How to create headphones EQ profile using REW

[Music1969]

take the harman_target.txt and edit it. or create a textfile from scratch

or try one of the other targets, e.g. harman without bass boost

Noted creating or editing text file was the backup.

But there's no solution to what I described?

I want to reduce the amount of bass boost of Harman curve but want to reduce the amount of 2k-5kHz boost (just to try and further tweak).

Is there an easier graphical way to create a target curve like i described - some program to connect dots, smooth the lines and export to .txt?

I guess the other way is to just change the EQ parameters in REW itself and I will visually see the predicted there anyway

 

[ernestcarl]

Is there an easier graphical way to create a target curve like i described - some program to connect dots, smooth the lines and export to .txt?

VituixCAD has an SPL trace tool where you can convert any image to a txt file target. My suggestion is drag and drop your present target curve into REW. Reshape via the EQ tool. Take an image snapshot of this new Target and import it in VituixCAD’s SPL tool... copy the raw trace to a new text file and drag and drop the file to REW. Apply psychoacoustic smoothing. Save and/export the target trace as a text file (within REW) to increase the resolution. The final file will be your new modified target response.

 

[Robbo99999]

Noted creating or editing text file was the backup.

But there's no solution to what I described?

I want to reduce the amount of bass boost of Harman curve but want to reduce the amount of 2k-5kHz boost (just to try and further tweak).

Is there an easier graphical way to create a target curve like i described - some program to connect dots, smooth the lines and export to .txt?

I guess the other way is to just change the EQ parameters in REW itself and I will visually see the predicted there anyway

Oratory has set up his EQ's so the user can easily adjust tonality re bass & treble. Have a look at his pdf file that I linked you, at the bottom he describes which filters to change. You just need to change the Gain of Filter #2 for the bass and Filter #5 for the treble. So implement those in REW and turn them into convolution, then demo the changes - you've got all you need there.

 

[Music1969]

VituixCAD has an SPL trace tool where you can convert any image to a txt file target. My suggestion is drag and drop your present target curve into REW. Reshape via the EQ tool. Take an image snapshot of this new Target and import it in VituixCAD’s SPL tool... copy the raw trace to a new text file and drag and drop the file to REW. Apply psychoacoustic smoothing. Save and/export the target trace as a text file (within REW) to increase the resolution. The final file will be your new modified target response.

This would be great if I had Windows :-(

Only use macOS though.

I'm just trying to smooth the transitions of the below out.

I don't want anyone to do it for me because i'll be trying all sorts of custom EQ curves so will need to figure a way myself.

The sharp corners aren't ideal for REW right? Better to have smoother transitions?

MS Excel can create a polynomial trendline and tell me the formula but 6th order isn't enough and is the maximum (its close). I need higher order LOL

 

[Music1969]

you've got all you need there.

I just want to try one more thing with auto EQ and if I can't do it, I will stick with oratory and adjust those 2x PEQ bands 'to taste'

It's just bugging me that i can't do something so simple like create a smooth curve .

 

[Music1969]

Ok so there's a few macOS apps that digitise plots... "plot digitizers".

Now I need to hand draw a curve LOL

Gotta be a better way.

 

[dasdoing]

You also shouldn't be boosting all the sharp dips at 3kHz & 4kHz on your Focal Elegia all the way up to the Harman Curve, those troughs are too sharp and too deep to EQ up completely to the Harman Curve.

REW's predicted curves are prety acurate afaik. I do't see a reason why EQ in this case wont work

 

[milus]

Ok so there's a few macOS apps that digitise plots... "plot digitizers".

Amongst them: WebPlotDigitizer

It's multi-platform, open source and you can install it or use it as a web service, launching it right from that page.

