Go Back   Cockos Incorporated Forums > REAPER Forums > REAPER General Discussion Forum

Reply
 
Thread Tools Display Modes
Old 11-01-2016, 07:26 PM   #1
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default Headphone Frequency Response Correction Using ReaEQ

EDIT 11/10/16: My finalized attempt is described here in post #60 though there is much good discussion between here and there. Also, Post #61 has a link to my ReaEQ presets in the Stash. In Post #69 I contemplate the effects of resonance of the ear canal, which I cannot measure, and Post #71 depicts the EQ Preset results of that idea. I believe the EQ preset in Post #71 is my best, final preset and that preset has been updated into the RPL file linked above.
-------------------------------------------------------
In a recent thread it had came up that I've created a preset for ReaEQ to compensate the frequency response of the headphones I use to mix which are Sennheiser HD 380 Pro.

DISCLAIMER: I cannot vouch for the accuracy of my response correction, but I'm willing to share the preset if anyone is interested.

Essentially, the idea started after I read a thread talking about Sonarworks' REFERENCE 3 HEADPHONE PLUGIN which costs $99 USD and can correct the response of your headphones IF you are using headphones listed on the page in the link. Since my headphone model was not in the list I began a quest to create my own corrective EQ using ReaEQ.

It was hard to find any response graph for the HD 380 Pro on the internet, but I found one at Golden Ears.
http://en.goldenears.net/13496

I wasn't confident that their reported graph was accurate, so I attempted to take my own measurements using a Behringer ECM8000 and REW Room EQ Wizard. I quickly learned that I simply did not have the correct measuring environment. To obtain a true, accurate headphone frequency response you need to have a Binaural Recording head placed inside an anechoic chamber. Of which, I had neither.

However, even as faulty as my measurement method was my results from 2 kHz and above were quite on-par with the graph reported by Golden Ears which leads me to believe that their graph is accurate. So, above 2 kHz I used my actual, measured response averaged from several tests in REW to create the filters needed to counter the response deficiencies and estimated the filters below 2 kHz based on the graph reported by Golden Ears.

Unfortunately, the filters in REW use Band-Width in Octaves Q to define the wideness of the filters rather than Q bandwidth in octaves like ReaEQ does. Fortunately, I found a conversion tool on the internet that would interpret the Q values for me.
http://www.sengpielaudio.com/calculator-bandwidth.htm

Ultimately, the corrective preset for ReaEQ turned out looking like this:

Putting it in action, I found that it sounded too harsh as-is, but at 70% wet mix did seem to flatten the response a bit. However, I have no good, scientific basis for comparison other than what my ears can tell.

The way I use this is to put this ReaEQ preset on the master when I'm mixing using my Sennheiser HD 380 Pro. Doing so prevents me from making too bright and too bass lacking mixes which is what seemed to happen when I played my headphone mixes back on my studio monitors.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000

Last edited by insub; 11-10-2016 at 07:24 AM.
insub is offline   Reply With Quote
Old 11-01-2016, 07:31 PM   #2
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by Tod View Post
I'm not sure what I'm looking at with that link.

I've got a pretty decent acoustic mic, and just for the heck of it I ploted out my low budget earphones a while back. I was just curious how it would turn out and this is basically it.

The link is indicating the frequency response in the first graph of the HD380Pro with subsequent graphs indicating other things for their purposes.

See my above post on whats required to get an accurate freq response for headphones.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-01-2016, 08:19 PM   #3
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
Unfortunately, the filters in REW use Band-Width in Octaves to define the wideness of the filters rather than Q like ReaEQ does. Fortunately, I found a conversion tool on the internet that would interpret the Q values for me.
Hi insub, take a look at the bandwidth in my EQ, ReaEQ is definitely in Oct. And I think it's pretty accurate too, .33 will give you about a third octave bandwidth. And ReaEQ has a really wide bandwidth, I don't know how accurate it is at the extremes but I use those areas a lot.

I'd have to go back and redo the tests to get you the actual plot, and I'll do that if it's important.

Quote:
To obtain a true, accurate headphone frequency response you need to have a Binaural Recording head placed inside an anechoic chamber. Of which, I had neither.
In regards to a "Binaural Recording head placed inside an anechoic chamber", that seems a little overboard to me. The volume levels coming from the ear phones as so low they can hardly be heard in my control room. I placed the mic right on top of the left, then the right, to see what they were doing. I ended up encapsulating the mic between the left and right.

