Equalize first, measure second: Share your results!
Apr 20, 2020 at 3:55 AM Thread Starter Post #1 of 10

HuoYuanJia

Previously known as Ultrazino
Member of the Trade: HEDD Audio
Joined
Jun 18, 2010
Posts
369
Likes
762
Location
Germany
Trying to figure out the "perfect" target response for my IEC 60318-4 raw measurements, I just couldn't agree with most targets out there. So I decided on a new approach.

Levering out any research, I have individually equalized multiple high-end IEMs with a sine tone generator so I could match the pressure from 80 to 10,000 Hz. This was not done through listening to music but only separate tones over multiple hours. Quite a tedious and fatiguing process. Matching the pressure needs a lot of attention...

I have tried adding lesser tier IEMs but I noticed that I could clearly hear the distortion. Comparing a 3 kHz tone over and over again with different IEMs makes you really susceptible to added noise.

Anyway, I have measured my results:
comparison.png


It looks wild, but it really isn't. The curves all hit the average within 2dB, which is a rather small difference. Unfortunately, the target is only useful for myself. I still have very good hearing, but there are many individual aspects that come into play. (I have done this two years ago with the InEar ProPhile 8 before. Back then I ended up with a similar curve but it showed less bass and more treble.)

Maybe this can inspire some more people to do the same and share their results.

target-1.png


Full article: Constructing an Audiophile IEM Reference Target
 
Apr 25, 2020 at 5:10 AM Post #2 of 10
Levering out any research, I have individually equalized multiple high-end IEMs with a sine tone generator so I could match the pressure from 80 to 10,000 Hz. This was not done through listening to music but only separate tones over multiple hours. Quite a tedious and fatiguing process. Matching the pressure needs a lot of attention...

I'm not sure I understand what you've done or how you've done it. Trying to match pressure "by ear" from 80Hz to 10kHz would be a very difficult and probably highly inaccurate. For example, taking a sine wave at around 3kHz as the 0dB reference, how would you accurately judge that a sine wave at say 1.5kHz is about 8dB higher, at 10kHz about 14dB higher and at 80Hz about 20dB or so higher? That's just 3 test tones, each additional test tone requires a different relative dB level and it's extremely difficult (near impossible) to judge dB differences "by ear" if the frequencies are different. Even with a great deal of training I highly doubt anyone could consistently achieve this feat to an accuracy within 2dB.

Matching loudness (rather than pressure) "by ear", it would be hugely easier to achieve accuracy within 2dB, though still tedious. However, what you would end up with is not a flat response but a response that includes an equal loudness contour. That's fine if all you ever listen to is test signals but NOT if you listen to music or sound recordings, because they have been created by artists/engineers who are obviously also subject to equal loudness contours and therefore the music or sound mix has already been compensated for an equal loudness contour. So, if you apply this target response, you'd effectively be compensating for an equal loudness contour twice. That might sound OK to some people but would be far from neutral/flat/accurate.

G
 
Apr 26, 2020 at 5:28 AM Post #3 of 10
Hi @gregorio, thank you for your reply!

Indeed I meant loudness and not pressure. The SPL curve as shown is the actual SPL and visual representation of my perceived linearity.

I agree, it is not easy and takes a lot of practice. I recommend to use a generator that tracks the mouse cursor. That way you can easily adjust the speed.

First step: Move quickly across the frequency range and notice where the volume drops or rises. Repeat with a slower speed until you find the right position. Add the EQ frequency point there.
Second: Adjust relative gain. Raise or lower the EQ gain to the sounds above and below.
Third: Adjust the quality factor of the PEQ so the transitions are smooth. If you can't smooth it out completely (two drops), shift the frequency point either up or down. If you locate three dips, lower the quality factor again and set a second frequency point.
Fourth: Adjust to reference gain. Now jump back and forth from where you located the peak/dip and a reference like 500/700/1000 Hz. Might have to lower the quality factor when the difference is very noticeable.