 

[Robbo99999]

REW's predicted curves are prety acurate afaik. I do't see a reason why EQ in this case wont work

Those sharp & deep dips are likely called by interferences which are not boostable by EQ - certainly the extremely sharp dip at 3kHz. Not only that the massive gain you need to apply to the filter means a greater negative preamp. One extra point, the further up the treble you go the less predictable it is from person to person, your dip in the treble may not be exactly in the same place, and then you'll be royally screwed if you end up creating a sharp & massive boost filter that is missing the dip, because then you'll end up with a massive sharp peak next to the massive sharp dip that you've missed - which would have been worse than not boosting that area. So the sensible approach is not too boost massive sharp dips, and that's why Oratory didn't boost those 2 areas, he knows what he's doing.

 

[Music1969]

One down side to this headphones stuff (without measuring gear) compared to speakers, is I can't verify the end result, to check for any major peaks that i've created.

Until measurement gear gets cheaper and better.

With speakers, I can verify how things measure in a room with convolution/EQ enabled.

 

[Jimbob54]

Those sharp & deep dips are likely called by interferences which are not boostable by EQ - certainly the extremely sharp dip at 3kHz. Not only that the massive gain you need to apply to the filter means a greater negative preamp. One extra point, the further up the treble you go the less predictable it is from person to person, your dip in the treble may not be exactly in the same place, and then you'll be royally screwed if you end up creating a sharp & massive boost filter that is missing the dip, because then you'll end up with a massive sharp peak next to the massive sharp dip that you've missed - which would have been worse than not boosting that area. So the sensible approach is not too boost massive sharp dips, and that's why Oratory didn't boost those 2 areas, he knows what he's doing.

Or, try it with filter on, filter off and somewhere between the 2 and see what sounds best. Easy if using PEQ, not so much with convolutions

 

[dasdoing]

Those sharp & deep dips are likely called by interferences which are not boostable by EQ - certainly the extremely sharp dip at 3kHz. Not only that the massive gain you need to apply to the filter means a greater negative preamp. One extra point, the further up the treble you go the less predictable it is from person to person, your dip in the treble may not be exactly in the same place, and then you'll be royally screwed if you end up creating a sharp & massive boost filter that is missing the dip, because then you'll end up with a massive sharp peak next to the massive sharp dip that you've missed - which would have been worse than not boosting that area. So the sensible approach is not too boost massive sharp dips, and that's why Oratory didn't boost those 2 areas, he knows what he's doing.

imho the only reason is the audio dogma to not use too much boost. we are not talking about a 50dB notch here (where boosting will just results in more cancelation). it make no sense to me personaly, since we all have massive headroom afaik. it probably made sense in the past. unless the headphones become too quiet or hiss emerges, there is no reason to limit boosts.....at the end, we are not boosting anything after we ajusted the gain, we ONLY atenuated

 

[Music1969]

imho the only reason is the audio dogma to not use too much boost.

The same exists in digital room correction.

Genelec's room EQ solution (GLM) only reduces, they are anti-boosters lol.

Dirac limit boosting to 9dB.

Audiolense (which I use) developer has no issues with boosting even higher for the reasons you mentioned.

 

[Music1969]

Easy if using PEQ, not so much with convolutions

Actually in HQPlayer I use I can create different convolution 'profiles' and switch on the fly, as music is playing.

It makes comparing easy.

 

[Robbo99999]

Or, try it with filter on, filter off and somewhere between the 2 and see what sounds best. Easy if using PEQ, not so much with convolutions

Yeah, it can be tried, why not.

imho the only reason is the audio dogma to not use too much boost. we are not talking about a 50dB notch here (where boosting will just results in more cancelation). it make no sense to me personally, since we all have massive headroom afaik. it probably made sense in the past. unless the headphones become too quiet or hiss emerges, there is no reason to limit boosts.....at the end, we are not boosting anything after we ajusted the gain, we ONLY atenuated

I think you're wrong in your assessment, at least one of those dips is an interference related dip, definitely the one at 3kHz, and probably the other one at 4kHz to a lesser extent, so boosting those will not actually lead to a boost if cancels out due to interference. Oratory didn't boost those areas all the way to the Harman Curve, he knows what he's doing. By all means try boosting that area perfectly to the Harman Curve, but I think it's the wrong approach.

