• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

AutoEQ Magic?

Joined
Jun 13, 2020
Messages
16
Likes
21
#1
I stumbled upon a link to AutoEQ in another thread and searched to see if it had it's own dedicated thread (this will be the first).

Short version:
Jaakko Pasanen (don't know him) took the Harmon target curves for 2018 (Headphones) and 2019 (IEMs) and crinacle's measurements database and wrote some python to output EQ recommendations to apply for each device in crinacle's data. It's a crazy huge library. I've been testing for the last couple hours and I think Jaako is on to something.

Steps:
1. Install Equalizer APO and the Peace UI.
2. Find your headphones in the EQ recommendation list on Github.
3. Copy the contents from the "ParametricEQ.txt" file into Notepad.
4. Add a new top line with the pre-amp settings (found in the text of each headphone entry), here's the one from the BLON BL-03: "Preamp: -7.2 dB"
5. Save the file with a name (e.g. the headphone make/model).
6. Open Peace.
7. Click the import button (see image below), load the file you saved in step 5.
8. Click the save button. It defaults to the name of the file you imported. I added a link to the EQ settings page from Github for each device.

Load EQ setting file in PEACE.png


Note: you can also use the Peace UI to change EQ settings, but then it forces you to use increments of 0.5 db instead of 0.1 db that the EQ recommendations use (edit: unless you change the default in PEACE's settings). Make the changes in the file for more accurate results. Also, you should know that Peace will disappear for a few seconds between switching EQ profiles. It's a feature.

At first, I didn't care for it. It made the music quieter and I had to turn of EQ and rapidly turn down the volume to A/B test the EQ settings vs stock tune. And as you'd expect, loading an EQ for a different headphone/IEM makes your music sound like hot garbage. I tried listening to BLON BL-03s with the EQ settings for FiiO FH7s... and it was upsetting. After EQ, I can hear all the details in the BL-03 that I can in the FH7, but just not as clearly. It doesn't make the BLONs into an "A-Tier" IEM, but it makes it a pleasing muddy audio experience. I already thought the FH7 was pretty solid, but I think it sounds fantastic with the EQ profile. Since the software and settings are completely free, it's kinda crazy not to try out. I was considering pulling the trigger on a hardware EQ deck to add bass to the Koss ESP/95X on the way from Drop, but this should do the same thing, but cleaner.

What's crazier? Jaakko posted the python code he used to create the recommendations. So if you have a measurement system for headphones/IEMS like this, you create an EQ profile for your actual headphone, not the sample that crinacle got his hands on. Could be clutch if you're into headphone modding but also want to chase a target sound curve.
 
Last edited:
OP
A
Joined
Jun 13, 2020
Messages
16
Likes
21
Thread Starter #2
I've been digging into the EQ recommendations for the last couple hours. I'm really happy with the FH7 with EQ applied. But looking at the measurements graph of the measured EQ response, there's still some slop in the treble range. This lead me down a rabbit hole trying to find what it would cost to find an IEM that really responds well to these EQ settings. I was looking hard at the Fearless S8F. It EQ's quite well and is generally considered an excellent IEM for under $500. The problem is I don't want to spend $500 for another pair of IEMS. Then I started going through the list looking for IEMs that had crossover switches, were $100 or less, and responded decently to an EQ. I settled on the CCA C12 which EQs almost as well as the S8F but is $50 instead of $490 (plus not dealing with shipping from China). I ordered a 2.5mm balanced cable just so I can give these guys a fair shake. I'm hoping I can get similar low and mid clarity that I'm getting from my FH7 (my hypothesis is the FH7 will be the better here) but with cleaner treble range once I load the EQ. Should know whether I was right in a few days :cool:. (btw, none of those are affiliate links, not in this for money but chasing the absolute best sound possible).
 

Kouioui

Member
Forum Donor
Joined
Nov 9, 2019
Messages
72
Likes
69
Location
Central FL
#4
Note: you can also use the Peace UI to change EQ settings, but then it forces you to use increments of 0.5 db instead of 0.1 db that the EQ recommendations use.
It's possible to change the increments using the Settings button in Peace UI to whatever default value you need.
 
Joined
Feb 17, 2020
Messages
34
Likes
26
Location
Lithuanian in Ireland
#6
I've tried it and Oratory1990 ( https://www.reddit.com/r/oratory1990 ) and I prefer Oratory setting. AutoEq much bigger library and I'm using that on my phone with Wavelet app.
 