For me it's not all that important because I have a pretty good control room. But I still find it very interesting and I like to experiment with things like this.
Tod is offline   Reply With Quote
Old 11-01-2016, 09:30 PM   #4
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by Tod View Post
Hi insub, take a look at the bandwidth in my EQ, ReaEQ is definitely in Oct. And I think it's pretty accurate too, .33 will give you about a third octave bandwidth. And ReaEQ has a really wide bandwidth, I don't know how accurate it is at the extremes but I use those areas a lot.
I must have got them mixed up. I guess it was the other way around, that REW uses the Q value.

Quote:
Originally Posted by Tod View Post
In regards to a "Binaural Recording head placed inside an anechoic chamber", that seems a little overboard to me. The volume levels coming from the ear phones as so low they can hardly be heard in my control room. I placed the mic right on top of the left, then the right, to see what they were doing. I ended up encapsulating the mic between the left and right.

For me it's not all that important because I have a pretty good control room. But I still find it very interesting and I like to experiment with things like this.
Maybe I should have said "most accurate". But, I could not get an accurate reading the way I tried before. However, I was trying to get separate measurements for each side. Thinking about it now, I should have just smashed the tip of the mic between the two ear cups allowing the cushions to make a seal around the capsule. I could have held the cups together with rubber bands. I may do that when I get home.

I made a new preset, based solely on the reported graph from Golden Ears. I used GIMP to invert the curve and align it to the ReaEQ graph. Then I set the filters visually. This is probably pretty good until I can get a better measurement of the actual response.

It looks like this, which is probably more realistic than what I'd done before:
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 08:09 AM   #5
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
Maybe I should have said "most accurate". But, I could not get an accurate reading the way I tried before. However, I was trying to get separate measurements for each side. Thinking about it now, I should have just smashed the tip of the mic between the two ear cups allowing the cushions to make a seal around the capsule. I could have held the cups together with rubber bands. I may do that when I get home.
Yeah, I'd like to see how that turns out.
Tod is offline   Reply With Quote
Old 11-02-2016, 10:42 AM   #6
DVDdoug
Human being with feelings
 
Join Date: Jul 2010
Location: Silicon Valley, CA
Posts: 2,779
Default

I can appreciate what you're trying to do, but practically I don't think there's a benefit.

The measurements are tricky and I won't say they are useless, but they are not 100% reliable either. You can compare different headphones made on the same simulated head & same set-up to get a general idea of the frequency response differences. i.e. If the Golden Ears test shows a Sennheiser headphone with a 5dB more bass than an AKG headphone, it's probably true. The high frequencies will be less accurate, but the measurements may still give you some good clues about the sound, and especially the difference in sound between different headphones.

But a different setup is going to give you different results. I assume Sennheiser's lab gets different results than Golden Ears, and I assume AKG's lab gets different results than Sennheiser, etc.

And in the real world, different people have different ear shapes (different resonances) and headphones fit different people differently, and that affects the sound.

And where it really gets tricky is that you don't want to completely eliminate the natural effects of the ear because if you do it won't sound natural!

Most mixing & mastering engineers will recommend against using headphones as your primary reference, so "accuracy" isn't that important anyway.

The HD380s are pretty good headphones and they should be "good enough" as-is, depending on your tastes/preferences. And if you prefer a different sound you can EQ by ear or choose a different headphone.

When you get into the price range of the HD380 (or higher) there are lots of good headphones and it largely becomes a matter of personal preference. ...In a blind listening test between the HD380 and any other good headphone, I wouldn't be surprised if half of the listeners chose the Sennheiser and half the listeners chose the other headphone.

So my recommendation is, buy a headphone that you like or EQ your headphone for a sound you like, or EQ by ear to whatever you need to make good mixes/masters.
DVDdoug is offline   Reply With Quote
Old 11-02-2016, 11:23 AM   #7
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Well apparently Sonarworks and all their customers believe there is a benefit. But, part of Sonarworks' offered services are to make custom response measurements of your particular pair of headphones. You send in your headphones to them for the measurement.

Even using my previous preset and available data, I believe there was a measurable improvement in my mixes. Could I have made just as much improvement by adding a high-shelf boost and nothing else? Perhaps. But, this way is attempting at something more specific and scientific than just, "Throw some EQ on it and go." Besides, I need to end up settling on something and sticking with it. Which is why I'm trying to approach this as scientifically as I can given my available means. If I keep experimenting around with different EQ settings at random then I'll never find a happy medium.

