[billqs]

I understand. I guess I'll use the HE1000v.2 settings. SE has some different sound but it's not completely out of the ballpark.
Thanks!

 

[Hercules]

Hi Jaakko,

Can the EQ generate from AutoEQ can be used on new AK players with Eqlists?

https://www.astellnkern.com/eng/con...ew.asp?search=&word=&idx=1000125&page=3&no=89

Thank You

 

[jaakkopasanen]

Hi Jaakko,

Can the EQ generate from AutoEQ can be used on new AK players with Eqlists?

https://www.astellnkern.com/eng/con...ew.asp?search=&word=&idx=1000125&page=3&no=89

Thank You

Based on what I see in the Youtube video linked to that page, the AK player has graphic eq with ~20 filters and a parametric eq that has fixed center frequencies. The parametric eq isn't quite compatible with AutoEq because of it's fixed center frequencies. However you should be able to generate graphic eq settings for the AK player by setting the center frequencies with --fc parameter and quality with --q. In this case the required quality is 2.45 if I'm not mistaken.

 

[Hercules]

Based on what I see in the Youtube video linked to that page, the AK player has graphic eq with ~20 filters and a parametric eq that has fixed center frequencies. The parametric eq isn't quite compatible with AutoEq because of it's fixed center frequencies. However you should be able to generate graphic eq settings for the AK player by setting the center frequencies with --fc parameter and quality with --q. In this case the required quality is 2.45 if I'm not mistaken.

a bit disappointing as Astell&kern only develop one EQ so far and stop doing it.

under the youtube link

 

[rodrigaj]

@jaakkopasanen
I have not seen any mention of using a minidsp SHD here. I have Sennheiser HD800S for which I wish to set aside one configuration solely for headphone use. I went into the SHD plugin, linked outputs 1 and 2 and proceeded to set the PEQ filters according to the table you have provided here. The SHD has a very easy to use GUI for setting up 10 PEQ filters. Here is a screenshot after I input the filters:



My question is about your statement: "It's very important to set preamp according to the value given in the result README.md document." The HD800S require a -6.2dB preamp setting. The SHD has a default 0dB reference as shown below on outputs 3 & 4. Interpreting your statement for the SHD, have I set the preamp value on outputs 1 & 2 correctly?



The SHD volume control would then be set to 0dB (i.e., full volume) and the SHD would then be used as a parametric equalizer with final volume control handled by my headphone amp.

Hopefully, I've got this sorted out correctly, and thank you so much for the work you've done and made available for free. It was exactly what I was looking for.

 

[hq136185925]

that's a huge amount

 

[rodrigaj]

To clarify post #50:

Here is the way signal flows in the minidsp SHD


Hopefully, someone can help me understand if I've set up my configuration correctly. (I tried at the minidsp SHD forums)

 

[playmusic]

Does anybody know whether all equalizers operate with sufficient headroom?

Here is an example for what I mean.
This eq is from github/autoeq for a Sennheiser HD 800S:
Type Fc Q Gain
Peaking 23 Hz 0.92 5.6 dB
Peaking 61 Hz 2.59 1.9 dB
Peaking 2657 Hz 0.75 6.5 dB
Peaking 9812 Hz 0.97 10.6 dB
Peaking 17437 Hz 0.07 -9.8 dB

Preamp gain is stated for this example as -6.2 dB.
This preamp gain is adequate for the sum of all filters.

But if the filters were applied one after the other by the eq software without additional headroom then +10.6 dB gain for filter 4 can potentially clip input which is close to full scale.

So, as I understand it, the statement "-6.2 dB preamplifier gain is adequate" is based on the assumption that all equalizers operate with sufficient headroom.
Is this the case?
Or should I take the maximal individual gain of filters as preamp to be on the safe side, so -10.6 dB in this case?

I would be in particular interested for Peace in Windows because this is the tool I use at the moment.

 

[rodrigaj]

Based on my experience with the SHD that was noted above, I tried with -6.2dB and 0.0dB attenuation. I did not hear any clipping, even with the most dynamic range music I have, so I left it at 0.0dB. I listen to classical music and keep the listening levels at "reasonable" levels.

It all depends on the setup that you have.

 

[playmusic]

I did not hear any clipping

I wish I knew more about digital audio.
But for the moment I have to rely on guessing what is going on.

Not hearing clipping with 0 dB preamp in your example can have two reasons:

1. There is no clipping
2. There is inaudible clipping.

Let us consider the critical frequency f (9812 Hz in the above example).
My understanding of eq is that if only a very small part of the total signal is in the region around f then a gain of 10 dB (or even 20 dB) around that frequency will not clip the total signal. So this would be case 1. I guess that this applies to most music actually.

