Frequency Response Graphs After DSP/EQ

[luft262]

Is there anyone posting frequency response graphs of headphones after EQ has been applied? I ask because if a pair of headphones that are inexpensive, such as the AKG 371's can match the Harmon Curve with minimal distortion using EQ there would be little to no reason to pony up the money for headphones like the Dan Clark Audio Stealth's, which are $4,000. I would be willing to bet that most headphone users are streaming their music from a phone or computer and would have access to DSP.

I like that Amir posts the frequency response graphs for headphones and their distortion, but why not also post the frequency response graph after applying DSP. Maybe also the distortion graph after DSP to see if the DSP is causing too much distortion.

If a pair of headphones in the $100 to $300 dollar range can hit the Harmon Curve using EQ with low distortion, what would be the point of multi-thousand-dollar headphones other than prestige, aesthetics, or the small number of users who can't use EQ?

Thanks for your information and for reading my post!

P.S. I have the utmost respect for Dan Clark Audio and any company that is pursuing perfection using science and not snake oil I'm just using them as an example because they have a very expensive pair of headphones that are incredibly accurate and there are other headphones for much less that are very accurate if not as accurate and it's possible the EQ would make the difference negligible, which would be great for those of us not willing or able to drop $4K on a pair of headphones.

 

[ADU]

This (unfortunately) gets back into the question of whether or not FR (and to some extent also distortion) is all that really matters in terms of a headphone's sound quality. And there are people who come down in different camps on that.

There is actually a fairly wide spectrum of views on this subject. And my personal view is that there is probably a little more to it than than just FR and distortion.

Not everyone agrees that hitting the Harman curve is the best approach to tuning a headphone as well. And this includes some noteworthy reviewers/graphers who use the Harman target as one of their reference points. I agree that it would be interesting to see the post-EQ measurements though.

If you just want to get an estimate of what the response would be after the EQ filters are applied, this is something that can be done fairly easily using something like Equalizer APO's Configuration Editor, by just combining the filters with the headphone's FR curve, and looking at the net result of this "stack" in the graph in the Analysis Panel. I could try to post one or two simple examples to illustrate this, if you think it would help.

There may also be other ways of calculating or estimating the post-EQ response. But using the Analysis Panel and stacking feature in EAPO's Configuration Editor is one of the simpler/easier methods that I know, and have already utilized with my own headphones.

 

[luft262]

This (unfortunately) gets back into the question of whether or not FR (and to some extent also distortion) is all that really matters in terms of a headphone's sound quality. And there are people who come down in different camps on that.

There is actually a fairly wide spectrum of views on this subject. And my personal view is that there is probably a little more to it than than just FR and distortion.

Not everyone agrees that hitting the Harman curve is the best approach to tuning a headphone as well. And this includes some noteworthy reviewers/graphers who use the Harman target as one of their reference points. I agree that it would be interesting to see the post-EQ measurements though.

If you just want to get an estimate of what the response would be after the EQ filters are applied, this is something that can be done fairly easily using something like Equalizer APO's Configuration Editor, by just combining the filters with the headphone's FR curve, and looking at the net result of this "stack" in the graph in the Analysis Panel. I could try to post one or two simple examples to illustrate this, if you think it would help.

There may also be other ways of calculating or estimating the post-EQ response. But using the Analysis Panel and stacking feature in EAPO's Configuration Editor is one of the simpler/easier methods that I know, and have already utilized with my own headphones.

Thanks! The only problem I can imagine with the estimate is any given pair of headphones may distorte before achieving the desired curve.

 

[JRS]

Thanks! The only problem I can imagine with the estimate is any given pair of headphones may distorte before achieving the desired curve.

That is definitely true for the bass response where the boost required is simply unrealistic. The other big deal that needs to be mentioned is that all ears are not the same, and by this I mean anatomically, and so unless one is using a deep dwelling IEM, the sound can vary a bit from that published. Also, if you take a moment and read the first few pages of several headphone reviews here, you will see a before and after spec for how well the headphone conforms to the Harman Intl "standard," which is certainly a good approximation, if not the last word, on what the proper frequency response is. It is anything but flat. So right there is pretty much the answer to what you are asking. If you read down into the comments some alternative EQ settings are provided that you may prefer. The Oratory recs are quite good, but in the case of the HIFIMAN Sundara's I am using a different set. Even then they shriek on some recordings, so bottom line, it's a bit more complicated than EQ, but it goes a helluva long ways.

 

[ADU]

The only problem I can imagine with the estimate is any given pair of headphones may distorte before achieving the desired curve.

That is definitely true for the bass response where the boost required is simply unrealistic.

This is why its only an "estimate". You both make excellents point though.

