autoeq.app Is a Web Application for Equalizing Headphones the Easy Way

[Gotriss]

First I wanted to say thanks for the app. It's absolutely great and much easier to use than the old Github page.

But I registered here to report that I noticed a strange bug. I have read the entire thread and no one seems to have pointed it out so I'm not sure if it's been unnoticed for this long or something is broken with my PC/browsers. Anyway, here it is:

Using Firefox 116.0.3 (64-bit):

First time opening the webpage and inputting HD 600 as headphone, then Peace as the option:

Do a refresh, and then put again hd 600 + peace as option.

As you can see in the screenshots, there's difference in pre-gain and slight variations in all values (Frequencies/Q/Gain)

I didn't touch anything else on the app other than selecting the headphone and peace as an option.

I thought maybe this was a Firefox bug, so I did the same using Edge 115.0.1901.203 (64 bits)


First go:

Second go:

Why does this happen? Does it matter?

 

[jb90]

Ok. I spend some time with this app both for eq headphones and speakers/monitors (.csv raw speaker measurements with measurement mic + .csv Harman in-room target for speakers) but I still have some questions about settings of this app to get the best results. Sorry for long post but I guess some users can have similar issues/questions Thanks for your patience and tips!

About AutoEQ settings:
1. "Sound signature smoothing" - what is that? How to use it? Don't understand this option.
2. "Max gain" - How much db should be the max gain for headphones and speakers? I know that probably "it depends" but I guess there are some "maxes" that sound good and not make phase issues and drop of the sound level.
3. "Max slope" - same here. How is too low and how is too much for eq (both speakers and headphones)? I guess I don't need steep db/oct to eq every narrow dip or peak?
4. "Smoothing", "Treble smoothing", "Treble gain multiplier" and "Transition region" settings - I'm not sure how to use it so I always set it as low as I can and transition region to the lowest possible to not "smoothing" my treble region of FR. I think that gives me the best results and not "fake" calibration in that regions. But maybe I do something wrong? How to use it efficiently for eq speakers and headphones?

Now I have questions about two types of EQ I use the most cause I think that those are the best ones when it comes to customization of settings - Convolution EQ and Custom Parametric EQ.

About Convolution EQ:
1. "Bit depth" - my audio interface and computer internal sound cards use 24 bit. So I should still use 16bit or use 32bit? (I use APO EQ as my main "eq platform").
2. "Phase" - I guess minimal setting for headphones and linear for speakers, right?
3. "Stereo" - I don't hear any difference. Should I care?

About Custom Parametric EQ:
1."Optimizer min and max freq" - How to setup it for headphones and for speakers? Should I set it empty or just use it like a frequency response range of particular headphones/speakers to prevent boosts below and above usable freq range?
2. How many "bands" (both shelves and bells) should I use at once to get best results but also don't make too much phase issues or any other artifacts?
3. How narrow or broad should be those bands and how much cut or boost should apply to have no issues (min and max values)?

And final question:
1. Which type of eq will be better for speakers and which one for headphones?

Thanks!

 

[jb90]

Anyone?

 

[hyperknot]

I've just found this great app, @jaakkopasanen thank you first of all for creating it! I'm a bit confused about the Zero RED measurement though. It selects this "Diffuse Field 5128 (-1 dB /oct)" automatically. I don't understand how does it related to Harman In-Ear and other classic curves. I've read it here that I basically have no choice but to use this because it was recorded on a particular rig?

So until someone publishes a new measurement done on a different rig, I have to use this target?

 

[alumnicesar]

-1db / octave (-10db tilt) is way too bass light. That can easily be seen when switching between Harman OE and IE with it. Why not something like -15 and see how that looks?

 

[hyperknot]

I've given up on the Zero RED + Diffuse Field 5128 (-1 dB /oct), it totally ruins the sound. This IEM sounds much better without any EQ compared to when EQ-d to this target.

 

[alumnicesar]

yah that target is very bad

 

[alumnicesar]

I've given up on the Zero RED + Diffuse Field 5128 (-1 dB /oct), it totally ruins the sound. This IEM sounds much better without any EQ compared to when EQ-d to this target.

i checked the -1 db / oct tilt vs the mark levinson headphones. which are near perfect for harman over ear. if you change the tilt, -1.5 is probably the best fit.
so - 1.0 db with a -0.5 tilt

so that means in-ears would be -1.85

 

[alumnicesar]

I've given up on the Zero RED + Diffuse Field 5128 (-1 dB /oct), it totally ruins the sound. This IEM sounds much better without any EQ compared to when EQ-d to this target.

try -1.8 or -1.9. So put the -1.0 target, then go into advanced and change the tilt to -0.8 or -0.9. How does that sound?

