• 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!

how to eq harsh upper mid - high frequencies?

I looked at some measurements of those speakers... I'd say they would very much benefit from eq'ing based on anechoic or quasi anechoic (gated) measurements with the aim to get a flat on axis response. I never touch anything over 500hz based on measurements taken at listening position in room, IME that never works. The room curve should be what a speaker w. flat on axis response produce in the room w. room bass problems fixed
 
Actually, I'm more curious about how the merged response curve appears within the Dirac Live program. REW measurements don’t provide the complete picture of how Dirac interprets or processes the multiple measurements.
and here is the problem i got in my room shown in REW wavelet. Too much energy in the high freqs

no EQ:
Wavelet without EQ.jpg


and after Dirac (not as much energy in the higher freqs but still):
wavelet after Dirac.jpg


@UprightCitizen
I'd say they would very much benefit from eq'ing based on anechoic or quasi anechoic (gated) measurements
but doesn't these wavelets show that my problems is with my room acoustic response? I don't think the JBL sound as harsh as they do in my room.
 
Last edited:
not sure which one you'd like to see so i made multiple snapshots. One measure is with 0deg UMIK1 and the other with 90deg UMIK1
Thanks.

I can't see which is 90° and which is 0°. You should use 90° as recommended by Dirac.

Looking at your curve, I’d suggest three adjustments:
  1. Drag the bottom curtain up to around 30 Hz, as this seems to be where the speakers naturally roll off. Currently, you’re forcing them to play below their intended range by boosting.
  2. The applied curve is boosting frequencies between 2.8 and 10 kHz, which likely isn’t ideal if you’re aiming to reduce harshness.
  3. As a starting point, turn off Dirac’s auto curve and apply one that closely follows the underside of the response curve while still maintaining the contour of the recommended curve.
1737351207439.png
1737351131912.png
 
So i did another day of measuring and testing and still have the same problems. High freqs are too harsh.
Dirac results are worse than REW EQ results with miniDSP PEQ afterwards.

Purple is no EQ, orange is Dirac and turquoise is REW EQ + miniDSP PEQ. With REW EQ + PEQ i can enjoy my speakers but the sound is muffled.

dirac_rew_eq.jpg


i don't know what to do next.
 
As you measure with REW, look at the waterfall graph of this speaker, and you will see where the trouble is and why you can't eq it out....
 
Prologue: make sure you have enough amp for your speakers and your intended listening volume. Consider that you need more headroom when you start to add gain on the EQ.

1. Direct sound. EQ your speakers for flat on axis response (not including reflectections from the room!), then make sure they are angled so that your ears are on axis.

2. Speaker placement and acoustic treatment. Find the best spot for your speakers and yourself. Many use an equilateral triangle. Myself, I tend to prefer a bit narrower stereo angle, about 45-46* instead of 60* (but that's another discussion).

3. Measure listening position with REW and sort out bass peaks, and fill out dips if it works for you. This is usually (for me at least) an iterative process. Measure>EQ and repeat. Look at the harman room curve for ideas how your target below the schroeder frequency could be. I'd say EQ for room problems is usually perfectly fine up to 200-300hz at least, but I never like it when I go higher than 500hz.

(3.5, take break and clear your head. Do some palette cleansing.)

4. Buy speakers with flat on axis FR, consistent off axis FR and low distortion.
 
Thta's the question! Before i spend money now on this Denon X3800H + Dirac Live license (which i can't refund or test?) would be there a better way? A dsp for example?
I believe that Dirac is offering 2 week free trial. I wasn't 100% sure but found it:


It is under the options for various packages.
 
As you measure with REW, look at the waterfall graph of this speaker, and you will see where the trouble is and why you can't eq it out....
You mean like this (attached with and without EQ)? How can i see it there?

make sure you have enough amp for your speakers
Hypex NC250 amp. 2x250 watt 4 Ohm

Direct sound. EQ your speakers for flat on axis response
how do i do that?

Speaker placement and acoustic treatment.
did all the placements and room treatment i could

Buy speakers with flat on axis FR, consistent off axis FR and low distortion.
now i have bought these expensive jbl and have to find a way to make it work with them

I believe that Dirac is offering 2 week free trial.
i already bought miniDSP and Dirac..


thanks guys
 

Attachments

  • waterfall with EQ+PEQ.jpg
    waterfall with EQ+PEQ.jpg
    197.7 KB · Views: 20
  • waterfall without EQ.jpg
    waterfall without EQ.jpg
    190.7 KB · Views: 24
I used one of these:
Rotated the Treble control a few degrees counter-clockwise. Problem solved. Loudness control is pretty cool too. I can always tone defeat or bypass the unit by switching to XLR input instead of RCA on my pre-amp. Anyway, worked for me. SINAD is reasonable, sounds great.
 
