Question about AutoEQ and REW for headphone EQ correction

[Volutrik]

Hello everyone. Today I did a little experiment: I downloaded the Harman 2018 over ear target curve from AutoEQ repository and also the headphones measurement available there. I then used REW to equalize the headphones to the downloaded Harman target curve and, unexpectedly, the result doesn't sound the same as the one that's already available as a "preset" on the AutoEQ repository.
I'm using IRs for more accuracy and the one I did using REW sounds slightly darker. Both measurements are using the Harman OE 2018 target, the difference is that one is already available on AutoEQ and the other one I made myself using REW.

Does REW have the same accuracy as AutoEQ or is AutoEQ more refined? I'm kinda confused right now. Both are using the same measurements and targets.

Below are the REW EQ settings I used for this as I want the results to be as close as possible to the target. Let me know if I should use some other settings

 

[staticV3]

I then used REW to equalize the headphones to the downloaded Harman target curve and, unexpectedly, the result doesn't sound the same as the one that's already available as a "preset" on the AutoEQ repository.
the one I did using REW sounds slightly darker.

The presets you find on the AutoEQ GitHub do not use the stock Harman target.
They use a modified version of Harman OE2018/IE2019v2 with 2dB less bass.

 

[Volutrik]

The presets you find on the AutoEQ GitHub do not use the stock Harman target.
They use a modified version of Harman OE2018/IE2019v2 with 2dB less bass.

I used the Harman OE 2018 v2 available from the AutoEQ Github and also the measurement from Oratory1990 too! Makes sense if you take into consideration that a sound with less low frequencies will be perceived brighter when compared to one with more low frequencies at the same perceived loudness. But the differences are much more than a matter of frequency tilting, they're very different on the mids/high mids, even above 5kHz. If you want I can provide you the Impulse responses, they're very different indeed, I'm kinda confused to why that is. Maybe REW doesn't do so much of a good job Auto EQing as Jaakko's Auto EQ and that's why it sounds different? I kinda like the result when comparing to the official Auto EQ one though

 

[staticV3]

You can plot the frequency response of both EQ presets in REW to compare them directly.

Here's how I do it with PEQ, but convolution filters should work just as well:

 

[Volutrik]

You can plot the frequency response of both EQ presets in REW to compare them directly.

Here's how I do it with PEQ, but convolution filters should work just as well:

Thank you for the tip! I tried doing it and the corrections don't look quite aligned to the target. The green one is the one downloaded directly from the Auto EQ Github and it has a bass boost by default, the orange one is the one I did using REW's auto EQ function and it is matched to the Harman OE 2018 v2 without the bass boost, and the blue one is the raw Harman target without the bass boost as well. Although I've been messing around with EQing headphones for some time, I'm kinda new to REW and to comparing the actual corrections to the target. My previous experiences were more based on just listening to the result and reading stuff online. Is this dip occuring between 2.2kHz and 7.3kHz a thing or is it an error? I'm wondering because the same occurs on the official Auto EQ one too...

 

[staticV3]

@Volutrik I think you forgot to attach the graph

 

[OK1]

Thank you for the tip! I tried doing it and the corrections don't look quite aligned to the target. The green one is the one downloaded directly from the Auto EQ Github and it has a bass boost by default, the orange one is the one I did using REW's auto EQ function and it is matched to the Harman OE 2018 v2 without the bass boost, and the blue one is the raw Harman target without the bass boost as well. Although I've been messing around with EQing headphones for some time, I'm kinda new to REW and to comparing the actual corrections to the target. My previous experiences were more based on just listening to the result and reading stuff online. Is this dip occuring between 2.2kHz and 7.3kHz a thing or is it an error? I'm wondering because the same occurs on the official Auto EQ one too...

Hello everyone. Today I did a little experiment: I downloaded the Harman 2018 over ear target curve from AutoEQ repository and also the headphones measurement available there. I then used REW to equalize the headphones to the downloaded Harman target curve and, unexpectedly, the result doesn't sound the same as the one that's already available as a "preset" on the AutoEQ repository.
I'm using IRs for more accuracy and the one I did using REW sounds slightly darker. Both measurements are using the Harman OE 2018 target, the difference is that one is already available on AutoEQ and the other one I made myself using REW.

Does REW have the same accuracy as AutoEQ or is AutoEQ more refined? I'm kinda confused right now. Both are using the same measurements and targets.

Below are the REW EQ settings I used for this as I want the results to be as close as possible to the target. Let me know if I should use some other settings

Hi Volutrik.

Looks like the key word here is algorithm. The method by which each processor derives their correction EQ, and therefore impulse files, is clearly not identical.

Which explains why their results are not identical.

I will add, the methods used are not set in stone. Once upon a time, REW would only correct frequencies up to 10Khz (or was it 8Khz)... but that has since changed. So earphone correction using EQ, is an emerging knowledge area, and just like we have in cars, or televisions, or any other area of product development, the solutions have the same objective, but their implementation is unlikely to be identical.

Was just reminded that we now have one more way to get to the International Space Station via a Boeing craft. Same purpose, but definitely not the same vehicle/design as Elon Musk's.