 

[pkane]

I think you're wrong in your assessment, at least one of those dips is an interference related dip, definitely the one at 3kHz, and probably the other one at 4kHz to a lesser extent, so boosting those will not actually lead to a boost if cancels out due to interference. Oratory didn't boost those areas all the way to the Harman Curve, he knows what he's doing. By all means try boosting that area perfectly to the Harman Curve, but I think it's the wrong approach.

The reason not to correct sharp dips in higher frequencies is that they are most likely the result of resonance and reflections inside the measurement device + headphone combination, often very sensitive to the specific position of the headphones.

Using EQ to correct for such a sharp dip can result in a large peak if you just reposition the headphones slightly, and that's probably not something you'd want to do to your ears

 

[Robbo99999]

The reason not to correct sharp dips in higher frequencies is that they are most likely the result of resonance and reflections inside the measurement device + headphone combination, often very sensitive to the specific position of the headphones.

Using EQ to correct for such a sharp dip can result in a large peak if you just reposition the headphones slightly, and that's probably not something you'd want to do to your ears

That was one of the reasons I listed in an earlier post too.

EDIT: although thinking about it, Oratory does many measurements at a number of headphone positions, and then creates an average curve from those - throwing out outliers or poor bass seal....so the average curve he shows for the Elegia has those 2 dips in that area, so they're probably not particularly headphone position dependant. Instead it's probably somekind of interference that is not related to headphone position. Either way we think it's unwise to boost those sharp dips.

 

[pkane]

This would be great if I had Windows :-(

Only use macOS though.

I'm just trying to smooth the transitions of the below out.

I don't want anyone to do it for me because i'll be trying all sorts of custom EQ curves so will need to figure a way myself.

The sharp corners aren't ideal for REW right? Better to have smoother transitions?

MS Excel can create a polynomial trendline and tell me the formula but 6th order isn't enough and is the maximum (its close). I need higher order LOL


View attachment 107718

You can use REW EQ filters to create a smooth curve. Just load the flat-line frequency response, then go to EQ and enter the desired filter to adjust it. Say LS12dB, 100Hz:

Then File->Export->Filter impulse response WAV file (just like you did for convolution). Close EQ window and import the newly created impulse response into REW by using File->Import->Impulse Response and specify the WAV file you just saved.

When loaded, select it and do export again: File->Export->Export measurement as text (choose no smoothing) and save it. This new file can then be used as the house curve in REW settings.

Needless to say, you can make the house curve as complex as you want by specifying multiple filter settings, not just one shelf filter as I did above.

 

[Music1969]

hen File->Export->Filter impulse response WAV file (just like you did for convolution). Close EQ window and import the newly created impulse response into REW by using File->Import->Impulse Response and specify the WAV file you just saved.

When loaded, select it and do export again: File->Export->Export measurement as text (choose no smoothing) and save it. This new file can then be used as the house curve in REW settings.

Needless to say, you can make the house curve as complex as you want by specifying multiple filter settings, not just one shelf filter as I did above.

Thanks for this!!

All good now. Well at least good for experimenting.

I must be an odd ball because I much prefer flat-ish to the Harman Curve

But its still early days , i need to experiment more.

After more time i may end up settling somewhere in between Harman and the below . Who knows where I'll end up

Appreciate everyone's help in getting me here !

I'm now in a position to help others with this

 

[Music1969]

I might actually go do a proper hearing test.

Now I know how to create custom impulse response files, for left and right ear separately, I can adjust not just for taste but also for my own hearing profile.

This could be more useful than spending money on fancy power cables and USB audio fixer gadgets ! (I used to be one of those people until I started with digital room EQ with speakers).