 

[jaakkopasanen]

Those headphones haven bee measured on a new system that's not compatible with Harman targets. There currently doesn't exist an equivalent for BK 5128 but the -1 dB/oct diffuse should be somewhat close. The bass amount you can and should adjust two your taste.

 

[markanini]

I've given up on the Zero RED + Diffuse Field 5128 (-1 dB /oct), it totally ruins the sound. This IEM sounds much better without any EQ compared to when EQ-d to this target.

Same experience with Red, and Moondrop Lan. Even when I tried modifying the tilt and low shelf. The mids always have a weird reverse-N shape coloration.

 

[Sean Olive]

AutoEq is a project for automatically equalizing headphones. While this project is now familiar to many people engaged in headphone hobby, it has always been a bit of a pain to use if you want anything else than the pre-computed presets.

autoeq.app is a web app using AutoEq but now with a graphical user interface in the browser without needing to install anything yourself. All the same features and power that has made AutoEq great is now available on any device in your browser. The main driver behind developing a web app was to give an easy way for users to tweak their experience and not needing to settle for the single preset.

It's well know at this point that the preferred bass and treble levels are very individual and therefore should always be optimized. There are a lot more than those two though and I encourage you to explore the advanced parameters in the app. The player in the app simulates whatever equalizer app you choose so that you can quickly tweak and compare different profiles without needing to alwyas import them to the eq app.

You can even create completely custom changes to the frequency response target with sound signature or copy a sound signature from another headphone to simulate (roughly) other headphones on your single pair.

A fair bit of design, development and testing has gone into this app but no doubt there are still problems and room for improvement. Please share your opinions and discoveries and I'll take them to consideration for future updates.

I've also updated all measurements in the same release so all the data you have in the app (and the Github repo) are now the latest and greatest from all supported sources.

This is a very cool app. Thanks for all your work and making it available to the public.
I tried it yesterday and reported the results here: https://x.com/seanolive/status/1721692088436478016?s=20

 

[Sean Olive]

The default targets are the best guesses for neutral. That's the best for mixing too. I've decided not to add reviewer targets as they have little evidence for why they would be good and AutoEq has extensive options for the user to tweak the sound to their individual liking.

Smart move.

It's a diffuse field response of the BK 5128 rig with a 1 dB per octave downward (dark/warm) tilt. There's no similar published research for 5128 as there is for GRAS rigs (Harman target) and this one is the best guess at the moment for the most neutral target for 5128.

 

[jaakkopasanen]

Smart move.

Thanks! The 5128 target remains a puzzle. My personal guess is that there should be a bass shelf added, maybe 4 dB, maybe something else. This isn't much of a problem with autoeq.app but with some equalizer apps that come with AutoEq presets built in it's not so easy to tweak things. Any guesses to how much the default bass shelf should be with 5128 or 4620 measurements?

 

[Sean Olive]

Hello, jb!

Actually, the Harman target is supposed to "mimic" the frequency response of flat speakers in a well treated room. It doesn't include the crossfeed aspect of speakers nor any of this, just the frequency response aspect. The preference part is kind of the icing on the cake, where listeners further "sculpt" the sound of the target curve. The treble preference didn't vary much from the original target, the main differences were in the bass region (105Hz and below, a low shelf).

Now, don't get me wrong, the Harman target isn't the definitive target curve, mainly because the GRAS rig doesn't include the ear cannal's impact on frequency response.
For contextualization on the above mentioned, every measuring rig have their own impact on the final measured frequency response. The GRAS rig was the one used to make the original Harman target, therefore you want to use measurements done with that rig to get results that properly translate with the target. The Oratory1990 measurements are the most well known and are mainly the ones available in Autoeq.

Currently, there is a new rig available made by Bruel & Kjaer called "5128" that includes the ear cannal's influence in sound. There isn't a equivalent to the Harman target for that rig yet, but there is an attempt that tries to mimic the sound of it. If you have headphones that were measured with that rig, I'd suggest you to try the EQ profiles. It's probably the best approximation to a flat sound nowadays, at least that I know of.
Here's a video that explains that attempt:

"GRAS rig doesn't include the ear cannal's impact on frequency response."

I think you need to clarify this statement. The response of the GRAS (KEMAR) and B&K 5128 rig at low frequencies is identical when you measure a sound source (e.g. a multiple speakers in a reference listening room). at the DRP. I just did this recently. They differ mostly a higher frequencies, and part of that is related to the head/torso/pinna - not just the coupler.

