Help with EQ and LCD-X

[BigTonkotsu]

I recently purchased the LCD-X 2021 revision and have the FR graph here:

I was hoping someone with more experience may have suggestions for EQ settings to get the FR closer to the Harmon target. Unfortunately the mids and upper-mids are quite recessed in my set, and I don't have much experience with EQ- Q factors are especially puzzling for me. I'll primarily be using Equalizer APO and Roon. Any help is appreciated, thanks!

 

[ProtegeManiac]

Simple way of doing it : Just boost 45hz and 2500hz or thereabouts. Unless Q has to be manually set then don't touch it because it will determine how and how much a particular curve type affects the frequencies other than your chosen center frequency. Start out with +3dB on both.

Complicated way to do it:

Band 1 : Low shelf, center freq 50hz, +3dB
Band 2 : High shelf, center freq 2000hz, +6dB

 

[BigTonkotsu]

Simple way of doing it : Just boost 45hz and 2500hz or thereabouts. Unless Q has to be manually set then don't touch it because it will determine how and how much a particular curve type affects the frequencies other than your chosen center frequency. Start out with +3dB on both.

Complicated way to do it:

Band 1 : Low shelf, center freq 50hz, +3dB
Band 2 : High shelf, center freq 2000hz, +6dB

Thank you for the recommendation! I actually tried simple shelf filters to start, and though it helped, I am trying to target the multiple recessions specifically. What I have so far is:

Band 1: low shelf, 33 hz, +2dB
Band 2: low shelf, 200 hz +2dB
Band 3: 650 hz, -1 dB, Q value: ? (I would like to target 550-750hz) i suppose 1.4-1.6
Band 4: 2.1khz, +5dB, Q: target 1.9khz-2.3khz
Band 5: 3khz, +9dB, Q: target 2.4khz-3.6khz
Band 6: 4khz +6dB, Q: target 3.2khz-4.8khz
Band 7: 8.1khz +5db, Q: target 7.5khz-8.7khz
Band 8: 11.8khz -2b: Q:narrow maybe- 3

I just dont know how to calculate the Q values and if ill end up boosting 2-5 khz too much.

 

[someguyontheinternet]

The AutoEQ Github is usually the place I start with. I tune by ear for my preference from there.
https://github.com/jaakkopasanen/Au...oratory1990/harman_over-ear_2018/Audeze LCD-X

 

[ProtegeManiac]

Thank you for the recommendation! I actually tried simple shelf filters to start, and though it helped, I am trying to target the multiple recessions specifically. What I have so far is:

Band 1: low shelf, 33 hz, +2dB
Band 2: low shelf, 200 hz +2dB
Band 3: 650 hz, -1 dB, Q value: ? (I would like to target 550-750hz) i suppose 1.4-1.6
Band 4: 2.1khz, +5dB, Q: target 1.9khz-2.3khz
Band 5: 3khz, +9dB, Q: target 2.4khz-3.6khz
Band 6: 4khz +6dB, Q: target 3.2khz-4.8khz
Band 7: 8.1khz +5db, Q: target 7.5khz-8.7khz
Band 8: 11.8khz -2b: Q:narrow maybe- 3

I just dont know how to calculate the Q values and if ill end up boosting 2-5 khz too much.

You don't need to target everything in the sense that you can end up with an even less flat curve in those regions due to how Q factor works. That slight dip is more likely to turn into a wide hump at Q1.0 or lower, or a weird peak at a higher Q value. Just target the bigger, problematic areas.

I take my headphones to as flat as possible on either side of 1000hz (stronger on the lower side), and on the HD600 that has more to do with flattening the 3500hz peak (the older versions have this annoying peak; newer graphs tend to not have it).

In your case if you're targeting the Harman Curve it should kind of look like how some people tend to do a stylized cursive V when signing their names (and depends on what the next letter is) so I just boosted everything with shelf filters. This is actually less complicated in some ways, but is more complicated in the sense that 1) it will require a more complex parametric EQ and 2) if you can control the Q factor, hooboy, it can get weird. If it's at around Q1 by default or you manually set it to Q1 everything above (high shelf) or below (low shelf) is boosted (or cut) by the same amount, then you have progressively weaker effect until the boost (or cut) fades out. Q1.4 or higher though at some point goes beyond fading to 0dB and actually introducing a narrow cut (if you boosted) or a narrow peak (if you cut).

 

[BigTonkotsu]

You don't need to target everything in the sense that you can end up with an even less flat curve in those regions due to how Q factor works. That slight dip is more likely to turn into a wide hump at Q1.0 or lower, or a weird peak at a higher Q value. Just target the bigger, problematic areas.

