[ADUHF]

The only aggression I hear is in the upper mids, but I only find it slight and dependent on the recording. I’ve pretty much gotten used to it now...

I think the problem is indeed in the 7-8 kHz range so that it also causes the upper mid to be raised

To both of your points, there does appear to be a sort of broad hump in the upper midrange formed by the two peaks at around 3 and 4.5 kHz, and the depressions at about 1.5 and 6 kHz. I've greyed out the rest of my plot, and also added a dashed yellow line between the two peaks to try to make it a bit easier to see what I'm referring to above.

If you want to smooth that area out a bit more (in addition to lowering the peak in the treble at ~7.5-10 kHz), then you could try lowering the peak at about 2.5-3 kHz by maybe a dB. And maybe also raising the areas of 1.25-1.5 kHz, and 6 kHz by a dB or so.

I would not try to do anything about the small notch between the two peaks at 3 and 4.5 kHz. Imo, that notch is probably too small in amplitude and bandwidth to really be noticeable.

 

[Ashimaru]

Deleting

 

[Ashimaru]

To both of your points, there does appear to be a sort of broad hump in the upper midrange formed by the two peaks at around 3 and 4.5 kHz, and the depressions at about 1.5 and 6 kHz. I've greyed out the rest of my plot, and also added a dashed yellow line between the two peaks to try to make it a bit easier to see what I'm referring to above.

If you want to smooth that area out a bit more (in addition to lowering the peak in the treble at ~7.5-10 kHz), then you could try lowering the peak at about 2.5-3 kHz by maybe a dB. And maybe also raising the areas of 1.25-1.5 kHz, and 6 kHz by a dB or so.

I would not try to do anything about the small notch between the two peaks at 3 and 4.5 kHz. Imo, that notch is probably too small in amplitude and bandwidth to really be noticeable.

Thanks for the explanation. But can you implement it in the Equalizer APO so that what you mean is easier for us to use?

 

[iFi audio]

Take the iFi Hip-Dac, or any of their line pretty much and they tilt more to the warmer side. They don't sound too warm though, it just adds a bit more body to everything like the lower mids and tames the bright spot

You nailed it. Just a touch of warmth is how we like it. Thanks!

 

[jasonb]

This is how I EQ mine. It's basically the autoEQ which gets it even closer to the Harman target than they already are. AutoEQ reduces the sub-bass by about 4dB which I didn't enjoy, so I modified the autoEQ profile to not have the sub-bass cut. This is based off of oratory1990 on Reddit's measurements of the K371 vs the Harman target. These just aren't the same fun headphone with the sub-bass boost taken away. With people talking about brightness they'd probably hate this EQ, but I listen at low volumes and like the extra detail and clarity.



On this chart -6 is 0 gain or cut. So a value higher than -6 is an EQ boost, any number lower is an EQ cut.

 

[ADUHF]

Thanks for the explanation. But can you implement it in the Equalizer APO so that what you mean is easier for us to use?

What is the Equalizer APO, Ashimaru? I'm not familiar with it.

 

[Deolum]

Thanks headfi for making me discover that gem <3

 

[plakat]

Thanks headfi for making me discover that gem <3

You're in luck... normally those good ideas flowing from head-fi are much more expensive...

 

[ADUHF]

This is how I EQ mine. It's basically the autoEQ which gets it even closer to the Harman target than they already are. AutoEQ reduces the sub-bass by about 4dB which I didn't enjoy, so I modified the autoEQ profile to not have the sub-bass cut. This is based off of oratory1990 on Reddit's measurements of the K371 vs the Harman target. These just aren't the same fun headphone with the sub-bass boost taken away. With people talking about brightness they'd probably hate this EQ, but I listen at low volumes and like the extra detail and clarity.



On this chart -6 is 0 gain or cut. So a value higher than -6 is an EQ boost, any number lower is an EQ cut.

Forgive me for playing devil's advocate, but if you don't trust its response in the sub-bass, then why would you trust it anywhere else?

I have looked at some of Jaakko's plots, and think he's done some amazing things with his apps. But prefer a somewhat different approach, which is based more off of actual measurements than an idealized target (which may or may not be relevant to the measuring system being used).

Imo, the Harman curve was never intended to be used as a target for equalization purposes the way Jaakko and some others are employing it. It was intended simply as a very rough guide for a headphone response based on the subjective preferences of listeners, and the sound of loudspeakers in a room. No more, and no less.

If you want to base your equalization off Oratory1990's graphs, then here is the approach I would use. Create a list headphones which you believe are as close to the sound you're after as possible, based on your own listening experience, and (if necessary) the opinions of others you trust. Average the responses of all those headphone together. And then use that averaged frequency response curve as your target for equalization.

Whether you would get a substantially different result with this type of approach than the one you are using now, I can't really say for sure. Imo though, an approach like this can potentially deliver a much more accurate result than a more automated system (like AutoEQ) will. It may still result in some over (or under) compensation though, simply because EQ-ing is never a 100% exact science. And headphone measurements are never 100% reliable.

Using actual headphone measurements as a basis for your target response curve will do a better job of capturing, modeling and eliminating (where necessary) the specific resonant characteristics and other idiosyncratic properties of the system being used to capture all the measurement data than a more idealized and smoothed target (such as the Harman curve) would imho... Others may disagree.

