Understanding Audio Frequency Response & Psychoacoustics (Video)

[solderdude]

Surprising since that was almost the whole message of the video.

Perhaps because what Amir talked about in the video is already known to me. So some short 'touching' of the subject didn't throw me off.

So if i understood correctly, in layman terms, humans are more likely to tell if the tonality is wrong if the peak/dip is in the high frequency, but they won't be able to tell (as easily) at which exact frequency range the dip/peak is happening.

When tonality is off it is off, this can be in the lows, mids, highs etc. It all can be heard.
Funny thing is that, especially when more than 1 thing is off, most people have no idea how to describe this or what to do about it.
That's probably why people love to get info on how to EQ headphones.
Those that have paid attention to post from @sweetchaos will have noticed that there is no such thing. Amir's, Oratory and Jaakko's ready made EQ differ like night and day with often differences exceeding 5dB (sometimes even 10dB). They can't all be right. Certainly when all of them are based on measurements on a similar rig.
I tend to agree with Amir. Only correct for the big ones with simple EQ. It is more likely to be correct.

 

[Robbo99999]

So if i understood correctly, in layman terms, humans are more likely to tell if the tonality is wrong if the peak/dip is in the high frequency, but they won't be able to tell (as easily) at which exact frequency range the dip/peak is happening.

However in lower frequencies the change in tonality wouldn't be as 'annoying' as in higher frequencies, but trained listeners will find an easier time telling at which exact frequency range the problem exists.

I reckon it would be pretty hard to tell exactly (perhaps not even roughly) where the dips/peaks were unless you were measuring them, so in terms of EQ'ing you'd be relying on the measurement, making EQ changes to cut peaks or boost dips, then you'd listen again to see if it's an improvement......kinda immaterial on being able to blindly recognise (without a measurement) where precisely the dip or peak is on the frequency response.

 

[ernestcarl]

Amir's, Oratory and Jaakko's ready made EQ differ like night and day with often differences exceeding 5dB (sometimes even 10dB).

I haven’t looked at either EQ profiles closely, but I’m very surprised the differences can be that big. Last pre-made EQ profile I used was Amir’s for the HD650 and it sounded okay to me in comparison to my other stock EQ profile where I personally preferred a tad less boost added in the bass due to the “woolliness” quality to it.

Either way, I don’t use headphones enough to make a judgment over which pre-made EQ profiles I find more “correct”.

 

[Dzhaughn]

Amir does good job in explaining how to read frequency response graphs and how the graph corresponds to our hearing perception. Unfortunately Amir made one big mistake in claiming that peaks and dips matter more in the bass range than in treble range when the opposite is true. This wouldn't be a big deal if he wasn't such an authoritative figure in audio for many. In this video Amir is spreading misinformation and most people don't question what he says and now there are who knows how many people who think that peaks and dips matter more in low frequencies when the truth is that they matter more in high frequencies. Pointing out this was what my first response was about. I wrote my second response because some people in this thread got confused about the logarithmic x-axis issue, which is essential in understanding why peaks and dips in frequency response graphs matter more in high frequencies. My intention was not to divert the discussion to away from the topic in Amir's video but to point out the mistake and then after that clarify some points which honestly I should have done already in my first response.

Here is another visualization of the filter widths in case the problem is still unclear to some readers. These represent rectangular filters at 9 arbitrary frequencies. See how the filters get narrower towards the higher frequencies. That is because x-axis is logarithmic in this plot, just like it is typically in frequency response graphs. The first filter has more room for dips and peaks which will be averaged by our hearing than the last one meaning that small peaks and dips in the high frequencies are more audible than in low frequencies.
View attachment 125942

Ok, at the root of the issue is Amir's rule-of-thumb claim "Only wide parametric equalizations are audible in high frequencies" where "wide" means "looks wider on this logarithmic frequency response graph." That's a straightforward (if not easy) empirical question.

What you (Jaakkopasanen) are saying here is that the ERB filter bandwidth growth can not explain that, even supposing it is true. That's because the ERB bands get wider but at a sub-exponential rate, so the (linear) width on the frequency response graph is still smaller at high frequencies even as the (numerical) width increases.

Is that it?

 

[solderdude]

Ok, at the root of the issue is Amir's rule-of-thumb claim "Only wide parametric equalizations are audible in high frequencies" where "wide" means "looks wider on this logarithmic frequency response graph." That's a straightforward (if not easy) empirical question.

3kHz BW at 20kHz is not wide.
3kHz BW at 4kHz is wide.

50Hz BW at 100Hz is quite wide, 50Hz BW at 20kHz would be super narrow. That was what Amir tried to explain.

 

[Dzhaughn]

3kHz BW at 20kHz is not wide.
3kHz BW at 4kHz is wide.