I take my headphones to as flat as possible on either side of 1000hz (stronger on the lower side), and on the HD600 that has more to do with flattening the 3500hz peak (the older versions have this annoying peak; newer graphs tend to not have it).

In your case if you're targeting the Harman Curve it should kind of look like how some people tend to do a stylized cursive V when signing their names (and depends on what the next letter is) so I just boosted everything with shelf filters. This is actually less complicated in some ways, but is more complicated in the sense that 1) it will require a more complex parametric EQ and 2) if you can control the Q factor, hooboy, it can get weird. If it's at around Q1 by default or you manually set it to Q1 everything above (high shelf) or below (low shelf) is boosted (or cut) by the same amount, then you have progressively weaker effect until the boost (or cut) fades out. Q1.4 or higher though at some point goes beyond fading to 0dB and actually introducing a narrow cut (if you boosted) or a narrow peak (if you cut).

Interesting. Again thank you for all the information and your help. So then maybe instead of targeting each dip in the upper mids maybe I will try this:

Keep the two low shelf filters of 33 hz +2dB and 200 hz +2dB. Add the high shelf, as you recommended, for 2khz +5dB. Then maybe lighter adjustments to the 3.5 k (+3dB) and 8.5 k (+4 dB) dips with a smaller Q factor, 1?, to gently round out the curve. And finally an additional high shelf filter at maybe 11.2khz -4dB to keep it from getting too peaky in the highs.

I just want be sure to get the 3.5khz dip up especially since that is a very noticable recession to the human ear. Maybe that is a simpler approach with less chance of a wonky FR. Or am I still overcorrecting?

 

[ProtegeManiac]

Interesting. Again thank you for all the information and your help. So then maybe instead of targeting each dip in the upper mids maybe I will try this:

Keep the two low shelf filters of 33 hz +2dB and 200 hz +2dB.

You understand that means you'll have an additional +2dB to 33hz and below and a little above that, right?

Add the high shelf, as you recommended, for 2khz +5dB. Then maybe lighter adjustments to the 3.5 k (+3dB) and 8.5 k (+4 dB) dips with a smaller Q factor, 1?, to gently round out the curve.

Putting all those boosts there wouldn't get as smooth a curve as you think.

If anything you have to make the Q higher at the 8.5khz to make the effect narrower instead of getting the 8.5khz trench boosted along with the peaks of the cliffs around it because again that's how Q factor works. You don't select just a single frequency, you're selecting a center frequency where the effect is centered but you're affecting frequencies around it too.

And finally an additional high shelf filter at maybe 11.2khz -4dB to keep it from getting too peaky in the highs.

A low Q (wide) boost at 3.5khz and 8khz coupled with a high shelf filter is more likely to make it piercing as you boost the 3500hz to 6000hz range several times over with all those overlapping boosts than whatever all those adjustments did to everything above 10,000hz since that aforementioned range is where non-time alignment sibilance tends to be at.

 

[BigTonkotsu]

You understand that means you'll have an additional +2dB to 33hz and below and a little above that, right?




Putting all those boosts there wouldn't get as smooth a curve as you think.

If anything you have to make the Q higher at the 8.5khz to make the effect narrower instead of getting the 8.5khz trench boosted along with the peaks of the cliffs around it because again that's how Q factor works. You don't select just a single frequency, you're selecting a center frequency where the effect is centered but you're affecting frequencies around it too.



A low Q (wide) boost at 3.5khz and 8khz coupled with a high shelf filter is more likely to make it piercing as you boost the 3500hz to 6000hz range several times over with all those overlapping boosts than whatever all those adjustments did to everything above 10,000hz since that aforementioned range is where non-time alignment sibilance tends to be at.

Yes, I was thinking two low shelf filters: one to level out the sub bass below 33hz and the 2nd to then lift the bass below 200hz closer to harmon.

I do understand that peak filters are for center frequency, and that Q factor effects an equal range around that frequency with larger Q being narrower range. I just am unfamiliar with how to calculate the proper Q for the frequency range I am seeking. I know there is an actual mathematical equation but I have not found a source that clearly explains that formula. I certainly want to avoid sibilance hence my adding the high shelf to attenuate the frequency response above 11k. So should my Q for 3.5k and 8.5k be around 3.5-4 with perhaps a smaller dB adjustment +2 - +3dB to avoid overcorrection? I think that may be closer in line with the information you've given. Thank you again for taking the time to give feedback!

 

[ProtegeManiac]

Yes, I was thinking two low shelf filters: one to level out the sub bass below 33hz and the 2nd to then lift the bass below 200hz closer to harmon.

I do understand that peak filters are for center frequency, and that Q factor effects an equal range around that frequency with larger Q being narrower range.