The larger the list of headphones you use to compute your average response curve, the better, generally-speaking btw. Because a larger sampling will generally do a better job of distilling/filtering any unwanted noise (or other strange behavior) from the individual headphones out of your final result.

 

[ADUHF]

Fwiw, this graph shows the difference between some smoothed and idealized target response curves (Olive-Welti and Harman Over-Ear), and a curve based on actual headphone measurements (Average of 19 HPs). Imo, the Harman Over-Ear curve is probably the least accurate of the three, because it smooths over too much important information in the treble. And will result in over-compensation in some spots and under-compensation in others. The Olive-Welti curve (which is based on the preferred sound of loudspeakers in a room) would do a much better job imo, with a few small adjustments to the level in the bass, the position of the two peaks in the upper midrange and treble, and possibly also the roll-off in the upper treble. But it would still likely miss, and probably try to over (or under) correct a few of the smaller details, such as the notch at 10.5-11 kHz on the Average of 19 HPs plot. (Some smoothing to the final equalization curve could possibly fix that though.)

The Average of 19 HPs shown above is still very much a work in progress btw. And it is based solely on raw frequency response measurements from the headphone review site, Rtings. So it's really only applicable to other raw headphone measurement data from that site. And would not deliver a reliable result with measurements from another source (such as Inner Fidelity or oratory1990).

 

[ADUHF]

Also, the Harman curve is not just one curve. It is actually a family of curves developed for different demographics and headphone types.

The generic Harman over-ear target curve (shown on the previous graph) has a rise of about 6 dB's (give or take) in the bass from the lowest point in the midrange at around 200-250 Hz. If you prefer a little more bass, then you might try using the target curve for younger males as a rough guide for choosing some of the headphones for your list of headphones with an ideal response. I believe that curve is a couple dBs higher in the bass than Harman's standard over-ear curve.

 

[jasonb]

Forgive me for playing devil's advocate, but if you don't trust its response in the sub-bass, then why would you trust it anywhere else?

I have looked at some of Jaakko's plots, and think he's done some amazing things with his apps. But prefer a somewhat different approach, which is based more off of actual measurements than an idealized target (which may or may not be relevant to the measuring system being used).

Imo, the Harman curve was never intended to be used as a target for equalization purposes the way Jaakko and some others are employing it. It was intended simply as a very rough guide for a headphone response based on the subjective preferences of listeners, and the sound of loudspeakers in the room. No more, and no less.

If you want to base your equalization off Oratory1990's graphs, then here is the approach I would use. Create a list headphones which you believe are as close to the sound you're after as possible, based on your own listening experience, and (if necessary) the opinions of others you trust. Average the responses of all those headphone together. And then use that averaged frequency response curve as your target for equalization.

Whether you would get a substantially different result with this type of approach than the one you are using now, I can't really say for sure. Imo though, an approach like this can potentially deliver a much more accurate result than a more automated system (like AutoEQ) will. It may still result in some over (or under) compensation though, simply because EQ-ing is never a 100% exact science. And headphone measurements are never 100% reliable.

Using actual headphone measurements as a basis for your target response curve will do a better job of capturing, modeling and eliminating (where necessary) the specific resonant characteristics and other idiosyncratic properties of the system being used to capture all the measurement data than a more idealized and smoothed target (such as the Harman curve) would imho... Others may disagree.

The larger the list of headphones you use to compute your average response curve, the better, generally-speaking btw. Because a larger sampling will generally do a better job of distilling/filtering any unwanted noise (or other strange behavior) from the individual headphones out of your final result.

I get what you're saying but I like what I'm hearing so it's good enough for me.

 

[ADUHF]

I get what you're saying but I like what I'm hearing so it's good enough for me.

All that really matters.

 

[jasonb]

All that really matters.

This EQ makes them sound smoother as in smoothing the dips and peaks, and brighter which I like. I have always been EQ'ing these to be brighter.

 

[ADUHF]

This EQ makes them sound smoother as in smoothing the dips and peaks, and brighter which I like. I have always been EQ'ing these to be brighter.

Understood.

Imo, your ears are always the best guide. And as mentioned in my previous posts, I believe there are also a few spots in the upper treble, and in the 6 kHz and 1.25-1.5 kHz areas where the K371 appears a bit too depressed, and lacking in some brightness. While there are other areas (such as around the 8-10 kHz region) where it is probably a bit too bright. And it doesn't appear to me like the EQ curve you're using from AutoEQ would necessarily be ideal for making adjustments to those particular issues. That is really all I'm saying here.

I have only listened to these headphones a few times at Guitar Center though (including yesterday), under less than ideal conditions. And am basing my opinions on how to perform the corrections mostly on my computed difference curve (based on the Average of 19 HPs), and to some extent also Rtings compensated graph. So my suggestions could be totally wrong, or off base, or "bogus" (as the kids like to say). The only way to know for sure though is to try some of them.

Fwiw, I'm not really trying to advocate for one approach or another here (even though it may seem that way). I frankly don't care what approach anyone else uses to correct or compensate their headphones or other equipment,... unless they're using it to produce music that I (and others) may eventually want to listen to on our correctly calibrated and adjusted gear. That is what tone controls are for though.