However, when you put a pair of headphones on these test fixtures you will observe differences at low frequencies which is related to the volume and acoustic impedance of the rear canals. The amount of bass differences will depend on the acoustic impedance of the headphone.

Thanks! The 5128 target remains a puzzle. My personal guess is that there should be a bass shelf added, maybe 4 dB, maybe something else. This isn't much of a problem with autoeq.app but with some equalizer apps that come with AutoEq presets built in it's not so easy to tweak things. Any guesses to how much the default bass shelf should be with 5128 or 4620 measurements?

 

[Sean Olive]

Thanks! The 5128 target remains a puzzle. My personal guess is that there should be a bass shelf added, maybe 4 dB, maybe something else. This isn't much of a problem with autoeq.app but with some equalizer apps that come with AutoEq presets built in it's not so easy to tweak things. Any guesses to how much the default bass shelf should be with 5128 or 4620 measurements?

This is something we are investigating for around ear and in-ear headphones. Do you mean a 4 db shelf added to the 5128 DF calibration?

The recent AES paper from Senselab where listeners evaluated different headphone response curves measured on a 5128 suggests that 4 db is about right for AE headphones

After measuring many IE and AE models on both GRAS45 and 5128 test fixtures there does seem to be on average 3- 4 db less bass on the 5128. It varies depending on the model. For in-ear models the difference can be attributed solely to the coupler. For AE models measured at both the unblocked and blocked ear canal (where the coupler is not part of the measurement) there is also less bass on the 5128 than the flat plate GRAS45. Less bass than when the same headphones are measured on humans.

This has to be leakage effects. So, this all points to a target curve on the 5128 that will have less bass than the original one specified on the GRAS 45CA RA 0045

I’ve also remeasured calibrated speakers in our Reference listening room using different mannequins including 5128 which gives us information on how they differ from their published free-field and diffuse field responses. Hopefully, this data along with more listening test results should point to what the best average target curve for the 5128 should be, and what the tolerances around it should be based ratings of preference and perceived accuracy to a reference.

 

[alumnicesar]

This is something we are investigating for around ear and in-ear headphones. Do you mean a 4 db shelf added to the 5128 DF calibration?

I think he means 4db boost on top of df -10db

 

[Mauro]

AutoEq is a project for automatically equalizing headphones. While this project is now familiar to many people engaged in headphone hobby, it has always been a bit of a pain to use if you want anything else than the pre-computed presets.

autoeq.app is a web app using AutoEq but now with a graphical user interface in the browser without needing to install anything yourself. All the same features and power that has made AutoEq great is now available on any device in your browser. The main driver behind developing a web app was to give an easy way for users to tweak their experience and not needing to settle for the single preset.

It's well know at this point that the preferred bass and treble levels are very individual and therefore should always be optimized. There are a lot more than those two though and I encourage you to explore the advanced parameters in the app. The player in the app simulates whatever equalizer app you choose so that you can quickly tweak and compare different profiles without needing to alwyas import them to the eq app.

You can even create completely custom changes to the frequency response target with sound signature or copy a sound signature from another headphone to simulate (roughly) other headphones on your single pair.

A fair bit of design, development and testing has gone into this app but no doubt there are still problems and room for improvement. Please share your opinions and discoveries and I'll take them to consideration for future updates.

I've also updated all measurements in the same release so all the data you have in the app (and the Github repo) are now the latest and greatest from all supported sources.

Never realized that “all the small things” is such a good test track for deficiencies in the treble!!

This is a really really really great tool Jaakko. I still use the Qudelix implementation but the web version is so intuitive.
Where can I find the instructions in the wiki or somewhere else to tune my headphones to any other headphone?
For a next release it would be cool to add in the target dropdown menu all other headphones.

Ps. In Opera there seems to be issues to open the window to let pick a csv. The window hangs forever after pop-up

 

[jaakkopasanen]

Never realized that “all the small things” is such a good test track for deficiencies in the treble!!

This is a really really really great tool Jaakko. I still use the Qudelix implementation but the web version is so intuitive.
Where can I find the instructions in the wiki or somewhere else to tune my headphones to any other headphone?
For a next release it would be cool to add in the target dropdown menu all other headphones.

Ps. In Opera there seems to be issues to open the window to let pick a csv. The window hangs forever after pop-up

Home

Automatic headphone equalization from frequency responses - jaakkopasanen/AutoEq

github.com

 

[Sean Olive]

I think he means 4db boost on top of df -10db

A 4 dB LF shelf does get it closer to the Harman IE Target. I measured the JBL Endurance Run 2 on both GRAS and 5128 last night and plotted them against both Targets. See thread here: https://x.com/seanolive/status/1722874973793243155?s=20