"Buy a headphone that you like" is not a viable option for many. I had to order my HD 380 Pro online based on other users' reviews because there was no place local to try them at the time. I selected these over DT 770 or ATH-M50 because of budget. Amazon had the 380 for $30 less than the other options. I don't think that novice/hobbyist level people would even benefit from going into a store to listen to them. They'd be choosing headphones that they liked the sound of better based on a short trial. But, that doesn't indicate if that person will make a better mix with one over the other. It was months before I could come to the conclusion that my mixes were consistently affected the same way due to my choice of headphones. Besides, even at the budget level, $150 & up per headphone set adds up pretty quick. I guess you could keep returning them, but that isn't very ethical in my opinion.
"Can I have my money back?"
"Is something wrong with them?"
"No, just... I make crummy mixes while using these."


I know that using monitors in a treated room is better. But, that is simply not an option for me right now. I have time to mess with audio while in my hotel room when traveling for work. So, headphones it is.

Regardless, I just have to get my nerd on until the itch is scratched, if you know what I mean.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 11:26 AM   #8
X-Tech
Human being with feelings
 
Join Date: Apr 2015
Posts: 589
Default

I have few pair of headphones with different characteristics and each have another resposne so i just changing headphones - not playing with EQ. For example audiotechnica is pretty clear headphones but bass is killing but at other hand i have byerdynamic and they more flat but not as much detailed at mid.
__________________
nonononononono
X-Tech is offline   Reply With Quote
Old 11-02-2016, 11:27 AM   #9
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

I highly recommend the iMM6 calibrated mic for use with iOS. $16 plus shipping. There's nothing fancy about the mic, except that every unit is individually tested at the factory and has a downloadable correction file. I've been using it to EQ my monitors+room and it is great.
Borodog is offline   Reply With Quote
Old 11-02-2016, 11:37 AM   #10
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by X-Tech View Post
I have few pair of headphones with different characteristics and each have another resposne so i just changing headphones - not playing with EQ. For example audiotechnica is pretty clear headphones but bass is killing but at other hand i have byerdynamic and they more flat but not as much detailed at mid.
That's a good point for having multiple pairs. One of the other reasons I chose the HD 380 Pro is because they fold flat and came with a semi-hard shell case for travel while the other options I was looking at did not. I am wanting to get a pair of DT 880 for at home, but they don't fold up very well. So, the 380s are probably still the better option for all the travel that I do.
Quote:
Originally Posted by Borodog View Post
I highly recommend the iMM6 calibrated mic for use with iOS. $16 plus shipping. There's nothing fancy about the mic, except that every unit is individually tested at the factory and has a downloadable correction file. I've been using it to EQ my monitors+room and it is great.
What software do you use on iOS? Did it have some that came with it?

One of the benefits of REW is that you can use it to generate the EQ filters necessary for the correction. Unfortunately, I wasn't able to figure out how to export those filter settings in a way that could be imported by ReaEQ or any other VST EQ that I have. REW will output the filter settings as a Text file.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 11:57 AM   #11
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Hi again insub, what kind of a mic do you have for testing. You might have mentioned it but if you did, I missed it.

I've got a Behringer ECM8000, I think it only cost $50 or $60, and seems to work pretty well.
Tod is offline   Reply With Quote
Old 11-02-2016, 02:01 PM   #12
juliansader
Human being with feelings
 
Join Date: Jul 2009
Posts: 3,714
Default

Would a dynamic EQ perhaps be better suited for compensating a headphone's frequency response?
juliansader is offline   Reply With Quote
Old 11-02-2016, 02:03 PM   #13
OPI
Human being with feelings
 
OPI's Avatar
 
Join Date: Nov 2014
Posts: 1,900
Default

Whoa... strange. I didnt see this thread.
OPI is offline   Reply With Quote
Old 11-02-2016, 04:10 PM   #14
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by Tod View Post
Hi again insub, what kind of a mic do you have for testing. You might have mentioned it but if you did, I missed it.

I've got a Behringer ECM8000, I think it only cost $50 or $60, and seems to work pretty well.
That's the same mic that I have. REW has a compensation file for it too.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 04:35 PM   #15
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
That's the same mic that I have. REW has a compensation file for it too.
Okay, are you up for a test without REW involved?

