim the observations false and draw hard conclusions that invalidate the observation without the ability to perform a complete analysis, it's just your opinion against his. That's not scientific, helpful, or worth while. It comes off as an outburst of ego. Many people turn a deaf ear to egomaniacs, regardless of if they are right or wrong.

Now, if we could collect more data, go to the room, duplicate the situation, make some observations of our own, make some measurements if possible, we might be able to state with authority what was going on and explain what was heard. If we can't do that, then we can explain why we can't be conclusive, and express doubt. We cannot state anything definitively, and all of this is just so many characters on the page.

 

[castleofargh]

someone said something, someone else is skeptical and asking for evidence. nothing out of the ordinary. IMO bigshot was just providing an anecdote and as such wasn't asking the all planet to instantly accept it as law.
as for what defines a flat setup, we all here know that we're not getting a flat at +/-1dB in the entire audible range at our listening position, no matter how expensive and calibrated our gear is. I personally took the statement of flat as meaning it was properly calibrated. so much closer to flat than uncalibrated stuff. nothing more. the rest is clearly a matter of how much content was to be found in the EQed area, and to make any sort of difference, I would assume it was fairly significant, or indeed there was a technical issue or an EQ a little too wide. but it's not like it contradicts the principle of masking. as for how that should feel, well it's an all different can of worms ^_^.

 

[pinnahertz]

I want to make sure this important point about the measurements I showed in the Genelec paper is not lost:

As you see, they used 1/3 octave smoothing. That is the wrong-headed system used to calibrate movie soundtrack recording studios to Dolby standards. It completely smoothes out peaks and valleys in low frequencies, making the response far more flat than it is. For bass region, smoothing should be limited to 1/12 octave.

Right, but then you also need spatial averaging, or you will chase your tail with the microscopic, single location excursions that are not actually audible.

What's the point? That the room wasn't perfectly flat, and no room is? We got it already.

 

[bigshot]

 

[amirm]

And you are insisting, in post after oost, that it can't be. Which of you knows the room and was there?

You once again insist that if someone says they saw a pink elephant, we can't have an argument about that being impossible. Bigshot said and I am quoting again: "You have to have a perfectly calibrated flat playback response for this demonstration to be clear. "

That is a made up statement. He didn't measure the room. He is putting that sauce on top of the meal hoping it will make it palatable.

If you are going to jump on the throat of any subjectivists comment about what they hear, you need to show the same here.

Nothing destroys audio science more than poor advocates of it.

When someone says there is a magical EQ that when you change midrange the high frequencies change, you need to speak up. When he says that is only heard in a "perfectly flat" response room, you need to doubly speak up. When he then says auditory masking is in play, then you need to speak up. This is all made up nonsense that has no place in discussion of audio science much less in a thread about audio myths. Or is it that the we are only interested in myths that subjectivists have and that we are so superior as to never be guilty of it?

I mean how wrong can a self-appointed objectivists on a forum can be before we speak up against it? I can't think of more egregious mistakes, myths and just plain nonsense than the above. Yet you are arguing with me about it? To what end?

 

[amirm]

Right, but then you also need spatial averaging, or you will chase your tail with the microscopic, single location excursions that are not actually audible.

Not really but that is for another topic.

What's the point? That the room wasn't perfectly flat, and no room is? We got it already.

The point is that the Genelec graph that shows variations is actually much worse that it even looked. You seem to be saying that what was shown was close enough to "flat" and I wanted to make sure you knew the measurements were highly filtered. Reality is much worse and indeed there is an effort to ditch the whole X-curve thing in SMPTE due to usage of 1/3 octave measurement/filtering.

But sure, if you now agree that the room Bigshot heard was not flat, then why is he saying we won't hear the effect because our rooms aren't? He should be able to replicate the experiment for us.

 

[amirm]

You haven't helped, though. I'm defending someone's right to post an anecdote.

Where would I find you doing that for subjectivists?

