Confused at correcting headphones/speakers using EQ
Apr 21, 2020 at 8:56 PM Post #16 of 51
You don't have to measure your cans yourself. Use published response curves and calibrate to Harman as best as you can. That is a good starting place. Listen to it like that with a variety of music and take note of what you like and don't like. Make a small correction- just one at a time, not a bunch all at once. Listen to it like that for a while and take note again. Rinse and repeat until you are happy. Odds are, your preference isn't much more than a few dB off of Harman. And the areas that you will probably be tweaking are between 2kHz and 5kHz and the bass range. As you get experience EQing, you learn what sounds the Hz and kHz numbers refer to. That makes it easier to make adjustments.

There are two Harman curves. I'd suggest using the newer one with more bass.

EQ only does more harm than good if you don't know how to use it. I think it's a very good idea to experiment and try to puzzle it out. That is how to learn. You don't get anywhere if you make excuses like "EQ can't fix everything." and "You can't test your cans yourself." Just get a good equalizer and try to learn how to use it well. It is the best way to improve the sound of just about any system. Response is the lion's share of fidelity nowadays.

  1. So how can I get the newer Harman curve, load it to EQ APO/Peace and change the parametric EQ and see how the headphones response curve changes in real-time?

  2. I believe I need to overlay the Harman curve on a graph window, then initially I will my headphones EQ as a flat line and then I need to play with the EQ values to get it as close as possible to Harman's curve?
I guess that's the "How To" that I'm missing. Where do I get the Harman response curve? how do I overlay it?

I found the Innerfidelity curve for my ATH-A900X (https://github.com/jaakkopasanen/Au...af-serious/Audio-Technica ATH-A900X/README.md), Thanks to @jaakkopasanen

But not for my AudioEngine 5+ speakers.

I guess that I need to import the parametric EQ to Equalizer APO, but then try to play with the EQ to get to Harman's curve but I'm not sure how to load that curve and try to match them.
 
Apr 21, 2020 at 9:30 PM Post #17 of 51
You line up the response curve measurement for your particular make and model and compare it to the Harman curve. Where there is a difference, figure out what that difference measures and apply it to your EQ curve. For instance, say the Harman curve is +3dB higher at 1kHz compared to your cans. Just apply +3dB to your EQ at 1kHz. If it's lower, subtract from your curve. Just go from one end to the other making the needed corrections, then pull the whole curve down overall below zero to keep it from clipping. Now when you switch on the EQ, you are hearing the Harman curve not the natural curve of your cans.

Here is your cans' response (green) https://diyaudioheaven.files.wordpress.com/2013/06/comp.png?w=831

Here is the most recent Harman Curve https://www.head-fi.org/threads/2017-18-harman-iem-target.881258/#lg=attachment2504681&slide=0

If you google, you might find more detailed graphs, but odds are the EQ you're using probably isn't precise enough to exactly match it anyway. Just get as close as you can. That is your starting point.
 
Last edited:
Apr 21, 2020 at 9:51 PM Post #18 of 51
You line up the response curve measurement for your particular make and model and compare it to the Harman curve. Where there is a difference, figure out what that difference measures and apply it to your EQ curve. For instance, say the Harman curve is +3dB higher at 1kHz compared to your cans. Just apply +3dB to your EQ at 1kHz. If it's lower, subtract from your curve. Just go from one end to the other making the needed corrections, then pull the whole curve down overall below zero to keep it from clipping. Now when you switch on the EQ, you are hearing the Harman curve not the natural curve of your cans.

Thanks!
The curve you kindly provided for my cans it probably not the right one. The curve display ATH-A900, where mine is ATH-A900X (https://www.audio-technica.com/cms/headphones/554203d7fee12d97/index.html)

This is a screenshot from the EQ software that I'm using (seems quite capable of achieving accurate results):
2020-04-22 11_49_34-Window.png

I googled for quite some time and didn't find the 900X response curve
 
Apr 21, 2020 at 9:59 PM Post #19 of 51
Then just use the 900 if you can't find your exact model.
 
Apr 21, 2020 at 11:10 PM Post #20 of 51
I found this: https://www.innerfidelity.com/images/AudioTechnicaATHA900X.pdf

The top-left graph, should I refer to the red or blue lines?

