Is the level matching needed when it is the physically same headphone without EQ, compared to the same one with EQ (no pre-gain)? Would one situation actually sound overall louder/softer?
In the Harman study, EQ was applied in all instances, right?
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Hifiman Sundara Review (headphone)
- Thread starter amirm
- Start date
In the Harman study, EQ was applied in all instances, right?
So, how would you compare headphones with and without EQ at home?
You seem to disagree with the way Amir does his verification of the effects of his EQ profile. As far as I understand he doesn't do any level matching.
To be strictly scientific you should level-match by loudness. However, in a study of the preferred low-frequency response of IEMs, Dr. Sean Olive found that loudness normalization did not really have a statistically significant effect on preferred bass shelf level and frequency:
Note this was only for IEMs (although I would expect it to hold for over-ear headphones), and only considered the bass shelf, so loudness effects may become more significant with more wide-ranging EQ changes. This study does suggest loudness matching, although desirable, may not be paramount though. What is however is blind listening, for which I suggested a method in my post #362 above.
Thank you for this.
Blind level-matched A/B tests are valid for some situations, such as amplifiers, DAC's etc.
But when it comes to one headphone, to test changes of its sound because of EQ'ing may not be achievable.
For one, level matching may not be achievable as the ear is more sensitive to certain frequencies and by changing those frequencies loudness using EQ, the ear may percieve loudness of the total sound not so accurately.
For two ! once you apply EQ that is audible (otherwise what's the point) , then the listener can tell which is which - so blind is out too.
Sometimes we may like a bit more sparkle with a certain song - but for another it may prove too much. To permanently alter the FR of a headphone may not yield desired effect.
If a headphone requires EQ at frequency ends (too much bass or too much top end), then EQ can help - if it is in midrange area, it will be hit or miss (mostly miss). If there is a suck-out in the frequency band, proceed with caution! because artificially boosting a frequency range on a headphone can increase distortion in the same range too.
My sennheiser IE 400 Pro IEM's has too much bass, so I use EQ to bring it down by 6dB, but that's all - after EQ it will sound right.
I tried all sorts of EQ on an Aeon flow open X - nothing worked! I have wrapped it up for eBay.
On the other hand, Sundara's may benefit from a slight boost at 1300 Hz (1.5dB), but it is fine without - I EQ on certain songs.
On classical music, where there is a lot of fine detail (micro dynamics) , EQ stays off, simply because my EQ system interferes with those fine details.
As I said previously, loudness (not simple level) matching is achievable, by following a frequency-weighted model such as ITU-R BS 1770 as Harman do. Whether this loudness normalization is strictly necessary given the above research on preferred IEM bass shelf I cited is debatable, but it does remove a potential confounding variable.
This is a common fallacy I see repeatedly on here. We are discussing preference tests, not ABX discrimination tests. As I outlined in my post #362, if you do the blind test before any sighted listening (even better without seeing the EQ either), you will not be able to tell which is which.
Last edited:
OK, this is my last post on the subject, as it seems I am up against really knowledgable members.
The understanding of Loudness, ITU/Harman curves are achieved by experiments done on many and averaging out the results, it is not set in stone.
We are but individuals, the results are just a guide line for measurement and for manufacturers to appeal to a larger proportion of their target clients.
Many on this forum believe passionately about EQ. they go as far as "all headphones NEED EQ to sound right" well good luck to them, I mean it.
I don't agree! That is how this latest back & forth started.
EQ may bring the FR of a headphone close to Harman curve (if measured by right gear), but how many also considered the damage such multiple digital EQ system does to the final sound??
FR is important - but transparancy matters too.
There are many headphones, that have been reported by many as sounding wonderfull, yet the FR curve does not match Harman's, so what gives?.
As for me, I do not need a level-matched blind test to confirm what I like - I already do!
Those who want to prove a point (or disprove) and those who are researching to design a new device can do all the tests they like.
I Just tried to high light difficulties in doing such tests on an individual level, you wanna call that Fallacy - more power to you.
Another fallacy, and what I suspected was the source of your reluctance to EQ. There will be no 'damage' done to the sound by well-chosen digital EQ filters such as Oratory's. In fact 'transparency' (to the source sound heard by the mastering engineer and intended by the artist) will increase. Please explain specifically what this supposed damage is that you speak of.
I have been creating them for the headphones I use. I could whip one up for ya.. I don’t own this one.. yet…
Here you go:
Last edited:
Sure if you want to
I tried myself but it is a bit confusing as the numbers i put in MathAudio does not show anywhere after i have entered them so i cant doublecheck that i entered them right. I found these numbers to put in for the revised version.
Dont know if these were what i was supposed to put in.
Peaking | 20 Hz | 0.81 | 5.7 dB |
Peaking | 2348 Hz | 3.3 | 2.5 dB |
Peaking | 5815 Hz | 2.92 | 0.6 dB |
Peaking | 7679 Hz | 3.04 | 1.4 dB |
Peaking | 9514 Hz | 3.52 | 1.7 dB |
Peaking | 234 Hz | 1.27 | -1.7 dB |
Peaking | 1148 Hz | 1.15 | -2.4 dB |
Peaking | 1697 Hz | 1.22 | 1.4 dB |
Peaking | 11516 Hz | 1.2 | 1.9 dB |
Peaking | 19828 Hz | 0.36 | -8.6 dB |
I just posted them to my post above. Check it out.Sure if you want to
I tried myself but it is a bit confusing as the numbers i put in MathAudio does not show anywhere after i have entered them so i cant doublecheck that i entered them right. I found these numbers to put in for the revised version.
