Master Complaint Thread About Headphone Measurements

[thewas]

Are headphones rated with or without EQ? I would suggest to use more clear categories.

Recommended
or
Recommended (only if you use EQ)
or
Not recommended (regardless of whether you use EQ or not)

Or
The good, the bad and the ugly

 

[amirm]

There should also be a way to measure the level of sealing provided by the earpads... Some headphones are sensitive to that.

I suffer a lot with some headphones in that regard as I try to measure them. If I remember, I will point it out.

Some others doing such measurements use clamps and such (especially those using the simpler GRAS 43 fixture) which exaggerates the kind of fit you would get on a real head or my fixture.

 

[amirm]

Are headphones rated with or without EQ? I would suggest to use more clear categories.

Recommended
or
Recommended (only if you use EQ)
or
Not recommended (regardless of whether you use EQ or not)

That is what I have done so far.

 

[Thomas_A]

That is what I have done so far.

Yes, I noted that on the HD800. But you also wrote that all headphones must be "EQued". That is confusing to me when it comes to the recommendation. Does it mean that recommendation "out-of-the box" is not an option?

 

[Thomas_A]

Or
The good, the bad and the ugly

That's another way of putting it.

 

[amirm]

Yes, I noted that on the HD800. But you also wrote that all headphones must be "EQued". That is confusing to me when it comes to the recommendation. Does it mean that recommendation "out-of-the box" is not an option?

Not at all. If it comes close to reference or sounds great as is, I will give it a recommendation on that basis and not even try EQ if not needed.

 

[Matias]

Well, since you asked, here are few suggestions.

1. Always show a normalized response to Harman's curve. Makes it easier to identify differences in tonality.

2. Always add the measurement file to the end of the review, so that users can use it to tweak EQ settings.

3. Have REW generate PEQ settings aiming for the Harman curve.

4. Subjective evaluation after the PEQ for Harman curve.

 

[Thomas_A]

Not at all. If it comes close to reference or sounds great as is, I will give it a recommendation on that basis and not even try EQ if not needed.

Ok thanks. Your sentence about EQ made me a bit confused. I personally do not use EQ other than room correction for speakers. For simplicity since I use various sources with my headphones. I use headphones with an already good target response, and there are not too many of those out there..

 

[amirm]

3. Have REW generate PEQ settings aiming for the Harman curve.

I haven't had success in generating these filters in REW for headphones. Does anyone have a write-up on it?

 

[Dreyfus]

I haven't had success in generating these filters in REW for headphones. Does anyone have a write-up on it?

I recommend the Auto EQ project by Jaakko Pasanen. It is very well documented and a powerful tool for equalizing headphones. You can define a lot of variables, including filter settings and compensation curves.

In relation to the topic of sealing, maybe try something like this?

Keith Howard: Better Headphone Testing

 

[spanayotov]

Hi All,

First, l would like to say just how much impressed I am of ASR, such a great community. I only wish it existed years ago when I listened to headphones mostly (as opposed to speakers) and my goto place was head-fi forums.

Regarding headphone measurements. I think it would be useful if we had input from someone who worked in headphone design. I am under the impression that many members here have experience in building loudspeakers and/or knowledge of the industry but are there any headphone engineers on ASR? Of course, I am aware that manufacturers are protecting their trade secrets and employees maybe cannot say much.

 

[pkane]

I haven't had success in generating these filters in REW for headphones. Does anyone have a write-up on it?

It's pretty easy. You just need to load your house curve, first, then load the GRAS measurement, set a few optional settings for EQ and run it.

Attached to this message is the target curve you posted, ready to load into REW. Open REW, click on the preferences icon, select House Curve option, then browse and select the downloaded curve:

Exit preferences, load your GRAS measurement file, and click on EQ icon at the top:

• On EQ screen, select Equalizer as Generic
• Under Target settings, Full Range speaker
• Unselect Add Room Curve
• Click on Calculate target level from response:

Under Filter Tasks, set the desired EQ options, be careful with max boost settings, I have them at max value. Click on Match response to target:

I then manually go through the suggested filter and add some shelves or other changes where I think I can improve the suggested the result. Press the EQ Filters button at the top of EQ window to see the filters.

