AutoEQ - Headphone Equalizer Presets

Where do I begin? Jaakkopasanen is a beautiful human who takes measurements from review sites like Rtings and makes EQ presets for graphical/parametric correction. Headphones are generally fit to the Harman Kardon curve of the year.

It won’t fix a really bad set of headphones, but if I have a pair that is fatiguing, I’ll throw some presets into Pulse Equalizer (linux) or HeSuVi (windows). His preset takes the T60RP’s, which are a tad fatiguing for me, and makes them a butter boat. It’s likely that my ear just prefers the Harman curve, as the half-dozen headphones I’ve listened to have all benefited from AutoEQ (the Tin T2’s the least).

He makes adjustments to his process, and recommended results are below:

3 Likes

This is quite the find. I just gave his AKG K712 EQ a try and the results are pretty good—much better than what I could came up. I would’ve really liked to see a Focal Elex EQ, but he didn’t have data for it for some reason.

Yeah, the only downside is that he is limited by the available measurements. For instance I’m waiting for the Coolermaster HM751/752 measurements to come out on Rtings.com so Jaakko can make an EQ preset.

On the other hand, I could buy a MiniDSP EARS system, which I’ve heard good things about – particularly for over/on-ear headphones. The trouble with IEM measurements is that depending on the length of the ear canal you get a node at different frequencies. I hear the EARS system has a shorter canal than the anatomical average. But I could be wrong, and the calibration files may account for this. Suffice it to say, if I had the extra $200 laying about, I’d buy the EARS and make filters with REW set to the Harman Kardon house curve.

To the extent that Rtings provides some of the measurements that Jaakko uses, I trust the source measurements. Rtings.com released a video showing their frequency response methods, and I’m very impressed with their methodology. They use ear canal blocking microphones to average 5 measurements across 5 listeners per headphone, to account for variance in pinna shape.

I’m trying to use this with Peace and I can’t figure out an easy way to create compatible txt files. I know there must be an easier way than copying the raw data and editing each line

Jaako explains that here: https://github.com/jaakkopasanen/AutoEq#usage

1 Like

I haven’t tried Peace – I use HeSuVi as the software that runs on top of Equalizer APO. I add a .txt with the graphical eq data to C:\Program Files\EqualizerAPO\config\HeSuVi\eq\Custom and select it in HeSuVi.

A few points.

The provided EQs are done using a different target for each measurer. Rtings EQs will give you an EQ based on Rtings’ own house flavour of the 2013 Harman target, while the Oratory1990 EQs will use the Harman 2018 target, for example. I suggest (as does Jaako) using the result as a starting point, especially to raise or lower the bass and highs to your own preference.

Also, these same data can be used to simulate headphone A on headphone B. I use a DT 1990 as my go-to headphone. But if I want to get some idea what an HD 660 S or an LCD-2C sounds like, I use that headphone as the target instead of something like a Harman curve. I’d be happy to help anyone get up to speed on this (or anything else that’s related).

1 Like

The recommended results are listed on https://github.com/jaakkopasanen/AutoEq/tree/master/results.

Please share how you’re using the data to make target curves that make one headphone’s frequency response sound like an other’s – that sounds like fun.

There is a CSV file provided for each headphone. This contains the raw numbers for all 600+ measurements with frequency in one column and the measurement for that frequency in the next column. So:

  1. download the CSVs for two headphones – I’ll call the one you own A and the one you want to emulate, B
  2. import them into your spreadsheet program
  3. copy the raw column from headphone B as a new column in the spreadsheet table for headphone A (or vice versa)
  4. create a new column with the formula to subtract the value for headphone A’s measurement from headphone B’s measurement at the relevant frequency in each cell in the new column.

That gives you the EQ curve you want to match in your equalizer software. I do my listening on a MacBook, so I’m limited to using a 31-band (1/3 octave) GEQ. But it’s not a big limitation, since the results are quite good. On a Win box I could use Peace GUI’s preset for 31-band input. But I prefer to use the graphical interface in Peace for this. Either way, I manually input the 31 dB values for each of the 31 frequencies (20 Hz, 25 Hz, 31.5 Hz, etc.). Set the pre-amp value to something sensible.

I assume one could go to the next level and automate in the spreadsheet to generate a text file suitable for EQ-APO or Peace input. In that case you could have the spreadsheet spit out a file containing all 600+ frequency values. I believe EQ-APO would ingest this quite happily. I guess one would use the default 1.41 Q value for each band?

If the above isn’t clear or sounds too complicated, let me know. These things are easier done than said in most cases.

BTW, if anyone is put off by the spreadsheet aspect of this but is fine with the EQ side, I’m happy to generate the 31-band numbers for anyone’s use. (I’m assuming this is so esoteric that I’m not likely to be swamped with requests, lol.)

That’s brilliant, thanks for sharing. I’ll give it a shot some time!