I highly recommend to use IEMs with low distortion. Added harmonics will become very noticeable. I suspect this would distort the outcome.

what you would end up with is not a flat response but a response that includes an equal loudness contour.
Well that's the whole point. And that is perceived flat under my conditions. The volume and equipment I listen to. As I said, this is not a scientific approach but sharing what "neutral" IEMs could look like - because not many will agree with the Etymotic or Harman curve. My aim is to motivate others to do the same and see what results they get.

And I think my target curve nicely shows the loudness correction in-line with Fletcher and Munson. A higher volume would drop bass and lift 3 kHz. In fact, I can confirm this when I repeat the process at a higher (uncomfortable) volume. It is one of the aspects other targets have not considered and why most sound designers tune their IEMs with more bass than recommended in the textbook.

That's fine if all you ever listen to is test signals but NOT if you listen to music or sound recordings
I'm not sure what music you listen to, but the topic of high-fidelity should be reserved to faithful recordings. Of course, most music is extremely heavily processed. Every track compressed and EQ'ed, every instrument altered. Every release has its unique sound so there is no way to factor this in unless you dedicate yourself to a specific genre and producer. From my experience, using my perceived linear target, it seems to be a very acceptable average. I still occasionally adjust bass and treble via the ADI-2 DAC, and in both ways. So depending on the song/album, this target might even still be bass-heavy.
 
Apr 26, 2020 at 6:09 AM Post #4 of 10
I've done things like that at some point, but as it was done with completely uncalibrated and simply not good measurement tools, the results have no meaning compared to say the expected raw response of a IEC711 coupler.
But in general, I can still identify some trends that are significant enough to exist on pretty much all calibrations:

I like to have about the lowest value in the, low to mid range, near 250Hz; and I leave the subs more or less the same on some vented dynamic drivers, or usually boost them near 60 or 80Hz on sealed stuff/BA drivers. So you can imagine something close to your target in the end but with a fairly wide reduction of maybe 3 or 4dB centered on about 250Hz.

Then I go for 9 or 10dB boost near 3kHz(I think I peak before 3kHz but who cares) so the all midrange would move up a little compared to you, or so I assume if by chance our rigs are similar enough.

I also like/need a reduction in the 4-4.5kHz area, almost all curves from all gears supposedly flat are annoying me at 4kHz and feel too strong there for me.

And I tend to keep going down gently after that(or so I assume). I have an almost dead zone at about 7.5kHz in my ears(like near 7 in one ear and closer to 7.5 in the other) where sine tones feel very quiet compared to the frequencies immediately next to that. So I can't really say anything about that area. and after that we enter the world of inconsistency for measurements so I wouldn't claim anything about anything.
 
Apr 26, 2020 at 9:02 AM Post #5 of 10
[1] Indeed I meant loudness and not pressure.
[2] Well that's the whole point. And that is perceived flat under my conditions.
[3] Every release has its unique sound so there is no way to factor this in unless you dedicate yourself to a specific genre and producer.
[4] So depending on the song/album, this target might even still be bass-heavy.

1. OK yes, it takes practice but it's far more achievable to accurately judge loudness than judging pressure "by ear" .

2. But that's the problem, test tones are flat but music/sound recordings are NOT, because ...

3. Yes every release is unique because every producer is somewhat unique but all producers are also somewhat the same. All producers are human beings and are therefore all subject to equal loudness contours. As the music/sound recordings they produce is based on what they hear when mixing, the resultant mixes/masters are NOT flat, they already compensate for the equal loudness contour. Your applied response curve then compensates for the equal loudness contour a second time!

4. Yes, all your songs/albums will be bass-heavy. Bass freqs in particular are perceived as much quieter at the same level/pressure (as per the loudness contours) and therefore, in order to sound balanced, the bass levels will be raised by the producer/engineers to compensate. As you're compensating for this feature of human hearing a second time, you're therefore adding far too much bass, which of course might be preferable to some people but is certainly not accurate.