Dont know if these were what i was supposed to put in.
I am not sure where you got your values from. To get the values, go the the 1st post in this thread and use the values Amir posted:
Just 3 filters to add to Math Audio. And yes, you can't see what you typed in after you type it in, but the graph should look like the graph Amir posted in his first post.
Math Audio:
Amir:
Great! I am totally new to the whole EQs thing so i got very confused what numbers i should have used but now i understand. Thank you!
Pio2001
Senior Member
Hi, I own Sundara, Sennheiser HD600 and Beyer DT-880 headphones.
My prefered one is the Sundara, although I find the HD600 to have different qualities. The DT-880 is the one I like least.
It seems to me that the HD600 is more natural sounding with solo instruments and human voice. The Sundara sounds artificial and "hollow" in comparison. For example, in this track, the guitar sound like a real guitar with the HD600. Quite less with the Sundara :
Current 93 - A Beginning
On the other hand, I find the Sundara to be more balanced overall, while the HD600 sounds a bit harsh. I like the Sundara better for orchestral music. For example, in this track, the orchestra and choir sound great with the Sundara, while they sound a bit smaller and smashed onto my ears with the HD600 :
Bach Netherlands Society - Mass in B minor BWV 232
I have tried the equalization posted above (with REW / Foobar convolver) for the Sundara.
First, I don't like the bass boost at all. In this track, the lowest notes sound way too loud compared to the higher ones.
Unknown artist - Soeren
I find it much more balanced with a flat frequency response below 100 Hz.
Then I tried the rest of the correction without the bass boost. But I wasn't convinced. Yes, it removes the "hollow" sound of the Sundara, but on the other hand, it also reduces its ability to reproduce orchestral music .
Can we be sure that the correction is exactly what is needed ? The spike at 6300 Hz is very sharp, and any inaccuracy caused by driver to driver variation, or ear canal resonance would ruin the equalization.
Is the spike exactly at the same frequency when these headphones are measured by others ?
Also, it is very unlikely that this spike has exactly the shape of a parametric correction.
My prefered one is the Sundara, although I find the HD600 to have different qualities. The DT-880 is the one I like least.
It seems to me that the HD600 is more natural sounding with solo instruments and human voice. The Sundara sounds artificial and "hollow" in comparison. For example, in this track, the guitar sound like a real guitar with the HD600. Quite less with the Sundara :
Current 93 - A Beginning
On the other hand, I find the Sundara to be more balanced overall, while the HD600 sounds a bit harsh. I like the Sundara better for orchestral music. For example, in this track, the orchestra and choir sound great with the Sundara, while they sound a bit smaller and smashed onto my ears with the HD600 :
Bach Netherlands Society - Mass in B minor BWV 232
I have tried the equalization posted above (with REW / Foobar convolver) for the Sundara.
First, I don't like the bass boost at all. In this track, the lowest notes sound way too loud compared to the higher ones.
Unknown artist - Soeren
I find it much more balanced with a flat frequency response below 100 Hz.
Then I tried the rest of the correction without the bass boost. But I wasn't convinced. Yes, it removes the "hollow" sound of the Sundara, but on the other hand, it also reduces its ability to reproduce orchestral music .
Can we be sure that the correction is exactly what is needed ? The spike at 6300 Hz is very sharp, and any inaccuracy caused by driver to driver variation, or ear canal resonance would ruin the equalization.
Is the spike exactly at the same frequency when these headphones are measured by others ?
Also, it is very unlikely that this spike has exactly the shape of a parametric correction.
Contrary to common belief (among many here at ASR), Equalizing is not a magic bullet!
Some headphones may benefit from it, but it won't cure a rainy day.
In case of Sundara's , as you have discovered, there is not much to cure to begin with.
What little there is, EQ is not the answer, it helps one corner but messes with another.
Just wait for "true believers" in EQ to pounce in . . . .
You are still wrong, and still you don't understand why.
First we need to determine whether you have the pre- or post-2020 version (this review is of the latter), because a change in the pads resulted in a significant change to the sound:
You're right though, the cut in the EQ on here to counteract that spike is too sharp, with a Q of 4 and gain of -4 dB. Oratory1990's EQ profile uses a less extreme -2.5 dB filter instead. He usually measures several units of each headphone model and averages the response before creating the EQ, in order to ameliorate unit variance issues. Also note he says this about the bass filters:
So you can just adjust those to your liking.
Wow +6dB on subbass including the already boosted lows. I tried but the total preamp gain reduction is huge. Need to wait for my amp to arrive. The standard Oratory1990 eq curve already sound good as it is for me
Pio2001
Senior Member
Thank you for the link !
I'll try these in REW and see if they work better for me.
I think that I've get the recent version : the earpads are thinner on the front side.
Similar threads