And how was that story an anecdotes when members were schooled on googling auditory masking?

I'm taking it one step at a time... Jnorris, have you googled "frequency masking" yet? What did you find? See the post just below this one if you want an example of what it means.

Our kind moderator post this great wisdom yesterday:

to EQ or not isn't the matter. you do whatever you like and listen to music however you enjoy it most. nobody but bigots would try to force you to do what you don't want to do with your music.
on the other hand when you present EQ as being wrong in general and claim it to be factual, now that's not something anybody should let pass. because this time you're pushing your reality onto others so it better be correct.

We were presented with an EQ experiment with conclusions that are flat wrong. And tied to audio science no less after the fact. So as mentioned, I am not giving him a pass. You seem to want to do that in a partisan way and I think that soils the reputation of our camp as a whole.

 

[jnorris]

Good Grief! Do you people ever stop your knob twiddling, measuring, adjusting, arguing, soap-boxing, pontificating and bloviating long enough to listen to some MUSIC???? Does anyone even OWN an equalizer that will do 1/12 octaves?

 

[pinnahertz]

You once again insist that if someone says they saw a pink elephant, we can't have an argument about that being impossible. Bigshot said and I am quoting again: "You have to have a perfectly calibrated flat playback response for this demonstration to be clear. "

!. That is a made up statement. He didn't measure the room. He is putting that sauce on top of the meal hoping it will make it palatable.

2. If you are going to jump on the throat of any subjectivists comment about what they hear, you need to show the same here.

3. Nothing destroys audio science more than poor advocates of it.

4a. When someone says there is a magical EQ that when you change midrange the high frequencies change, you need to speak up. 4b. When he says that is only heard in a "perfectly flat" response room, you need to doubly speak up. 4c. When he then says auditory masking is in play, then you need to speak up. 4e. This is all made up nonsense that has no place in discussion of audio science much less in a thread about audio myths. Or is it that the we are only interested in myths that subjectivists have and that we are so superior as to never be guilty of it?

I mean how wrong can a self-appointed objectivists on a forum can be before we speak up against it? I can't think of more egregious mistakes, myths and just plain nonsense than the above. Yet you are arguing with me about it? To what end?

Well, I see I aroused the beast.

1. I don't think the experience was made up. I think the anecdote lacks sufficient information for proper analysis. Can we at least agree that "perfectly calibrated flat playback response" is open to interpretation, since there is no such thing? If so, then can we agree that the statement provides no information as to what "perfectly calibrated flat playback response" means?

2. I will decide when, and on whom to jump on, thank you.

3. I disagree. Nothing destroys audio science more than myth. Myth is generated in several ways. I've found that, regardless of how much science I apply, my efforts to bust myth have been largely futile. But in this case we don't have myth, we have an observation without sufficient detail. Know the difference.

4a. Let's go back to the original post in question:

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.

Anything wrong there, outside of a complete lack of technical detail? Do you not know that you actually can adjust an equalizer in one area of the spectrum and end up with a subjective change in the other with good quality equalizers? Sure you can! I don't know about his example, there's not enough information to replicate it, but the general principle is well known. So I chose not to discount it on face value, it's just an anecdote with no detail.

Then, he added this:

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. Treble is a big part of what people describe as "detail".

Anything wrong there? I would be drawn to "narrow bands" because bandwidth is not specified, and that causes a problem because we don't know what is meant by "narrow". Is it 1 octave? 1/2 octave? Less? It's not wrong, it's just not stated. But if we ignore "narrow" the rest of the statement is actually fine, if again, missing a whole lot of detail.
Then this:

If you have an equalizer, try to isolate the frequencies of the sisss sound of the cymbal and then adjust a narrow spike one octave below it. You'll find the spot. It's really quite amazing. Not at all intuitive.

Ok, still not enough data, but at least an invitation to experiment. Have you tried this? For real?