G
 
Apr 26, 2020 at 1:48 PM Post #6 of 10
Dear @gregorio, thanks for chiming in. However, I completely disagree with you at this point. I think it is very bold of you to speak for all producers. If anything, many will factor in that we are not listening in a perfectly treated room and will tone down the bass in the mix to compensate room resonances - completely opposite of what you claim. There are different trends for different genres and also a development through the years that constantly changes. And there are enough releases that disregards trends completely.

But that has nothing to do with my initial post. How one wants to mix and master his music is something else. There are plenty audiophile records that capture the true high-fidelity nature. It is then up to the equipment to reproduce it accurately and that was the whole purpose of my post. Let's take the discussion of a completely different topic somewhere else. Thanks.
 
Apr 26, 2020 at 3:29 PM Post #7 of 10
Recently, I've run across several multichannel SACD and blu-ray audio releases where the sub bass is severely attenuated. There is one record label where almost every release has next to nothing below 80Hz. I'm not sure why this is, but the reviews never seem to mention it. I suspect a lot of people who have multichannel speaker systems in their homes don't have much sub bass response. They don't even notice. Would there be an advantage to filtering off the sub bass?
 
Apr 26, 2020 at 4:44 PM Post #8 of 10
Recently, I've run across several multichannel SACD and blu-ray audio releases where the sub bass is severely attenuated. There is one record label where almost every release has next to nothing below 80Hz. I'm not sure why this is, but the reviews never seem to mention it. I suspect a lot of people who have multichannel speaker systems in their homes don't have much sub bass response. They don't even notice. Would there be an advantage to filtering off the sub bass?
I don't know the exact details but there is a matter of a +10 dB boost required for the LFE channel of some (or I think even most) multichannel recordings. I think the playback system should automatically take care of the right setting without the user having to worry about it. Could it be that for some reason this doesn't work properly with some discs (on some systems)?
 
Apr 26, 2020 at 5:08 PM Post #9 of 10
Some of them aren’t even attenuated. It’s just chopped off under 80. Very weird.
 
Apr 26, 2020 at 7:16 PM Post #10 of 10
[1] However, I completely disagree with you at this point. [1a] I think it is very bold of you to speak for all producers.
[3] If anything, many will factor in that we are not listening in a perfectly treated room and will tone down the bass in the mix to compensate room resonances - completely opposite of what you claim.
[4] There are different trends for different genres and also a development through the years that constantly changes. And there are enough releases that disregards trends completely.
[5] But that has nothing to do with my initial post. How one wants to mix and master his music is something else.
[6] There are plenty audiophile records that capture the true high-fidelity nature. It is then up to the equipment to reproduce it accurately and that was the whole purpose of my post.
[7] Let's take the discussion of a completely different topic somewhere else. Thanks.

1. You completely disagree that producers are human beings and therefore have human loudness contours?
1a. I can't think of anything LESS bold or controversial than stating producers are human beings!

3. Compensating for bass room resonances with speaker playback is another part of the equation, a different part of the equation to automatically compensating for loudness contours, with significantly different compensation values. So not the opposite of what I claim!

4. Yes there are different trends for different genres but there has NOT been any development or evolution through the years of human hearing. We all have loudness contours and have had since long before recording technology was ever invented.

5. How one wants to mix and master his or her music is entirely up to them. Having a loudness contour is not though, all producers have them and mix according to what they hear!

6. I don't know of any commercial music recordings that are not monitored and adjusted by human beings. I completely agree that it's up to the equipment to reproduce the recordings accurately but the "whole purpose of your post" is incorrect because you are not reproducing the recordings accurately, you are reproducing them with a compensation for a loudness contour, a compensation which has already been applied!

7. What different topic? You stated you are matching loudness and are therefore compensating for a loudness contour.

G
 

Users who are viewing this thread

Back
Top