Here's what I get after a couple of minutes of playing:

Looks better with some smoothing

 

[thewas]

Please keep in mind that narrow positive gain notch filters can cause more audible problems than the increased distortion due to the higher power, one of the reasons why EQ cannot fully correct a flawed transducer:

-Positive notch causes ringing in time domain. This is more likely be audible above low frequencies and high Q-values.
From https://support.genelec.com/hc/en-u...-does-GLM-not-fix-dips-on-frequency-response-

I used to also use such extreme auto-EQ settings in REW but nowadays prefer more moderate manual filters which may not look from FR as nice but sound better to me, so can only recommend everyone testing several EQ strategies and chose the one that sounds best to him.

 

[pkane]

Please keep in mind that narrow positive gain notch filters can cause more audible problems than the increased distortion due to the higher power, one of the reasons why EQ cannot fully correct a flawed transducer:

-Positive notch causes ringing in time domain. This is more likely be audible above low frequencies and high Q-values.
From https://support.genelec.com/hc/en-u...-does-GLM-not-fix-dips-on-frequency-response-

I used to also use such extreme auto-EQ settings in REW but nowadays prefer more moderate manual filters which may not look from FR as nice but sound better to me, so can only recommend everyone testing several EQ strategies and chose the one that sounds best to him.

I usually use REW filters as a starting point, eliminating any sharp notch filters with Q above 7 or 8.

 

[Vini darko]

Grumble grumble hurumph

 

[pwjazz]

1. Thanks for the work you do
2. Thanks for soliciting feedback
3. Please use a logarithmic scale on distortion graphs so we can see details in the lower distortion areas, as these sometimes provide useful insights into strange driver behaviors at certain frequencies
4. Please give stock tunings a chance. The Harman target provides a good baseline preference curve, but that doesn't mean that a headphone with a different tuning can't sound good to some people on some material some of the time.

 

[Dreyfus]

Please keep in mind that narrow positive gain notch filters can cause more audible problems than the increased distortion due to the higher power, one of the reasons why EQ cannot fully correct a flawed transducer:

-Positive notch causes ringing in time domain. This is more likely be audible above low frequencies and high Q-values.

That is probably also the reason why a lot of HRIRs have such a bad reputation. They often do not subtract all the comb filters of the room and distortion of the playback/recording system. The sound can get really messy when you apply all that to your headphone.

Did Olive et al comment anything on this (high Q ringing) topic when working on their virtual headphone simulator?
As far as I remember they applied a lot of smoothing.

 

[Feelas]

That is probably also the reason why a lot of HRIRs have such a bad reputation. They often do not subtract all the comb filters of the room and distortion of the playback/recording system. The sound can get really messy when you apply all that to your headphone.

This, especially because hearing resolution isn't as good as the current measurement rigs' resolution, thus oftentimes this one notch filter might contribute a lot of additional THD, yielding no real improvement to the perceived sound, since the resolution is lower anyways.

@amirm, someone's mentioned before that DSP (like the ones in ANC headphones) could use the track itself as the reference point, so I've got a point to discuss.
Would you think that running tests comparing some "raw tracks", just EQ'd to the target curve with what was played & recorded back, could contribute anything of value? I wonder if running some non-synthetic tests could tell us anything interesting. Maybe even plotting the original signal spectrum vs the played-back would be enough, or are the GRAS's microphones completely unsuited for real musical material & something akin to "null testing"?

 

[amirm]

Regarding headphone measurements. I think it would be useful if we had input from someone who worked in headphone design. I am under the impression that many members here have experience in building loudspeakers and/or knowledge of the industry but are there any headphone engineers on ASR? Of course, I am aware that manufacturers are protecting their trade secrets and employees maybe cannot say much.

Both ODMs (companies that make headphones under branding for others) and manufacturers have contacted me privately. They don't want to comment in public though as you say.

 

[amirm]

Would you think that running tests comparing some "raw tracks", just EQ'd to the target curve with what was played & recorded back, could contribute anything of value?

I don't think so. Problem is that your headphones will be different and so you are not hearing what I am hearing. I think the best ratification we can have is people taking the EQ settings if they have the same headphone, and apply them themselves. If the results are mostly positive, then we know it is working.