What does your Dirac curve look like in your latest attempt?
You should aim to create a sloping curve similar to the one I shared earlier (the green line). It doesn’t seem like you’ve achieved that yet, or if you have, it might need to be steeper to address some of the harshness you’re noticing.
Experimentation is key -Dirac isn’t a one-click solution for perfect sound.
 
What does your Dirac curve look like in your latest attempt?
You should aim to create a sloping curve similar to the one I shared earlier (the green line). It doesn’t seem like you’ve achieved that yet, or if you have, it might need to be steeper to address some of the harshness you’re noticing.
Experimentation is key -Dirac isn’t a one-click solution for perfect sound.
I tried your green line and many variations of it. Even much steeper but nope, no good results. The orange line in my rew pic is your green line but like -5db steeper. It seems like dirac doesn't let it go that deep
 
I tried your green line and many variations of it. Even much steeper but nope, no good results. The orange line in my rew pic is your green line but like -5db steeper. It seems like dirac doesn't let it go that deep
Maybe try to convert shelves to points (right mouse click on curve and convert to points) and apply steeper curve which follows your speakers natural roll-off. Check if curve follow natural roll-off for Both speakers.
Try to adjust curtains and don't let Dirac boost HF.
IMG_20250123_123207.png


With control points You will be able to create your own slope, remove points between 1kHz and top end, adjust slope moving last point or add points every octave from 1kHz, 2kHz, 4kHz, 8kHz for more control and your own taste:
create own slope and points.jpg
 

Attachments

  • Dirac control points.jpg
    Dirac control points.jpg
    122.2 KB · Views: 14
  • Dirac convert to points.jpg
    Dirac convert to points.jpg
    133.7 KB · Views: 16
  • Dirac HF boost.jpg
    Dirac HF boost.jpg
    132.8 KB · Views: 17
  • Dirac HF natural roll off.jpg
    Dirac HF natural roll off.jpg
    134.5 KB · Views: 15
  • Dirac NO boost.jpg
    Dirac NO boost.jpg
    261.2 KB · Views: 17
  • Dirac shelves.jpg
    Dirac shelves.jpg
    130 KB · Views: 15
Last edited:
i think i did it!

I made a new measurement with Dirac but now with a tight pattern and changed the curve as you told me and now it sounds crazy good. Not only is the harshness almost gone (in some songs it's still there but not as harsh), the sound in general is much more linear and "right" sounding and not muffled! Fantastic!

Thanks to you all guys

Bildschirmfoto 2025-01-24 um 13.41.04.png



rewdirac5.jpg
 
i think i did it!

I made a new measurement with Dirac but now with a tight pattern and changed the curve as you told me and now it sounds crazy good. Not only is the harshness almost gone (in some songs it's still there but not as harsh), the sound in general is much more linear and "right" sounding and not muffled! Fantastic!

Thanks to you all guys

View attachment 423508


View attachment 423509
Great to hear! Thanks for the update. View Dirac as a tool rather than a magic solution.. It needs a bit of trial and error.
 
Switching amps won't solve a problem like this, unless the original amp was broken or an old/cheap load dependent class D. EQ can solve this, but the statement

is kinda weird, because there isn't one EQ in higher frequencies. There's a near infinite number of setting combinations so you couldn't possibly judge them all by testing a handful or even just a single one.

That being said, your room with a wood ceiling, laminate flooring and no big windows looks pretty good to me. I would recommend you start with two simple things: Rotate the speakers outwards in 5° steps and see if that reduces the hot treble. If this doesn't help, add a medium sized carpet in front of the speakers and below the table. You could just borrow one from another room or use a big blanket for testing.

If both of those options fail, you can think about EQ. Due to the amount of options an EQ gives you, it is very powerful but can also be time consuming and tedious to tune. That's why I would initially recommend the two "easier" steps above.
Why is laminate flooring good? I would think it quite reflective.
 
Why is laminate flooring good? I would think it quite reflective.
I didn't say that laminate flooring specifically is good. But the combination of materials in that room is far from a worst-case scenario. A wood ceiling with large beams is okay, small-ish window surface area is great, laminate flooring is okay. I would assume that it's better than tiles, concrete or marble, but worse than carpet. Didn't check any data on this, though.
 
Back
Top Bottom