Creating convolution files for Roon in the new Room EQ Wizard (REW) with Inversion

Thanks for the info OCA.

Just for clarification, because I am just a beginner I’m a little confused about some things. Maybe even a lot confused…LOL

With the group of files you sent me most recently in the zip file, the FLFR-48k wav has no correction above 500 Hz correct?

And the REW file that’s in the zip file contains that (under 500Hz) correction?

Correct! The first one I sent with wrong xo point doesn’t go beyond 500Hz either.

Is that better? Then the original PDF that does full range?

also I’ve been playing around with my original REW file that i first sent you, and even when i start over from saving it at step 1…it automatically adds the 2700ms to IR windows Ref time. So i didn’t type it in. I even cleared my settings, and its just there after i create the R inverted. Would you have any idea why?


With digital correction less is usually better :wink:

Your speakers really have a smooth response in the high frequencies and HF response is very location specific (think millimeters).

1 Like

I don’t know, you might be using an old buggy version of REW besides trace arthmetic volumes and window types changed slightly in the newer versions of REW and I cannot replicate your results, it produces the minimum phase versions correctly in my REW (attached):

Ok. Thanks again and again.

I’m very glad you made that under/500Hz filter for me, it sounds great and it’s probably the one I will stick to for a long time.

But, I guess I also wanted to play around with the full correction too. That’s why I was digging into it. Having that option to tame the mids and highs to see what that sounds like is interesting….even if it might not sound as good.

1 Like

So, I use Roon for playback. In the PDF you instruct for Roon to have REW output 2 files(one for 44.1 and one for 48K sample rate) and then zip them together. But, the phase corrected file that you sent me before(with correction below 500Hz) its only one file labeled 48K.

Should i have 2 files for that too. one for 44.1 and 48k?

Thanks OCA, let me know when i should stop bugging you…LOL

I might as well also ask you this…

You mentioned earlier with the correction that factors the crossover(the under 500hz correction) that I can just change the slope of the target curve to increase bass if I wanted. Do you have an instructional video or did you make a PDF on how exactly to compile it all? Unfortunately, I have no idea how to compile a new wave file from that.

As I said, before, I’m just learning all the stuff… and I would love to be able to know how to do it myself.

That PDF that you made is fantastic, really, I’ve never seen anything as detailed and thorough as what you made on corrections.

I think I’m just finding I need a little bit more bass than what that correction has.

You have been extremely helpful and generous with your time. I have learned more about REW in the past few days than I had in the past year.

Here is the filter with more bass:

48kHz is REW’s native sampling rate so it produces the cleanest filters are at that rate. Let Roon do the up/down sampling. It has better algorithms for that than anything else.


That sounds great! Thanks :pray:

Hi Serkan, thanks a lot for sharing this video.

I am new in using Roon and new in DSP processing. So not everything in the video is clear to me. At the moment I have two questions and would be happy with your assistance.

First is the measurement of my Stereo System. I have REW on my macbook and connected this via HDMI/RCA Splitter to my analog stereo amplifier. I owned an Umic 1 microphone. Everything works.
As far I understood I measure both speakers at my listening position. Would you recommend only one measurement for each speaker at the middle of listening position or is it better to measure several times in a cube of 50 cm around the listing position an to calculate an average measurement?

Is it correct to increase the volume so that the Level shown in REW is around 75DB?

You use a Harman Target Curve in your video. Where can I find this file. You told me that there is a link in the comments. Sorry, but where can I find this comments? Is there any PDF or manual where I can read the content of your video in detail?

Sorry for the perhaps dumb questions.

Best regards

As long as you align them with cross correlation, you will have advantages with multiple measurements around the LP. The link to Harman curve and the pdf tutorial is in the decription under the video but I will paste them here again for you:

Hello OCA

Absolutely loving the filter, truly it sounds amazing.

So, is that PDF you put out exactly what YOU did for me? Including the last part at the end too?

Or was it a different method?

Hello Serkan,

thanks a lot for your reply. Just found the target curve and the PDF tutorial in then description of the video.
Done the first measurements with the UMIK1 yesterday and tried to follow the description in the video and the PDF. I got the correction curves for each speaker. But if I have a control look Speaker Original + Correction over the target curve there is nearly no change. The filter export and import in Roon works fine, but the sound is very disharmonic and clips.
I tried to find a mistake in my steps and tried the same twice more with same results. No idea for me what is going wrong. Do you have an idea, where I did my mistake?

Looks like you’re adding the convolution to the original response instead of multiplying with it. Convolution in time domain is multiplication in the frequency domain.

If all else is correct, the clipping in Roon is normal and expected. You’ll need to add some offset for headroom. Make sure you export the minimum phase 1/A convolution impulses with no normalization and windowing.

If you can’t sort out, share your mdat file if you want me to have a look.

Hello OCA,

thanks a lot. Yes this was a mistake. I calculated A+B, not AxB. So now the response curve look better. So visual it look better, but the sound keeps bad. Can you please explain what you mean with offer for headroom in Roon?

I can share my measurement files. It would great if you would have a look?
Where can I share the measurements for you?

Best regards

Hello OCA,

I just found the headroom offset in Room and had some tries. I have to decrease for 12dB got get no clipping. Is this OK?

Thanks a lot

It’s a bit too much for just magnitude inversion but it depends on your speakers’ response really. I average around -8dB headroom in Roon with my corrections for the worst records for clipping.

You can upload an mdat directly I guess or share a google drive link here.

No need to agonize about headroom adjustments audibly degrading the signal, if that’s your fear, since calculation is performed with 64 bit precision to keep effects many orders of magnitude below human perception limits.

Read the excellent article by @DrCWO who wrote an article presenting relevant measurements regarding Roon’s DSP volume control , which performs just like the headroom adjustment.

1 Like