50Hz BW at 100Hz is quite wide, 50Hz BW at 20kHz would be super narrow. That was what Amir tried to explain.

I don't agree that is what he is saying. Look at 7:40, he is pointing at a graph, and saying "your ERB is this wide." And to paraphrase "This little wiggle" does not matter perceptually while "the same wiggle over here" doesn't.

(Note please I am agnostic on whether it matters perceptually, although I tend to take Amir's word for that. But the ERB width is not as what explains this, because it is not getting wider.)

 

[solderdude]

It's a log scale so in absolute sense it is wider at higher frequencies when looking at Hz. Not as a 'band'

 

[audiofooled]

For people who would like to dig a bit deeper, anywhere you click you're in more deep...

http://hyperphysics.phy-astr.gsu.edu/hbase/Sound/ear.html#c1

http://hyperphysics.phy-astr.gsu.edu/hbase/Sound/hearcon.html#c1

http://hyperphysics.phy-astr.gsu.edu/hbase/Sound/loud.html

http://hyperphysics.phy-astr.gsu.edu/hbase/Sound/soucon.html

Most of the graphics are interactive and full of useful links. Great source imo.

 

[Andrej]

I am sorry but I have to stand in the defense @amirm .

Even though he may had best intention in mind, in my humble opinion @jaakkopasanen is the one who has created a confusion here. I must say that he did a brilliant job in explaining how we should interpret what we see on the graphs when we want to apply EQ and we should all thank him for this, but I think he actually did no favor to Amir's point of view in the video.

@amirm did a great job in explaining what we actually hear vs. what we see on the graph. He used ERB as a point of reference with regards to psychoacoustics to make a point and teach us why is it that we hear something the way we do. But for @jaakkopasanen to turn this other way around in the same thread on the forum to make a polar opposite point explaining what we see vs. what we hear, IMHO must not be done without previous explanation, just for the sake of avoiding confusion.

In a similar fashion, one could make a video explaining why we see colors the way we do, and use the very presence of light as a point of reference. Someone else could try to make his own point in the same thread about why we don't see colors in the absence of light and create similar confusion. And the discussion goes in the opposite direction, and now everyone is trying to prove weather or not the colors even exist in the absence of light.

Although the knowledge provided is doubled for someone, for someone else it may be just confusion. IMHO, threads in which we discuss what we hear (vs what we see on the graph) and the threads in which we discuss what we see on the graph (vs. do we hear it or not) should be miles apart. Simply because of possible misinterpretation.

I hope I did not create even more confusion here. But if you think I did, just please follow your intuition and learn both aspects of the same thing, but please don't further undermine Amir's work here, because he did his best not to create confusion in the first pace. We hear with our ears and we see with our eyes, both are interpreted by the brain simultaneously. Audio is about what we hear.

I believe that I was the first to notice and point out the one specific error in Amir's video and also in his referenced article. There is no confusion about what the error was nor what the correct interpretation is. This was generously explained by @jaakkopasanen who took the time to provide progressively more detailed explanations in several subsequent posts. His effort to provide accurate factual information should be applauded and recognized. Indeed, that is the very same thing Amir has been doing for a long time, and which is why we are all grateful for his efforts and the knowledge he so generously shares. We all make mistakes from time to time. Amir made one here. No big deal, if it is recognized and corrected, and we move on.

 

[Robbo99999]

Ok, at the root of the issue is Amir's rule-of-thumb claim "Only wide parametric equalizations are audible in high frequencies" where "wide" means "looks wider on this logarithmic frequency response graph." That's a straightforward (if not easy) empirical question.

What you (Jaakkopasanen) are saying here is that the ERB filter bandwidth growth can not explain that, even supposing it is true. That's because the ERB bands get wider but at a sub-exponential rate, so the (linear) width on the frequency response graph is still smaller at high frequencies even as the (numerical) width increases.

Is that it?

I think you've got it there. As you go up the frequency range the logarithmic x-axis "effect" overides the ERB filter bandwidth growth in terms of comparing equally wide (in terms of pixels on the graph) peaks or dips....therefore you can conclude that hearing sensitivity increases as you go up the frequency range in terms of comparing the audibility of equally wide (in pixels) dips or peaks.

 

[audiofooled]

In addition to my previous post(s), more about auditory filters:

And some more about pitch perception:

These are some practical examples on how far apart the frequencies have to be in order for auditory filters to actually tell the difference in loudness and meaning, or pitch. Further more, how do we perceive pitch and how far apart in the frequency spectrum signals have to be in order to tell the difference in pitch. Also, when played simultaneously, how far apart in the frequency spectrum signals have to be in order to be perceived as separate harmonics, vs just an average of the two, modulated.