Is my assumption correct, that I need to adjust my EQ to reach the ATH-A900X and then overlay that with Harman's target response and compensate the EQ to match the Harman target?
Or, can I match the Harman target from flat EQ?

I've done this in the meantime:
2020-04-22 11_49_341-Window.png

I just don't know how to cut/drop from 15000Hz like the Harman target.
 
Last edited:
Apr 22, 2020 at 12:09 AM Post #21 of 51
I found this: https://www.innerfidelity.com/images/AudioTechnicaATHA900X.pdf

The top-left graph, should I refer to the red or blue lines?

Is my assumption correct, that I need to adjust my EQ to reach the ATH-A900X and then overlay that with Harman's target response and compensate the EQ to match the Harman target?
Or, can I match the Harman target from flat EQ?

I've done this in the meantime:
2020-04-22 11_49_341-Window.png

I just don't know how to cut/drop from 15000Hz like the Harman target.

Blue is left channel, right is right channel. Just take the average.

Be very careful when doing what you're doing: you need to take into account the fact these curves may correspond to the same 'space' as what the headphone's measured in. All headphone measurement rigs have compensation: to do any form of accurate EQ'ing, you need to take into account the compensation.

What I would do instead is grab the impulse response for the AD900X and start with that. Someone did the hard work already, might as well use it!
 
Apr 22, 2020 at 12:16 AM Post #22 of 51
What I would do instead is grab the impulse response for the AD900X and start with that. Someone did the hard work already, might as well use it!

Impulse respone? Not Frequency response?
And by 'grab' you mean, set the EQ to resemble the A900X (not AD) frequency response and then decide how to change the EQ?

Sorry, I don't quite understand.
 
Apr 22, 2020 at 1:02 AM Post #23 of 51
Impulse respone? Not Frequency response?
And by 'grab' you mean, set the EQ to resemble the A900X (not AD) frequency response and then decide how to change the EQ?

Sorry, I don't quite understand.

Sorry, frequency response. And yes, grab the A900X, then evaluate what the difference is between the A900X and A900, then compensate based off that.
 
Apr 22, 2020 at 1:14 AM Post #24 of 51
Sorry, frequency response. And yes, grab the A900X, then evaluate what the difference is between the A900X and A900, then compensate based off that.
ok, will do. I thought to grab the A900X and then evaluate the diff between that to the Harman IEM 2017 frequency response.

What's not clear to me is why do I need to adjust the EQ to match the frequency response of the A900X and not straight away to the Harman freq. response?

If in my case it's System wide EQ -> SoundBlaster DAC -> A900X Headphones, why not set the EQ to Harman's? What benefit do I get by setting it to the A900X first?
 
Last edited:
Apr 22, 2020 at 4:57 AM Post #25 of 51
I just don't know how to cut/drop from 15000Hz like the Harman target.

Above 8 or 9 kHz or so is the least important part of the range to get right. Focus on getting the middle right.

Just ballpark correct to try to reach Harman by figuring out the difference between them and compensating. Once you do that listen to it. If you can polish it a little closer, do that. If you don't like where it's sitting, make small incremental changes and make sure they are in the right direction by going slow.

People are going to give you all kinds of complicated advice. Since you're just starting out, forget complete accuracy and just try to get it as right as you can. There is plenty of time to fine tune as you gain experience.
 
Last edited:
Apr 22, 2020 at 5:07 AM Post #26 of 51
People are going to give you all kinds of complicated advice. Since you're just starting out, forget complete accuracy and just try to get it as right as you can. There is plenty of time to fine tune as you gain experience.

I wish I knew what right was :)
Ok, so I think I got the Harman target correct, I just don't like that 3Khz peak. Other than that, I'm still uncertain on why I need to set my initial target to match the A900X and then try to match the Harman.
 
Apr 22, 2020 at 4:48 PM Post #27 of 51
I wish I knew what right was

Right is what sounds right to you. The Harman curve is just a starting place. That is what sounds good to the majority of people. YMMV.

If you are listening with the A900X, you are already hearing its signature. You don't need to set your EQ for that curve. You just need to look at the measurements for those cans and calculate how to adjust to reach Harman.