No, every kind of EQ type has a "center frequency," and even a peak boost or cut does not literally have the "center frequency" dead center in its area of effect.

As a very rough guide Q=~1.0 means if you set it to 3500hz, the effect will apply from 1750hz to 7000hz. Set it to 5000hz, and the effect is from 2500hz to 10000hz. You increase the Q value, and it can go to say 2625hz to 5250hz, or 3750 to 7500hz; decrease the value, and it can be something like 875hz to 10500hz or 1250hz to 20000hz.

The effect varies for low/high shelf filters. Q=1.0 on low sQhelf 45hz means you get +/-XdB at 45hz and below, but there is a gradually decreasing effect all the way to around 90hz; on high shelf 4000hz means 4000hz and higher gets +/-XdB, but there's a gradually fading effect until around 2000hz. Put that value to something like =1.8 and it gets really complicated. The +3dB at 45hz low shelf can have a fading effect until say 60hz, but then turns into a cut to around -2dB centered at 75hz before goes up to a boost at 82.5hz before finally settling down at 90hz.

These aren't exact figures, but I'm illustrating how complicated this can get well past how "center" on peak boost/cut isn't even literally dead center.

I certainly want to avoid sibilance hence my adding the high shelf to attenuate the frequency response above 11k. So should my Q for 3.5k and 8.5k be around 3.5-4 with perhaps a smaller dB adjustment +2 - +3dB to avoid overcorrection? I think that may be closer in line with the information you've given. Thank you again for taking the time to give feedback!

Your problem here, like I previously discussed, is that sibilance or any overbearing treble is more likely to come from a strong 3000hz to 8000hz than anything above 10000hz.

That's why not everybody prefers the Harman Curve. The Harman Curve to me is basically some bass, as opposed to what the graph suggests is "lots of bass," because it's getting overwhelmed by this.

 

[BigTonkotsu]

No, every kind of EQ type has a "center frequency," and even a peak boost or cut does not literally have the "center frequency" dead center in its area of effect.

As a very rough guide Q=~1.0 means if you set it to 3500hz, the effect will apply from 1750hz to 7000hz. Set it to 5000hz, and the effect is from 2500hz to 10000hz. You increase the Q value, and it can go to say 2625hz to 5250hz, or 3750 to 7500hz; decrease the value, and it can be something like 875hz to 10500hz or 1250hz to 20000hz.

The effect varies for low/high shelf filters. Q=1.0 on low sQhelf 45hz means you get +/-XdB at 45hz and below, but there is a gradually decreasing effect all the way to around 90hz; on high shelf 4000hz means 4000hz and higher gets +/-XdB, but there's a gradually fading effect until around 2000hz. Put that value to something like =1.8 and it gets really complicated. The +3dB at 45hz low shelf can have a fading effect until say 60hz, but then turns into a cut to around -2dB centered at 75hz before goes up to a boost at 82.5hz before finally settling down at 90hz.

These aren't exact figures, but I'm illustrating how complicated this can get well past how "center" on peak boost/cut isn't even literally dead center.




Your problem here, like I previously discussed, is that sibilance or any overbearing treble is more likely to come from a strong 3000hz to 8000hz than anything above 10000hz.

That's why not everybody prefers the Harman Curve. The Harman Curve to me is basically some bass, as opposed to what the graph suggests is "lots of bass," because it's getting overwhelmed by this.

Lol. Wow. I had no idea shelf filters still affect both sides of the target frequency. I thought low shelf only adjusted dB below the target frequency and vice versa for high shelf. That certainly does muddy things further. Thank you for your patience with me btw- I am finding, through this discussion, that at least for now I should keep it simple and do a deep dive into studying EQ later when I have the time.

Ok, so in terms of what I am trying to do: leveling out the sub-bass, raising bass below 200hz, raising mids/upper-mids/treble, correcting the 3.5k and 8 5k recessions, and lowering the 11k-12k peak- you would recommend just the two shelf filters. Is there any merit whatsoever in adding mild adjustments to the two mid recessions and one treble peak? Perhaps leave the 8.5k and 11.5k variances alone and just trying to bring up the 3.5k range (very mildly)- since the human ear is very sensitive to dips in that frequency?

 

[ProtegeManiac]

Ok, so in terms of what I am trying to do: leveling out the sub-bass, raising bass below 200hz, raising mids/upper-mids/treble, correcting the 3.5k and 8 5k recessions, and lowering the 11k-12k peak- you would recommend just the two shelf filters. Is there any merit whatsoever in adding mild adjustments to the two mid recessions and one treble peak?

I'd simplify it to just one of those options and generally avoid too many EQ overlaps.

1. High shelf and just boost 2500hz and above uniformly. You'l still have those dips, but then you can get most of that range up.