I can help you get setup for that if you do.

This test would just involve Reaper, ReaEQ, your mic, and Span.

Span is a great spectrum analyzer and works very well for this.

At any rate let me know what you think.
Tod is offline   Reply With Quote
Old 11-02-2016, 04:39 PM   #16
OPI
Human being with feelings
 
OPI's Avatar
 
Join Date: Nov 2014
Posts: 1,900
Default

Also interested tod i wouldnt mind some collateral knowledge
OPI is offline   Reply With Quote
Old 11-02-2016, 04:49 PM   #17
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

I'm still out of town. I might be able to make new measurements after tomorrow.

I could also measure it in MEqualizer, but it has a limited number of EQ bands and SPAN has none. Is there a way to export from SPAN into ReaEQ?

I don't see an advantage to using SPAN over REW. Also, REW will average multiple responses, which can't be done in MEqualizer and I'm not sure about SPAN in that regard. Can SPAN be made semi-transparent. That, at least, would be useful. I think that Blue Cat's FreqAnalyst can be made transparent, but I can't remember if it is freely resizable.

I know that the ECM8000 is an Omni pattern, but I still want to take multiple measurements with the capsule entering the ear cups from different directions. I'll probably do 4, front, back, top and bottom. Then average them together. I think I can make 31 bands of EQ in REW, IIRC.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 05:22 PM   #18
semiquaver
Human being with feelings
 
Join Date: Jun 2008
Posts: 4,923
Default

but phase !!!!
semiquaver is offline   Reply With Quote
Old 11-02-2016, 05:43 PM   #19
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

Quote:
Originally Posted by insub View Post
That's a good point for having multiple pairs. One of the other reasons I chose the HD 380 Pro is because they fold flat and came with a semi-hard shell case for travel while the other options I was looking at did not. I am wanting to get a pair of DT 880 for at home, but they don't fold up very well. So, the 380s are probably still the better option for all the travel that I do.

What software do you use on iOS? Did it have some that came with it?

One of the benefits of REW is that you can use it to generate the EQ filters necessary for the correction. Unfortunately, I wasn't able to figure out how to export those filter settings in a way that could be imported by ReaEQ or any other VST EQ that I have. REW will output the filter settings as a Text file.
AudioTools. Another $20, but worth it imo. There are other, free tools that can load the correction file. It's just a txt file.

You can export a reading as a text file, but I haven't figured out how to use it as an EQ correction yet. Just been adjusting the HF controls on my monitors plus some manual guess and check correction via ReaEQ.

My head is killing me but I'll try to post some with/without measurements later.
Borodog is offline   Reply With Quote
Old 11-02-2016, 05:45 PM   #20
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Good point Semiquaver.

Do you think it's better to deal with the pre-ringing of linear-phase EQ? Maybe I should use ReaFir instead?

I haven't noticed any trouble with phase relationships due to monitoring through ReaEQ. Think about it. At some point an EQ is going to wind up on the master bus. What difference does it make if it's at the beginning of the project?
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 05:59 PM   #21
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
I could also measure it in MEqualizer, but it has a limited number of EQ bands and SPAN has none. Is there a way to export from SPAN into ReaEQ?
Ho there isn't but the way I do it, you're using span to adjust the EQ, so there's no need for that.

Quote:
I don't see an advantage to using SPAN over REW. Also, REW will average multiple responses, which can't be done in MEqualizer and I'm not sure about SPAN in that regard. Can SPAN be made semi-transparent. That, at least, would be useful. I think that Blue Cat's FreqAnalyst can be made transparent, but I can't remember if it is freely resizable.
I'm not sure what you mean by "semi-transparent", but Span is resizable to 3 different states.

Quote:
I know that the ECM8000 is an Omni pattern, but I still want to take multiple measurements with the capsule entering the ear cups from different directions. I'll probably do 4, front, back, top and bottom. Then average them together. I think I can make 31 bands of EQ in REW, IIRC.
Humm, are you saying you have 4 different speakers with your earphones?

I do have ReaEQ set up for a 27 band graphic EQ, and that's also an interesting way to approach it. But I'm mainly talking about starting out at Zero and building your EQ from there. You have to have a little knowledge about how set the bandwidth up, but basically you adjust the EQ according to what you see happening with Span.