So, it turns out when we listen to sinevawes, on a logarithmic scale (look at the description) and at constant volume:

We will perceive differences in loudness, which may be huge or not, depending on FR.

However, if we play some music (I usually do), which is played in KEYS at the pitch differences in cents, to be able to discern how loud close tones are is just a wishful thinking, especially when we combine them with reflections. Remember that the higher in frequency we go, the higher the frequency difference must be to be able to discriminate the difference in pitch. Close tones on the ERB scale would be either masked by others in the same filter bank, or perceived as an average of the two but short enough in modulation that we perceive only one average amplitude. As for the perceived loudness, combination of the close tones will be perceived just as slightly louder than each of them separated because they will share the same hair cells at the Organ of Corti. The higher in volume we go, the protective mechanism also comes into play...

EDIT: If we put into the equation the variable that all mechanisms of auditory perception are highly individual, also age dependent...?

 

[ctrl]

For all those who still find the criticism of Amir's statements confusing, here is an attempt to present it as simply as possible - hopefully not so simplified that it is incorrect

Yep, in the video Amir has phrased it wrong or in a way that could be misleading - more precise at 07:47 and especially at 08:06.

The starting point of his argument are visually identical resonances on the logarithmic frequency scale (of a loudspeaker measurement) and the loudness perception when using ERB.

Here is a definition of how loudness perception is explained using ERB - Amir touched on this very briefly in the video:

Our loudness perception is based on excitation of the basilar membrane found in the inner ear.
The human ear processes sounds in parallel, meaning it handles the different sound components simultaneously. This involves breaking down the incoming sound signal into different characteristic frequency bands (known as “frequency groups”) and forming the partial loudness of each frequency group.
In the field of psychoacoustics, the loudness impression produced by a sound (or the “overall loudness”) is computed by adding up the partial loudnesses produced in the ERB frequency groups. The unit of the frequency groups is the equivalent rectangular bandwidth (“ERB”). In some cases, the older (and now obsolete) bark scale is used instead of the ERB scale.

Source: https://www.schleske.de/index.php?id=35&type=123&L=2

or another definition, in this case using the Bark scale (somewhat older classification of critical bands)

To measure the loudness, the sound pressure values in the 24 frequency groups (when using the Bark scale) are determined and converted into partial loudnesses (in sone) using graphical or numerical methods.

The sum of the partial loudnesses results in the total loudnesses, whereby masking effects must still be taken into account as corrections. If one determines the loudness in phon from the total loudness in sone found in this way, this value agrees quite exactly with that which one finds directly by auditory comparison with the 1000 Hz tone finds.

Source: Translation from "Reinhard Lerch · Gerhard M. Sessler · Dietrich Wolf - Technische Akustik - Grundlagen und Anwendungen"

Applied to our resonances, this means that a resonance is heard better the more ERB's are involved (for "identical" resonances on the log Frequency scale, same Q factor) - so I hope this is not too simplified.

Now let's consider three "identical resonances" (200Hz, 1kHz, 10kHz - same Q factor) on the usual frequency scale (analogous to what Amir did in the video).
For each resonance center frequency the critical bandwidth is plotted, once for the ERB scale and once for the Bark scale.

According to this, the resonances at 1kHz and 10kHz should be much more audible than the resonance at 200Hz because multiple frequency bands are excited there.

Experimentally, comparable results are obtained:

Source: Toole - Sound Reproduction

With the same Q factor, a resonance at higher frequencies is audible at lower relative sound pressure levels than at lower frequencies.

 

[jcarys]

So I'm interested in the Audyssey comment in the video. Amir states that you should bump up the target to about +5 db at 20Hz and then there's a gentle rolloff to some frequency I missed. So two questions - what is the frequency I should have this back to the flat line? I already don't let Audyssey adjust anything over 500Hz. And I assume this is just for the front channels in the Curve Editor? If this needs to move to an Audyssey discussion, just let me know. Thanks.

 

[abdo123]

So I'm interested in the Audyssey comment in the video. Amir states that you should bump up the target to about +5 db at 20Hz and then there's a gentle rolloff to some frequency I missed. So two questions - what is the frequency I should have this back to the flat line? I already don't let Audyssey adjust anything over 500Hz. And I assume this is just for the front channels in the Curve Editor? If this needs to move to an Audyssey discussion, just let me know. Thanks.

generally you should look for a gentle downward slope from 100 Hz to 20KHz for about 6 dB so 100Hz = 0 db and 20KHz = -6 dB.

and the subwoofer should be 4 dB louder than the main speakers so 80Hz and below should be at least +4 dB than the frequency at say 1KHz . Harman further suggests an upward slope below 100 Hz.

Here is the entire Harman target curve, frequency first, amplitude second.