OP
A
Joined
Jun 13, 2020
Messages
16
Likes
21
Thread Starter #8
Nice! Thanks for bringing Oratory1990 to my attention. I made a quick helper sheet to create Equalizer APO EQ settings based on those settings in Google Sheets.
 
Joined
Feb 17, 2020
Messages
34
Likes
26
Location
Lithuanian in Ireland
#9
I would also recommend on trying some crossfeed (APO has two implementations), because our brain is wired to receive sound crossfed and headphones isolate that. Most music is mastered with speakers in mind.
 
OP
A
Joined
Jun 13, 2020
Messages
16
Likes
21
Thread Starter #10
Did some more testing. I can report that Oratory1990 EQ recommendations sound better to me, but they cover vastly fewer makes and models. Currently listening to CCA C12s w/ AutoEQ settings and they sound nearly perfect. Drums sound like drums, voices sound normal, pianos sound like pianos. Very happy.
 
OP
A
Joined
Jun 13, 2020
Messages
16
Likes
21
Thread Starter #13
AutoEQ took several databases of measurements and EQ'd them to the Harmon curves. That's why you see the Oratory database in the Github. I think with the AutoEQ python scripts you can EQ to any target you want. I haven't tried a diffuse field target though, so can't contribute there =/
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
2,895
Likes
2,772
Location
UK
#14
Joined
Mar 2, 2020
Messages
18
Likes
12
#15
That's not too surprising. It probably comes down to the way that the software calculates the filters. For example it appears to me that AutoEQ does not incorporate High-Shelf or Low-Shelf filters in their EQ settings.
That shouldn't worry you too much though. The single filters might be different, but if you add them the result can still be (essentially) the same. It's like writing 5 as 2+3 or 1+4. Different components, same result.

So much for the theory. In practive there is of course a difference.
Since the number of filters is limited you cannot get arbitrarily close to the target curve. Instead you have to make trade-offs and decide where you want to be more precise and where you allow some more room for errors.
It appears that the Oratory curve is more "lax" when it comes to high frequencies >10k Hz. There's only a single filter in that range. The AutoEq settings however have 3 filters in that range (and one at 9.6k Hz). So the AutoEq curve is matching the target curve in the high frequency range much better.
In contrast, the Oratory curve has more filters assigned to the mid-range so it will be more precise there.

In the end it comes down to personal preference and to tuning the algorithm that calculates your filter parameters.
 

Soniclife

Major Contributor
Forum Donor
Joined
Apr 13, 2017
Messages
2,895
Likes
2,772
Location
UK
#16
It appears that the Oratory curve is more "lax" when it comes to high frequencies >10k Hz. There's only a single filter in that range. The AutoEq settings however have 3 filters in that range (and one at 9.6k Hz). So the AutoEq curve is matching the target curve in the high frequency range much better.
In contrast, the Oratory curve has more filters assigned to the mid-range so it will be more precise there.
They do seem to prioritise different ends of the spectrum.
650.jpg

This is the 2 compared, they should sound different, going to play with it later on.
 

Fluffy

Addicted to Fun and Learning
Joined
Sep 14, 2019
Messages
752
Likes
1,106
#17
This is definitely not the first thread on this topic, and I'm really tired of getting into this discussion over and over again. Suffice to say that AutoEQ and similar measure-and-apply-filter-to-match-target-response solutions are very simplistic in their approach and ignore many other considerations of sound quality and personalization. It's worth to search deeper in the topics of AutoEQ and Oratory1990 to get a more complete picture.
 

dasdoing

Active Member
Joined
May 20, 2020
Messages
237
Likes
95
Location
Salvador-Bahia-Brasil
#18
I just tried out the CSVs in REW and they are compatible.
So anybody can use REW with the data provided.
This one is just a test with REWs auto eq generator.
I use the tons of headrooms I (and probably you guys) have to make the target lower


autoeq.jpg
 
OP
A
Joined
Jun 13, 2020
Messages
16
Likes
21
Thread Starter #19
This is definitely not the first thread on this topic, and I'm really tired of getting into this discussion over and over again. Suffice to say that AutoEQ and similar measure-and-apply-filter-to-match-target-response solutions are very simplistic in their approach and ignore many other considerations of sound quality and personalization. It's worth to search deeper in the topics of AutoEQ and Oratory1990 to get a more complete picture.
1593523365422.png
 

Similar threads

Top Bottom