You can also have 2 graphs (with the free version of Span), both left and right showing at the same time, but that takes some setting up and depends on what you want to achieve.
Tod is offline   Reply With Quote
Old 11-02-2016, 06:11 PM   #22
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

EQ only causes phase problems when it's differentially applied. A single EQ that's the last thing across the master buss will not cause any audible phase issues

I think.
Borodog is offline   Reply With Quote
Old 11-02-2016, 06:14 PM   #23
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Ok. I think I get where your going with this.

You run pink noise into the headphones and monitor SPAN while adjusting the EQ in realtime?

There is only a single driver in the headphones. I just want multiple measurements for accuracy checks. The drivers are at an angle, I believe, to the outer surface. Multiple measurements are likely unnecessary, but it's just in case there is some difference based on the direction which the mic enters the headphones.

Using REW, I'd be generating the frequency response from a sine sweep rather than using noise. The pink noise method is much faster though. Honestly, I hadn't considered trying that method.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 06:29 PM   #24
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

Ok, so this is not a finished product yet. Still a work in progress. Here's the room+monitors before adjustment:



And here's the room after adjusting the HF controls on the monitors and some manual experimentation with ReaEQ:



This room is completely untreated; laminate floors, drywall, big windows. Not even so much as a rug.

I'd sure some room treatment could smooth out the nooks and crannies a bit, but right now it's not in the budget.

This thread has me wanting to measure my Sennheiser HD 280 Pros.
Borodog is offline   Reply With Quote
Old 11-02-2016, 06:31 PM   #25
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

PS I think the second graph is a little more detailed because it was on 1/12th octave not 1/3 octave smoothing like the first one.
Borodog is offline   Reply With Quote
Old 11-02-2016, 06:37 PM   #26
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by Borodog View Post
This thread has me wanting to measure my Sennheiser HD 280 Pros.
Found these, quick search.
http://en-de.sennheiser.com/download...7a2d0206a6.pdf
http://reference-audio-analyzer.pro/...hd-280-pro.php
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 06:42 PM   #27
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Same site has a measurement for the HD 380 Pro too. Quite a different graph than what Golden Ears had, especially for the bass response.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 08:08 PM   #28
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

Here's my 280s:

Borodog is offline   Reply With Quote
Old 11-02-2016, 08:49 PM   #29
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

That looks fairly close to what RAA reported on their graph. Can your iOS app display the frequency graph logarithmic-ally? The grid divisions don't make much sense. Is it a 1/3 octave grid?
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-02-2016, 09:32 PM   #30
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
Ok. I think I get where your going with this.

You run pink noise into the headphones and monitor SPAN while adjusting the EQ in realtime?

There is only a single driver in the headphones. I just want multiple measurements for accuracy checks. The drivers are at an angle, I believe, to the outer surface. Multiple measurements are likely unnecessary, but it's just in case there is some difference based on the direction which the mic enters the headphones.

Using REW, I'd be generating the frequency response from a sine sweep rather than using noise. The pink noise method is much faster though. Honestly, I hadn't considered trying that method.
Yes, you've got the right idea, the thing about using pink noise is that you'll be doing it in real time. Of course you do have to understand what you're looking at and then how to make the correct adjustments.

Like I said, I also have ReaEQ set up with a 27 band graphic EQ that kind of takes a lot of the guess work out of it. It works pretty well but isn't as accurate as pin pointing the freqs.
Tod is offline   Reply With Quote
Old 11-03-2016, 11:41 AM   #31
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Here's a pict of the 27 bands of 3rd octave graphic EQs I'm been talking about. I use three ReaEQs for this, each with 9 bands. The reason for the 3 EQs is so they are all visible and I can adjust them with out having to scroll.

I've got 2 actual 3rd octave 27 band graphic EQs on my studio monitors and I mimicked them with this EQ.



The main and only advantage to using this setup, is that it takes the guessing out of setting the bandwidth and the frequencies. However, there are times that slight adjustments of the bandwidth to make it little narrower or wider works well too.
Tod is offline   Reply With Quote
Old 11-03-2016, 01:07 PM   #32
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Nice.

Well, I discovered some MAJOR errors that I made using REW previously. First of all, REW does not slope the graph like SPAN or MAnalyzer, etc. I confirmed this after my new measurements by using the REW real-time analyzer with a pink noise source. I will post some pictures soon.

I will also compare my results using the REW method and using just pink noise + SPAN.