Clipping can of course happen, e.g. for a full scale sine wave of frequency f.
I just generated some examples of sine wave clipping and the most dominant effect is a third harmonic 50 dB below the fundamentel - which is likely inaudible, at least for f > 7 kHz, or even lower. So this would be case 2.

So, my understanding for this example is:
preamp -10.6 dB is fully safe, no clipping
preamp -6.2 dB is fully safe if the eq software uses sufficient headroom
preamp 0 dB has probably non audible consequences.
(Actually, preamp -10.6 dB reduces the dynamic range by 10 dB, so this seems to be a negative side effect.)

I am looking forward to the feedback of more knowledgeable people who might also point out that my reasoning in this post is not correct.

 

[UpTo11]

Just tried your eq data out of curiosity with my 1more triple driver over-ear and all I can say is: it works!
The sound improves not only in the tonal balance and control, but also in soundstage, expecially depth. There is more "breath" into the music.
Quite impressed! Kudos! It's an amazing job!

On Mac OSX the only drawback is that AUNBandEq is limited to 5 octave bandwidth AFAIK.

Why don't you include DIY Audio Heaven measurements too?

 

[headdict]

I use AUNBandEq in conjunction with TDR Nova to overcome AUNBandEq's limitations and quirks whenever needed.

 

[jaakkopasanen]

Just tried your eq data out of curiosity with my 1more triple driver over-ear and all I can say is: it works!
The sound improves not only in the tonal balance and control, but also in soundstage, expecially depth. There is more "breath" into the music.
Quite impressed! Kudos! It's an amazing job!

On Mac OSX the only drawback is that AUNBandEq is limited to 5 octave bandwidth AFAIK.

Why don't you include DIY Audio Heaven measurements too?

DIY Audio Heaven uses flat plate coupler for measurements ie. it's missing the outer ears. This kind of measurements are good in terms of reproducibility and useful for example in quality control. However they are not an accurate presentation of how human hears and therefor should not be used as a basis for equalization.

 

[snatex]

Anyone here with better computer skills than me that can take Audiofanatyk's LCDi3 measurements for the Cipher V2 and enter them into Jaako's AutoEq to generate a 10 band EQ?

Here are the links for the measurements and the AutoEQ instructions:

https://www.audiofanatyk.pl/wp-cont...-compare-frequency-response-graph-1900-02.jpg

https://github.com/jaakkopasanen/AutoEq

 

[HapaxLegomenon]

Hi Jaakko,

First of all, I honestly can't thank you enough for all the work you've put into AutoEQ. I've been using AutoEQ for months now and I cannot understate how revelatory it has been: this is the most fun I've had with headphones in years. For me, AutoEQ is now just as, if not more important, than the headphone itself. Recently I've been playing around with sound signatures, and I love how I can almost "audition" any other headphone I might be interested in (or at least its general sound signature).

I had a question about AutoEQ that I was wondering you could answer. I've equalized both my HIFIMAN HE4XX and HIFIMAN HE400i 2020 to the Focal Clear's sound signature (with a 1.7dB bass boost), and I was wondering why the target curves for both results' generated graphs (attached) are different.

Here are the commands I used:


HE4XX to Focal Clear:

python autoeq.py --input_dir="measurements\oratory1990\data\onear\HIFIMAN HE4XX" --output_dir="my_results\[UAPP 10-band Parametric] HE4XX to Focal Clear (1.7dB Bass Boost)” --compensation="compensation\harman_over-ear_2018_wo_bass.csv" --sound_signature=”results\oratory1990\harman_over-ear_2018\Focal Clear\Focal Clear.csv” --equalize --parametric_eq --max_filters=10 --bass_boost=5.7 --max_gain=25​

HE400i 2020 to Focal Clear:

python autoeq.py --input_dir="measurements\oratory1990\data\onear\HIFIMAN HE400i 2020" --output_dir="my_results\[UAPP 10-band Parametric] HE400i 2020 to Focal Clear (1.7dB Bass Boost)” --compensation="compensation\harman_over-ear_2018_wo_bass.csv" --sound_signature=”results\oratory1990\harman_over-ear_2018\Focal Clear\Focal Clear.csv” --equalize --parametric_eq --max_filters=10 --bass_boost=5.7 --max_gain=25​

I thought that since the compensation, sound signature, and bass boost are the same for both commands, the target curve should be the same for both results. However, when looking at the automatically-generated graphs, the target curves are different (but the general shape is the same). I played around some more and found that if I use the same compensation, sound signature, and bass boost but a different headphone's (e.g., Sennheiser HD 600) measurement input_dir, the target curve shifts up/down (but retains the same general shape), so I'm sure there's a very good reason for why this is the case. I guess I'm just wondering why this happens.

Thanks for everything, and if you have time please let me know!