Then, unfortunately, this (the Edit is the unfortunate part):

I'd be more motivated to go to the trouble of making examples to prove things to you if you had honestly answered our questions about whether you kept the volume at normal listening level at all times when you "heard" the noise floor of 16 bit... So I'll just suggest that you take a solo violin recording, figure out where the important upper sheen is, calculate one and two octaves below that, and try applying some notch filters.

At least I tell you how it's done.

* EDIT: I just thought of something that is important. You have to have a perfectly calibrated flat playback response for this demonstration to be clear. If you have imbalances in around the frequencies you're working with, or in even octaves above or below them, the effect can get muddled.

If we take out the word "perfectly" and "flat", it starts to make sense though. And since we all know there is no actual "perfectly flat", we pretty much have to take that out just to make it make sense. Do you not agree that trying to EQ in a monitoring environment with poor response would, at very least, yield poor results? Would you disagree that at least basic room cal to a reasonably flat target curve is essential for good judgement when applying EQ? I believe that is the reality, it was just expressed in an exaggerated, and unrealistic way.

That's why I don't specifically take exception to his post. I recognize it as anecdote, and with a little understanding (he's not an engineer, or calibrator, he works in film sound production), and run the comments through that filter, his observation seems, at very least, to merit further investigation. I see no need to myth-bust the post, and yet I hold science in sound to the highest esteem. I would welcome the opportunity to investigate the observation.

But you go and accuse his friend of using a poor equalizer that is so sloppy as to have a reciprocal gain effect outside of its primary center frequency. Seriously, when was the last time you saw that happen on a mixing desk? Last time I was even aware of the phenomenon it was an ancient Baxandall tone control on a hifi device, and that problem was engineered out years ago. And then you pick on him for saying "perfectly flat". Come on, man, that's now just looking for a nit to pick. He later detailed that the room was well calibrated, and that the engineer in question had years of calibration experience. I might still have questions about that, like is he still using 1/3 octave EQ, or is he into PEQ, or perhaps something even better, like an FIR filter? And what is his personal tolerance, target curve, etc.? Well, we don't know, do we? Do you think all of this silliness is going to somehow magically generate more data for you to chew on? I doubt it. Or is it really just that all-fire important for the big dog to bark?

And then, clearly out of desperation, you decide to pick on me for somehow loosing my calling as a sound science professional. To that, I take strong exception, not that it will matter to you.

I think someone needs to take a chill pill. It might be me, or it might be someone else. I'm betting it's not me.

 

[pinnahertz]

1. Good Grief! Do you people ever stop your knob twiddling, measuring, adjusting, arguing, soap-boxing, pontificating and bloviating long enough to listen to some MUSIC???? 2. Does anyone even OWN an equalizer that will do 1/12 octaves?

1. I don't believe the two are mutually exclusive. I do both every day, thanks.
2. Yes, several. And they can be set to much higher Q than 1/12th, with precision Q, frequency and gain, multiple bands. Not even expensive. You could get one, if you wanted, for less than a garden variety power amp.

 

[bigshot]

This conversation is ridiculous. I'm reminded of the Brobnigagians arguing over which side of an egg to break.

Making exists. It's possible to demonstrate it in music if you find something with energy across several upper octaves, like large cymbals and probably violins. Room acoustics have nothing to do with what I was talking about. Feel free to go try the experiment for yourself. Calibrate your system tightly or sloppily. I don't care. I'm with jnorris on this one.

 

[castleofargh]

Good Grief! Do you people ever stop your knob twiddling, measuring, adjusting, arguing, soap-boxing, pontificating and bloviating long enough to listen to some MUSIC???? Does anyone even OWN an equalizer that will do 1/12 octaves?

what is this "MUSIC" you're mentioning? is it a new kind of sine sweep?
I use DMG Equilibrium as go to EQ for my headphones/IEMs. the Q value goes to 50 so I'm in. yeah!!!! did I win? is there an amateur category? because it's a little unfair to be lined up against the pros.