Speaking of slopes, SPAN's default slope is 4.5 dB/Oct which should result in pink noise making a slight rise as the frequency increases. Whereas, MAnalyzer uses a default slope of 3 dB/Oct which should result in a flat pink noise response. I recommend changing the slope setting of SPAN to 3 dB/Oct.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-03-2016, 01:49 PM   #33
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Using REW with a sine sweep
I think it was a good idea to take multiple measurements. I measured with the tip of the microphone entering the ear cups as they were held together using the cushion pads to form a seal. The four measurement directions were the mic entering from the front, back, top, & bottom. Three of the measurements were quite similar, while the front measurement had a much lower bass response.

Then I averaged the 4 measurements together.

REW does have an option for bandwidth rather than Q factor if you use the FBQ2496 Equaliser setting that allows for up to 20 filters, peak type (parametric) only (no shelves, etc.). I was able to set the target resonse partially to a -3 dB/Oct slope as shown in this image as the purple line.

This shows the original averaged response and the predicted response after the filters are applied.

I was able to get a flattish response using 12 filters.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-03-2016, 02:36 PM   #34
Borodog
Human being with feelings
 
Join Date: Dec 2014
Posts: 153
Default

Quote:
Originally Posted by insub View Post
That looks fairly close to what RAA reported on their graph. Can your iOS app display the frequency graph logarithmic-ally? The grid divisions don't make much sense. Is it a 1/3 octave grid?
It is logarithmic. Just different divisions and labels. Major divisions are a factor of 2 (ie an octave): 31.25, 62.5, 125, 250, 500, 1k, 2k, 4k, 8k, 16k. Minor divisions are 1/3 octave (hence 3 per major).
Borodog is offline   Reply With Quote
Old 11-03-2016, 03:15 PM   #35
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
Speaking of slopes, SPAN's default slope is 4.5 dB/Oct which should result in pink noise making a slight rise as the frequency increases. Whereas, MAnalyzer uses a default slope of 3 dB/Oct which should result in a flat pink noise response. I recommend changing the slope setting of SPAN to 3 dB/Oct.
Yeah, I forgot to mention that.
Tod is offline   Reply With Quote
Old 11-03-2016, 08:34 PM   #36
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

CONCLUSION
The average response from the four measurements I took matches VERY closely to the graphs published by Golden Ears and RAA when taking slope into account. REW's graph has no slope at all, Golden Ear's looks to be about 1.5-2 dB/Oct, and RAA's looks like a 3 dB/Oct slope. In my opinion, their graphs are good representations of the Sennheiser HD 380 Pro at least.

I believe that using the REW method produced a more realistic corrective EQ action than using pink noise in conjunction with a real-time analyzer. Here is the result of my correction EQ using pink noise + SPAN:


And, here is a direct comparison of the EQ presets, REW vs Pink Noise:


Notice that the pink noise corrective EQ required more bands with higher gain amounts and narrower bandwidth settings. I believe the same occurred with Tod's EQ posted above. Notice some extremely sharp filters.

From my testing, I don't believe there will be a worthwhile result from setting the correction based on pink noise. Furthermore, the pink noise method appeared to be considerably affected by ambient sound even though the HD380 Pro's have excellent noise rejection (up to 32 dB according to Sennheiser). While I was trying to set the EQ filters the furnace fan turned on. I'm in my basement and the furnace is about 14' away. There was a considerable low frequency jump after the fan came on.

I don't think that sine sweep responses are as susceptible to ambient interference to as large a degree. After all, I took 5 response measurements, each lasting about 6 sec and they all came back approximately the same. Although, I don't know if the furnace fan was on or not during those measurements. But, considering that my measurements matched pretty closely to those I found on the internet, I believe they are pretty accurate.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-03-2016, 11:27 PM   #37
Tod
Human being with feelings
 
Tod's Avatar
 
Join Date: Jan 2010
Location: Kalispell
Posts: 14,745
Default

Quote:
Originally Posted by insub View Post
Notice that the pink noise corrective EQ required more bands with higher gain amounts and narrower bandwidth settings. I believe the same occurred with Tod's EQ posted above. Notice some extremely sharp filters.
Hi insub, and thankyou for your excellent tests. One thing to keep in mind when adjusting the EQ to pink noise in real time, it's very easy to start chasing the frequencies. In fact looking at the results of the pink noise test, it appears that may have happened. As you adjust one EQ band up, it often affects one, or both, of the adjacent bands, causing you to adjust those bands in the opposite direction.