31.5 6
40 5.9
50 5.5
63 4.9
80 3.7
100 2.5
125 1.3
160 0.6
200 0.2
250 0
1000 -1
32000 -6

 

[jcarys]

Thanks, abdo123. I went back to the video and paused and got very close to this. Fat fingers have now edited the curve editor and I'm trying it out now. I'm going to try the Steven Wilson surround mix for Tull's A on it and see if the bass rattles the floor or not.

 

[Arc Acoustics]

I'm very sorry for who gave me a reply on the following thread, no matter positive reaction or not.
I received some reply alerts but I couldn't read them because the thread was deleted (as I predicted) for "Personal" insults.
Deleting my post was kinda ok, but the guy's replies with effort were not, I wanted to read them, I'm very sorry.

Guys in the original thread "modestly" pointed out amirm 's mistake for him, and I waited for him to retract or correct this video for a whole week but he seems to decide shamelessly to ignore his rather shameful mistake.
Probably he can not admit his basic fault because he believes he should keep strong and flawless Amirm to a bunch of ignorant followers.
So now we should regard audiosciencereview as just an audioreview.

Of course, he can decide what he does, I do not have any right to compulsion.
He or admin can delete my post, he can ban me, he can do everything he want and I will not complain about that, but it's pathetic.

 

[bobbooo]

I'm very sorry for who gave me a reply on the following thread, no matter positive reaction or not.
I received some reply alerts but I couldn't read them because the thread was deleted (as I predicted) for "Personal" insults.
Deleting my post was kinda ok, but the guy's replies with effort were not, I wanted to read them, I'm very sorry.

If you have email alerts set up on here you should have received each reply as an email (check your spam folder). If you find my reply, you have my permission to post it here.

 

[Arc Acoustics]

If you have email alerts set up on here you should have received each reply as an email (check your spam folder). If you find my reply, you have my permission to post it here.

Thank you for your suggestion.
Unfortunately, I saw (which deleted as well) multiple reply notification (upper right) at least but all I've got on email was only two messages (from Vict** and vood*****) with the first sentence of the reply, sorry.
Probably because the notification email is delayed so it might have some time lag.
I really would like to know the citizen of Amir Kingdom's thought, was it right, wrong, or some neutral opinion, but it might be dangerous a bit. lol

 

[Robbo99999]

Thank you for your suggestion.
Unfortunately, I saw (which deleted as well) multiple reply notification (upper right) at least but all I've got on email was only two messages (from Vict** and vood*****) with the first sentence of the reply, sorry.
Probably because the notification email is delayed so it might have some time lag.
I really would like to know the citizen of Amir Kingdom's thought, was it right, wrong, or some neutral opinion, but it might be dangerous a bit. lol

Insults are of course wrong, no question. However you keep making these subtle offensive jokes about citizens of Amir Kingdom and believers in Amir's dogmas (see previous page). Just saying.

I have to say I can't remember the last time Amir said he was wrong in something, and I'm sure he's wrong with his video here, either through it being easily misunderstood or he had a genuine illegitimate viewpoint.....generally hinging around his assertion that "visually identical" peaks/dips as viewed on a log x-axis scale in high frequency are less audible than in low frequency, which proved wrong. Humility is good, especially if it aligns with science. He does some really great work and this website is a credit, but I think he's reluctant to clarify or admit weak points. I think the positive side is people like @jaakkopasanen did some really clear explanations of the reality of the situation, thereby clearing up either any misunderstandings or falsifications related to the video.

 

[bobbooo]

Thank you for your suggestion.
Unfortunately, I saw (which deleted as well) multiple reply notification (upper right) at least but all I've got on email was only two messages (from Vict** and vood*****) with the first sentence of the reply, sorry.
Probably because the notification email is delayed so it might have some time lag.
I really would like to know the citizen of Amir Kingdom's thought, was it right, wrong, or some neutral opinion, but it might be dangerous a bit. lol

Luckily I've found my reply was still in my clipboard (slightly edited):

Making mistakes is absolutely, 100% ok. What's not ok is obstinately refusing to admit to and correct those mistakes, especially from a projected and perceived authority in audio to many, on a YouTube video presented as educational, with the potential to misinform a large audience as this one has. Every good, scientifically-minded or simply fact-based channel I've seen immediately and openly corrects mistakes made in previous videos at the start of the following video. The great thing about this site is the live, fluid peer review process from thousands of bright, rational minds, and the admission and correction of mistakes when highlighted is an essential part of this process, in order to maintain credibility and truth. Failure to address and censorship of valid criticism just acts to erode trust, the complete opposite of I presume the intent of attempting to project an infallible image of authoritative knowledge. (The Streisand effect comes to mind.)