EQ is too generic a term. Fixed freq vs Variable freq. A 6 band parametric is vastly more flexible for adjusting a speaker/headphone in the FR domain than a fixed freq EQ of even 31 bands.
I use 6 and 10 band PEQ’s on every headphone I have, even ones that are supposed to be “flat”. I start out with Harman 2018, but I always lower the settings under 100 Hz and in the 2.5-4k area since I find the Harman tuning flawed as is the methodology they use to derive it.
How do you make the distinction between Harmon being fundamentally flawed and you being one of the people who simply don’t like the Harmon curve?
My understanding is Harmon is a “most people” thing, not “everyone must like this”.
I don’t recommend it to most users. But I still use EAPO’s Configuration Editor. It’s not the most user-friendly. But it is fairly powerful/flexible, and offers quite a few different types of filters, including both PEQ and GEQ. I use mostly the latter. And I like the stacking feature in EAPO, which allows me to see the effects of my EQ filters on my headphone’s FR curve, among other things.
For the few here who might be interesting, there is a topic on Head-Fi that covers most of the basics on using the interface. I learned how to use it mostly through trial and error.
This is still a work in progress, but here’s an example GEQ curve I’ve been working on for my DT-770’s in EAPO’s Config Editor. The narrow band peaks may be too high. But it doesn’t sound that bad…
My target is the original 1/3-octave HBK 5128 DF curve, stacked with a sound power slope in the -1.0 to -1.5 dB/octave range. This target is pretty consistent with the responses of neutral and Harman-tuned headphones measured on the 5128. And it seems to work quite well, based on tests I’ve done so far with the DT-770. I refer to this target as the DF+SP model, for simplicity.
The DT-770 is not an ideal headphone for this kind of thing though, because it has what appear to be cancellation effects in both the bass and upper mids that are hard to correct with precision (if at all). And because pad wear seems to influence the FR fairly noticeably. I try to take these factors into account as best I can though.
GEQ’s can be either fixed or variable frequency in EAPO’s Configuration Editor. I use both types, and sometimes convert between them. The EQ curve above uses two variable GEQ filters, for example. One is a simple -0.6 dB/octave slope with just two points at 20 and 20k Hz. The other uses 24 points at various frequencies designed to match features I want to correct in my headphone’s response curve.
I tend to find GEQ’s easier and faster for my needs. But would encourage users to try both PEQs and GEQs. I generally wouldn’t recommend using EQ curves with such prominent high Q peaks as the ones shown on my graph above btw.
Well I was weaned on live unamplified classical music at home (piano and violin of parents) and my fathers Quad ESL-63’s and going to the BSO many times from age 11 up.
The Harman curve is not an absolute, it has elements of user preference, and also uses a crap room with bass nulls and peaks as a guide.
As a side business I improved, rebuilt, and had built from scratch audio rooms for about 20 years.
So, now that I’ve put it out there that I have educated ears according to others, I make the claim for myself. It’s flawed IMO - take it or leave it.
Also I am a physical giant, with big bones, and its likely I hear bass more than the average person which could account for some extent my dislike of bass cannons.
I have migraines as well, so very bright trebles are a no no as well.
My gear speaker: BMR Philharmonic Monitors and ProAC Tablettes with a ribbon tweet (Fountek X3, and a ScanSpeak 12RC w/ Jantzen caps.
Headphones: HFM 6 SE/500, Senn HD-600; looking at new ones in the $1.8k+ range. HFM HEK SE or ZMF Auteur (a clone of the two would be nice), or other.
Interesting, do you tune by your own ears or do you have developed your own target curve and if so where do you deviate from Harman?
Yes I do, i use about 20 cuts to get it in the area (about 5 cuts off the Chesky sampler, Norah Jones, various classical string quartets, solo piano, Mapleshade label jazz stuff) then live with it long term, and make adjustments not for one cut but if several types of music seem to have the same problem area(s).
My ear would be classed as Audiophile and also the not very used term “New England Sound” which is fast deep but not overhung bass, flat mid bass (maybe a bit dry for most), and slightly down in the 2-4k area.
I also adjust beyond FR for things like CSD/ringing in the treble (common for planars) since you cannot separate that from FR, I use very steep settings Q of 6-9 and usually no more than 2.8 db so that the sound I’m getting is akin to what a headphone or speaker would be doing w/o that problem w/o the CSD. My HE-500 and HE-6 SE I probably spent 6 months each and about 40 settings until I got what I wanted. My new speakers (BMR Philharmonic Monitors) will be much easier - room bass nodes, and maybe a db or 2 down around 3 k - time will tell but it will be easier then planar cans.
Another version of my DT-770 EQ curve in progress, superimposed over the previous curve. The high Q peaks are more blunted in this version. And there are a few minor changes in the tonal balance.
A -0.6 dB per octave slope is also shown in blue. And the curves are all normalized at -5 dB at 630 Hz in the midrange.
For various reasons, I don’t try to precisely target or correct all of the finer-grained and “high Q” details in both the lower and higher frequencies. Especially above 16 kHz, where I believe my DF+SP target significantly overshoots a neutral response!
The DF curve that I use for my target is still HBK’s 1/3-octave 5128 DF btw.
Current 250-ohm DT 770 EQ, still under construction.
Instead of just using HBK’s 1/3-octave DF with a slope or speaker’s sound power for my target, I’m now working with the average response of several headphones that are good approximations to the DF+SP model as my target. The benefit to this is that it offers more detail in the treble, and more realistic falloff in both the higher and lower frequencies than HBK’s DF and a slope.
These changes aren’t reflected yet in the above curve. But this is what I’ve gradually been working toward all along. And it’ll hopefully allow a little better precision in my EQ adjustments. So far, the results on my end are pretty encouraging. And it should only get better as more headphones close to the DF+SP model become added to the sampling of the average FR curve.
I’ve deleted a few of my previous graphs here btw, because of their similarity to Rtings’ new compensated 5128 FR graphs of the 250-ohm DT 770. This was an unintentional gaff on my part. So my apologies to them.