You'll find it is easier to judge what is correct for you by using well recorded, full range acoustic music. Electric instruments and synthesizers and highly processed mixes aren't a good baseline. Orchestral music, solo piano or guitar, chamber music, small group jazz with saxophones, church organs... those will give you a clearer read. You just need to analyze what you hear and experiment in baby steps so you don't wander off into some crazy setting you can't figure your way back from. Just search for natural sound and think about what you are hearing. It isn't just finding a sound you "like". It's finding the one that is real.
 
Last edited:
Apr 22, 2020 at 9:20 PM Post #28 of 51
If you are listening with the A900X, you are already hearing its signature. You don't need to set your EQ for that curve. You just need to look at the measurements for those cans and calculate how to adjust to reach Harman.

Ok, I think I understand now but please validate if I got it:
1. Assuming the system EQ is flat and not filters applied, I'm hearing the A900X signature.
2. The A900X signature equates to the A900X Frequency Response as measured by Inner Fidelity as seen here.
3. Tweaking the flat EQ to match Harman's suggested response curve is NOT the correct way, because the flat EQ already represents my A900X signature.
4. By looking at the A900X response curve, I should tweak the EQ in such a way that it will match Harman's response curve by looking at the differences between them.
Example: For the A900X, the response curve at 2Khz is -5db. Harman's response curve at 2Khz is +7db. Do I need to add +12db gain at 2Khz to reach a +7db?

Additional questions:
1. Looking at the A900X response curve, do I need to look at the grey lines? or average the blue/red lines?
2. Is there an automated tool that can do that? It's almost impossible to know the gain and Q values to use just by looking at pictures instead of raw numbers.
3. Does this table represent Inner Fidelity's measurements as seen in the graph from the above link? If looking at 20Hz, how can I read this table and conclude what parameters to use in my EQ settings to reach Harman's curve? maybe @jaakkopasanen can help? :)

Thank you! I feel like there's progress.
 
Apr 22, 2020 at 10:00 PM Post #29 of 51
I don't know what graph of the Harman Curve you are using. This one https://www.innerfidelity.com/content/harman-tweaks-its-headphone-target-response (green line) peaks at +10dB over zero, while your A900X chart tops out at zero. So you will want to 'normalize' the Harman curve down -10dB to put them in the same ballpark level-wise. So if the A900X reads -5dB at 2kHz, and the normalized Harman curve reads +5db, then +5dB minus 10dB normalization makes it -5db at 2kHz- you don't have to do any correction at that frequency at all. See my edit below. In general, it looks like you are going to have to cut below 2kHz and boost above it. You'll also need a boost at the bottom of the bass. You'll end up with a U shaped curve on your equalizer.

No, there is no automatic way of doing this. You have to do math. Write down the correction at each frequency band so you remember and don't have to do all the math again to get back to your Harman correction.

Generally, you want to EQ subtractively, rather than applying corrections that boost over zero. Thats' why you need to normalize the curve. Otherwise you would push your equalizer into clipping. Also, the frequencies you want to focus on are the ones under 10kHz. Above that, it doesn't matter nearly as much because those frequencies are only slightly audible. Headphones aren't terribly accurate up there anyway as you can see with all the zig zags.

I'm going to say it again. Don't worry about being totally accurate. Just ballpark it and get a feel for how it works. After doing it once, you'll be more familiar with how your equalizer works and the process, so you can try again with more precision. As you do this, try to get a feel for what the various frequency ranges sound like, and what sound goes with what number. That will help you when you start deviating from Harman to suit your ears.

Edit: One other monkey wrench in the works... I googled InnerFidelity and they are applying a compensation curve to their chart already. You can read about it here... https://www.innerfidelity.com/content/new-compensation-curve-innerfidelity-measurements That probably means that you have to subtract Inner Fidelity's compensation curve from their response measurement before correcting to Harman. This would affect the frequencies above 1kHz. Everything below that should be flat. This chart shows their compensation curve... https://www.innerfidelity.com/images/170913_Blog_IFCompensation_Graph_InnerFidelityCurve.jpg
 
Last edited:
Apr 22, 2020 at 10:11 PM Post #30 of 51
So to summarize... Take InnerFidelity's measurement of your cans and subtract their compensation curve. At 2kHz you would subtract 5dB. Your cans measure -5dB at 2kHz, so we subtract the IF compensation curve, and that makes it -10dB. The normalized Harman curve is -5dB at 2kHz, so you would need to boost 2kHz +5db on your equalizer.
 

Users who are viewing this thread

Back
Top