2. Correct the 3500hz and 8500hz dips to bring them up to where the high points above them are and slightly raising them, but if you go too far raising these chasing the Harman curve, you can end up reversing that curve and the former high points will be left as dips. If what you're shooting for is a smoother curve, this is what you go for, but again this is for smoothening that curve, not for raising it up to a Harman Curve.

Perhaps leave the 8.5k and 11.5k variances alone...

I wouldn't really worry about 11500hz. Unless it's a super narrow peak that will make it more obvious vs surrounding frequencies (in which case unless I'm using a high shelf filter I'm more likely to just cut that frequency at a high Q) or really just super high vs 1000hz, anything above 10000hz is not going to be super sharp to your ears. 3000hz to 6000hz/8000hz peaks are more audible for sibilance. In which case for my ears boosting 8500hz far enough can actually result in sibilance.

...and just trying to bring up the 3.5k range (very mildly)- since the human ear is very sensitive to dips in that frequency?

Depends on how sensitive your ears are.

If you boost it up to roughly where 2000hz and 5000hz are then you'll hear a difference in tone even without taking it up to full Harman Curve.

By contrast if there's any strong response at that range my brain hears scratching on blackboard. For example my KZ ZSN is just painful without EQ on my smartphones, but even on my AP80 the rough edges are rounded out but that range is strong enough ie Harman Curve that I can't crank it up to hear the bass as I end up hearing more of that range as the output level goes up. Good thing this is just my beater IEM - my Westone2 is smoother above 1000hz, and so is my Aurisonics ASG1.3, but despite the higher sensitivity on these (even with all measured at 1000hz or 1200hz) I can crank them up higher and hear more of the bottom end (which is far more susceptible to being affected by ambient noise).

 

[dan3952]

I recently purchased the LCD-X 2021 revision and have the FR graph here: <graph omitted> I was hoping someone with more experience may have suggestions for EQ settings to get the FR closer to the Harmon target. Unfortunately the mids and upper-mids are quite recessed in my set, and I don't have much experience with EQ- Q factors are especially puzzling for me. I'll primarily be using Equalizer APO and Roon. Any help is appreciated, thanks!

My curve resembles yours somewhat.

I would do the following, in order of importance:

1. Raise the normal Audeze dip at 4 kHz or so, with not more than 2.5 db of gain in your case. Mine had a dip in the same place, but more severe.
2. Repair the dip between 1.4 kHz and 2 kHz, flattening it out as much as possible with a high shelf or a low shelf.
3. Drop the peak at 11 kHz, but keep in mind you get only percussion overtones at that frequency. I can't hear above 13.5 kHz anymore, due to my age. I thus ignore any peaks above 12 kHz. When I was a teenager, it was probably 19 kHz.
4. Low shelf the dip below 35 Hz, but keep in mind there is little content in the music there, and you won't hear this change unless you listen to a lot of electronic instruments. Planar headphones already have curves that are far flatter than that of most dynamic headphones below 100 Hz. The lowest note on a bass guitar is at 41 Hz. The lowest note of a pipe organ at a church, is 28 Hz.

I had ignored the bump at 700 Hz, it's not audible to me. There's already a gentle rise in the mid bass, which I hadn't low shelved any further.

I use a plugin called Frequalizer in JRiver (for the visuals), and then their built in Parametric Equalizer. Which frequencies are most active, fundamentals , and overtones will become more apparent if you use a real time analyzer. I had learned a lot.

 

[dilbertprogrammer]

Below is the Parametric Equalization for the Audeze LCD-X from the AutoEQ project.

Preamp: -5.49 dB
Filter 1: ON PK Fc 21 Hz Gain 2.9 dB Q 1.05
Filter 2: ON PK Fc 226 Hz Gain -3.0 dB Q 0.71
Filter 3: ON PK Fc 818 Hz Gain -4.3 dB Q 1.28
Filter 4: ON PK Fc 3106 Hz Gain 6.6 dB Q 1.48
Filter 5: ON PK Fc 8818 Hz Gain 4.9 dB Q 5.39
Filter 6: ON PK Fc 62 Hz Gain 0.1 dB Q 1.56
Filter 7: ON PK Fc 1098 Hz Gain -1.1 dB Q 8.35
Filter 8: ON PK Fc 4085 Hz Gain 3.1 dB Q 7.04
Filter 9: ON PK Fc 10489 Hz Gain 5.8 dB Q 1.27
Filter 10: ON PK Fc 19959 Hz Gain -15.3 dB Q 0.21

Hope others find this helpful!

 

[albertmuc]

After some EQ tinkering I arrived at Amirm‘s EQ, I like it.

 

[szore]

My simple EQ for the X.