Many open dynamic driver headphones in particular will likely begin to become noticeably and unacceptably distorted in the lower frequencies well before reaching the level of the Harman target in that range. And simply may not be able to reach the levels at all in the lowest sub-bass frequencies, because they can't maintain the air pressure necessary to accomplish this in their earcups.

Open planar-magnetic headphones can supposedly handle the adjustments in this range a bit better though, because they are often better extended and have less distortion in the lower frequencies to begin with than the dynamic driver headphones. (This is what many users claim anyway, and it seems to jibe with the FR and THD measurements on at least some planars.)

Many closed dynamic headphones should be able to hit (or possibly even exceed) the generic Harman target in the bass though, without producing an intolerable level of distortion. Particularly those which are already well-extended in the bass and sub-bass to begin with.

So does this mean that closed dynamic headphones are always better than open ones?... Not necessarily. Because (going back to the original point in my previous post) there could be more at work in the sound quality of these headphones than just their extension in the bass, or their ability to conform to a given target frequency response curve.

Most audiophiles seem to favor the open dynamic and planar headphones because they often to seem to have a broader, more immersive, and more believable sound than closed back headphones, in their opinion. This might be due to the addition of external sounds blending with a recording playing through the open headphones, and perhaps combining with its sound in a more believable way. Or perhaps it could be due to some other factors (like maybe some differences in their acoustic impedance, for example?). These are questions we probably can't fully answer, or only make some guesses about at this point.

 

[luft262]

That is definitely true for the bass response where the boost required is simply unrealistic. The other big deal that needs to be mentioned is that all ears are not the same, and by this I mean anatomically, and so unless one is using a deep dwelling IEM, the sound can vary a bit from that published. Also, if you take a moment and read the first few pages of several headphone reviews here, you will see a before and after spec for how well the headphone conforms to the Harman Intl "standard," which is certainly a good approximation, if not the last word, on what the proper frequency response is. It is anything but flat. So right there is pretty much the answer to what you are asking. If you read down into the comments some alternative EQ settings are provided that you may prefer. The Oratory recs are quite good, but in the case of the HIFIMAN Sundara's I am using a different set. Even then they shriek on some recordings, so bottom line, it's a bit more complicated than EQ, but it goes a helluva long ways.

Those are good points, but I don't believe Amir ever posts a frequency graph after DSP. He'll post one without DSP and he'll post what peq filters he would recommend, but what I want is a frequency graph after DSP has been applied and preferably also a distortion graph so we can see if the headphones were able to achieve the Harmon target with acceptable distortion using dsp. If a cheap pair of headphones is able to do this they might be worth it over an expensive pair, such as the Dan Clark Stealth, because I always have access to DSP and I'd imagine most other headphone users do too.

 

[Sgt. Ear Ache]

I think many people would be shocked at the reality of just how little is gained by spending $4000 on a pair of headphones. I don't think it's at all out of the question that a much less expensive set could sound every bit as good...

 

[JRS]

Those are good points, but I don't believe Amir ever posts a frequency graph after DSP. He'll post one without DSP and he'll post what peq filters he would recommend, but what I want is a frequency graph after DSP has been applied and preferably also a distortion graph so we can see if the headphones were able to achieve the Harmon target with acceptable distortion using dsp. If a cheap pair of headphones is able to do this they might be worth it over an expensive pair, such as the Dan Clark Stealth, because I always have access to DSP and I'd imagine most other headphone users do too.

I know I have seen them somewhere, including the more useful difference plot before and after EQ. Maybe it's Maiky's awesome contributions such as this one: https://www.audiosciencereview.com/...hifiman-sundara-review-headphone.22529/page-2

 

[Sgt. Ear Ache]

It's not like the PEQ effects are particularly unpredictable. I mean if you are applying a 3db cut between 3khz and 4khz to counteract a 3db boost in the same range shown in the measurements you can pretty much assume what the outcome will be...

 

[luft262]

I know I have seen them somewhere, including the more useful difference plot before and after EQ. Maybe it's Maiky's awesome contributions such as this one: https://www.audiosciencereview.com/...hifiman-sundara-review-headphone.22529/page-2

Nice! That's what I'm talking about!

 

[luft262]

It's not like the PEQ effects are particularly unpredictable. I mean if you are applying a 3db cut between 3khz and 4khz to counteract a 3db boost in the same range shown in the measurements you can pretty much assume what the outcome will be...

True, but it could cause distortion. The point is if a cheap pair of headphones can hit the Harmon target with DSP there would be little objective reason to buy more expensive headphones that can do the same without DSP.

 

[dc655321]

I think many people would be shocked at the reality of just how little is gained by spending $4000 on a pair of headphones. I don't think it's at all out of the question that a much less expensive set could sound every bit as good...