It takes a great deal of patience, practice and trial & error to come up with something reasonable.

Even looking at your REW test, it appears that could have done a little chasing to, but not as bad.

One of the main problems with that, is with the phasing between the individual EQ bands. I know I've worked on EQing my monitors and gotten an almost perfectly flat response, only to find that they did not sound good at all.

Consequently, I've become more conservative with my adjustments.

Quote:
From my testing, I don't believe there will be a worthwhile result from setting the correction based on pink noise. Furthermore, the pink noise method appeared to be considerably affected by ambient sound even though the HD380 Pro's have excellent noise rejection (up to 32 dB according to Sennheiser). While I was trying to set the EQ filters the furnace fan turned on. I'm in my basement and the furnace is about 14' away. There was a considerable low frequency jump after the fan came on.
Oh yeah, outside noise like that will definitely affect the results. My studio is very quiet, the only extraneous noise would be my chair if I happened to move, which I'm very aware of.

One thing I do with Span, is often click on it to rebuild the envelope.

Quote:
I don't think that sine sweep responses are as susceptible to ambient interference to as large a degree. After all, I took 5 response measurements, each lasting about 6 sec and they all came back approximately the same. Although, I don't know if the furnace fan was on or not during those measurements. But, considering that my measurements matched pretty closely to those I found on the internet, I believe they are pretty accurate.
Yeah, I'm sure that might be true.

Last edited by Tod; 11-03-2016 at 11:34 PM.
Tod is offline   Reply With Quote
Old 11-04-2016, 06:13 AM   #38
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Quote:
Originally Posted by Tod View Post
Hi insub, and thankyou for your excellent tests.
You're welcome. We are such nerds to be interested in this stuff, I think. Thank you for being interested and helping me think through this endeavor!
Quote:
Originally Posted by Tod View Post
One thing to keep in mind when adjusting the EQ to pink noise in real time, it's very easy to start chasing the frequencies. In fact looking at the results of the pink noise test, it appears that may have happened. As you adjust one EQ band up, it often affects one, or both, of the adjacent bands, causing you to adjust those bands in the opposite direction.

It takes a great deal of patience, practice and trial & error to come up with something reasonable.

Even looking at your REW test, it appears that could have done a little chasing to, but not as bad.

One of the main problems with that, is with the phasing between the individual EQ bands. I know I've worked on EQing my monitors and gotten an almost perfectly flat response, only to find that they did not sound good at all.
I don't know what you mean here exactly. Do you notice some kind of transient smearing at the crossover points between the EQ bands? If so, how do you propose to reduce the problem?

I could try to make an EQ preset that is primarily subtractive. The problem is the shape of the EQ bands. Wide enough to sufficiently suppress a large peak also starts to draw down the valleys on either side. Ultimately, my main goal in the above filter sets was to use as few filter bands as possible.

Here is the REW based ReaEQ preset I made with the phase showing:
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-04-2016, 07:24 AM   #39
insub
Human being with feelings
 
insub's Avatar
 
Join Date: Mar 2014
Location: Louisville, KY, USA
Posts: 1,075
Default

Check this out!
Using primarily subtractive EQ I was able to get a predicted flat response using only 8 filters! I only needed one boost filter.

In the screenshot, the red line is the original response, the dark green line is the predicted response after the EQ, and the dark purple line is the target. I'm not sure how I glitched the colors make them differ from the legend (Response line red instead of light green). Also, yesterday I couldn't get REW to let me make a continuous 3 dB/Oct slope, but I was able to today by setting both the "LF Rise Start" and the "HF Fall Start" to 200 Hz and both slopes set to 3 dB/Oct.
__________________
Everything you need to know about samplerates and oversampling... maybe!
My Essential FREE 64bit VST Effects, ReaEQ Presets for Instruments
Windows 10 64 bit; MOTU 828 MKII, Audio Express, & 8PRE; Behringer ADA8000
insub is offline   Reply With Quote
Old 11-04-2016, 08:55 AM   #40
JHughes
Banned
 
Join Date: Aug 2007
Location: Too close to Charlotte, NC
Posts: 3,554
Default

Please see ReaDave's contribution to this very subject.
http://forum.cockos.com/showthread.p...ight=headphone
JHughes is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 05:25 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.