Indeed.
I have ciems retailing for ~$2000, although I paid substantially less. I still apply EQ to get them to where I think they should be.

I did the the same to the iems I previously used, and they sound remarkably good too. And they’re an order of magnitude less $…

 

[charleski]

The principal reason is that the high-quality signal generator Amir uses doesn’t have the facility to produce an EQ’d output. He could, of course, set up a separate rig to measure EQ’d response, though this would be a lot of effort to produce information that can really be predicted from the base responses. You can estimate the distortion caused through boosting the bass by interpolating the distortion responses at different levels. In reality you’re going to listening at the 94dB level or less, so you know that a 10dB bass boost will bump distortion no higher than the 104dB line.

 

[luft262]

Indeed.
I have ciems retailing for ~$2000, although I paid substantially less. I still apply EQ to get them to where I think they should be.

I did the the same to the iems I previously used, and they sound remarkably good too. And they’re an order of magnitude less $…

Even a $150 set if cans like the AKG371 could probably be EQ'ed to within a hair's breadth of the quality of the industry leading Stealth's. I agree with Amir that we want to award good manufacturing practices and good companies, but I agree that, especially with DSP, the dimishing reruns are very steep.

 

[Jimbob54]

Just to clarify, @Maiky76 does not measure after eq, he generates his profiles using measurements (Amir and others) and shows the predicted post eq response. @oratory1990 does similar on his work over on Reddit. One or both I believe have said that remeasuring after eq would give results very similar to those predicted. I'm sure they will correct me if I'm mistaken.

 

[oratory1990]

Correct. As long as the characteristic curve of the headphone is sufficiently linear, it is reasonable to assume the headphone to do exactly what we expect. Meaning that when we feed it with 3 dB more energy into 300 Hz, the headphone will emit 3 dB more energy at 300 Hz.

How do we know that the characteristic curve is sufficiently linear? By looking at the distortion. If the harmonic distortion is low enough (lower than ~1-5%) we know that it must be linear enough for this purpose (if the characteristic curve were nonlinear you would necessarily see it via increased distortion)

 

[ADU]

Just to clarify, @Maiky76 does not measure after eq, he generates his profiles using measurements (Amir and others) and shows the predicted post eq response. @oratory1990 does similar on his work over on Reddit. One or both I believe have said that remeasuring after eq would give results very similar to those predicted. I'm sure they will correct me if I'm mistaken.

Completely forgot about Ora's PDFs, though I frequently use his graphing tool. Thank you for mentioning this, Jimbob54.

 

[ADU]

I've also posted a few before and after EQ plots of my DT-770's response in a couple threads here, which were also based on Oratory1990's graphs.

These used a simple sloping 2-point GEQ, rather than parametric EQs. Though the same approach could be used for either kind of filter (or filters). The DT-770 "After EQ" curve shown below came from the Analysis Panel graph in EAPO's Configuration Editor, as I described above. So it represents only an estimate of the response with EQ.

SLOPE EQ CURVE:

DT-770 FREQUENCY RESPONSE BEFORE SLOPE EQ:

ESTIMATED DT-770 FREQUENCY RESPONSE AFTER SLOPE EQ:

The DT-770's FR is represented by the bold black curves in the two graphs above. The other colored curves in the background are the same in both plots, and are combination of a number of different headphone responses from Ora's graphing tool that I'd consider to be a good reference or starting point for adjusting the response of my headphones.

I should also mention that the DT-770 FR curves shown above are based on an average of Oratory's Fresh and Worn DT-770 plots. And that I'll also use just the regular Fresh or Worn plots as well for EQ-ing, depending on the condition of the pads I'm using and what I think sounds best with them.

All of the headphone FR plots above are also compensated with Ora's diffuse field target compensation curve (in case that wasn't already apparent). And they are all normalized at a point in the middle of the 20 Hz to 20 kHz frequency range, at around 630 Hz.

 

[luft262]

Correct. As long as the characteristic curve of the headphone is sufficiently linear, it is reasonable to assume the headphone to do exactly what we expect. Meaning that when we feed it with 3 dB more energy into 300 Hz, the headphone will emit 3 dB more energy at 300 Hz.

How do we know that the characteristic curve is sufficiently linear? By looking at the distortion. If the harmonic distortion is low enough (lower than ~1-5%) we know that it must be linear enough for this purpose (if the characteristic curve were nonlinear you would necessarily see it via increased distortion)

Given your extensive experience do you feel a well measuring, but not perfect, inexpensive pair of headphones, using DSP, can reproduce music as well as an expensive pair of headphones that can match the Harmon Curve without EQ? Thank you for your insight!