A guide how to do room correction and use it in Roon

(Steve) #121

Nothing wrong with that. Just the way I like it too!! :slight_smile:

(kevin) #122

Hi all, @Magnus and I worked together to turn this guide into a post on our blog – check it out below! We also linked to it on our Facebook page (go like it!).

A big thanks to Magnus for taking the time to put this together. It has already helped so many of you, and we’re glad to spread the word.

(Mark Allen) #123

Awesome. I’ve expressed this to @Magnus already, but this guide is fantastic. Applying this guide as really made my system come alive.

(Jeffrey Robbins) #124

Need help on a 5.2 surround system. Has anyone used REW to measure up for this and imported the filters into Roon? JCR

(Magnus) #125

I have no experience with multi speaker setup, but maybe something like this will work:

  1. Measure front and each satellite and adjust them (both frequencies and total volume) so they match each other. This would be step 6-7 in the guide.
  2. Do a measurement with all speakers running (including subs) and measure at listening position(s) and adjust.

I don’t know if you can save more than stereo impulse files in REW, but if not than it still should be possible with Roon using Procedural EQ containing one Parametric EQ for each channel, and then one Parametric EQ affecting all channels. You have to input the EQ numbers manually though.

You can also use Roon to enable/disable individual speakers (Procedural EQ -> Mute channels), and play the Pink PN noise from Roon while doing measurements in case you can’t select individual satellites in REW (you can save the Pink PN from the Generator to a wav file and play it from Roon). Just make sure you disable all DSP in Roon while doing the measurement.

This way might not be 100% optimal for 5.1/5.2 systems, but it should get you part of the way at least.

(Scott Fletcher) #126

Just received my UMIK-1. Its occuring to me i’m running my Roon core in a QNAP TVS-471…
Clearly this is adding further complexity in terms of the process articulated very nicely above. Thanks Magnus.

Has anyone on here been in the same position and implemented successfully? Any tips?

(Magnus) #127

You can save the “Pink PN” sound from the “Generator” window and play it from Roon (either as a .wav file or convert it to flac first). Just make sure you disable all DSP in Roon before playing and doing measurement. If you plan to do the extra fine-tuning in 16a, you need to do that anyway.

If you have the computer a long way from the listening position, you probably want a friend to help with the start/stop RTA recording, and an USB-extension cable (or use a laptop, but place it on floor behind/beside you so it’s fans don’t affect the measurement).

Let us know how it goes!

(Pascal Filion) #128

@Magnus Thank you for providing such a detailed walkthrough.

However the sound feels dull and lifeless after applying my Convolution filters.

You mention “Be wary of trying to boost deep and steep holes, and if REW does it for you it might be best to manually remove that filter”. Would you care to elaborate how one should tame the REW default Task Filters ?



(Magnus) #129

Check this post, but I don’t think that’s your problem. Boosting destructive nodes tends to produce to much unwanted noise.

Getting a flat frequency response will make the sound more focused and narrow, which if you are used to the opposite will make it feel like you get less body and life. Try listening to it for a day or two, and then switch back and compare.

If you still don’t like the room corrected sound, you can limit the correction to where its needed most (usually lower frequencies), for example 20 - 500 Hz. You can also tell REW to have a higher tolerance by changing the Flatness target in the EQ window before generating the filter (for example, setting it to 5 dB will make REW only correct those places with a >5 dB difference).

Using room correction software with ROCK
(Pascal Filion) #130

The filter sounds too extreme compared to the native B&W sound. That is probably it, I will try limiting the REW tolerance levels.

I will report back once I get the chance to start tweaking!



(Steve) #131

We’re all different but when I first started with EQ I found it took me a week or two to adjust.

A ‘bright’ sound I’d aclimatised to over months/years, sounded ‘flat’ when corrected. But then after a few weeks this new sound became normal, and disabling EQ bought back a horrible brightness. I’ve learnt that in sound terms I’m quite sensitive to change, so need time to readjust. That’s why I pretty much gave up comparing cables/components etc (well, nearly on the components). A lively, brighter sound can also seem really good initially but can be tiring after a while - I think more than a few speaker manufacturers crack up the treble to lure people in.

I also found that default Dirac style curves weren’t my preferred - I like to keep a bit of the bass boost the room gives (but tame it). I think a lot of music I listen to assumes (or benefits from) a bit of room interaction!

Anyway, my main Room EQ experience wasn’t with Roon /REW, but thought it might be useful to share. I would definitely set and try and forget for a few weeks and not bother even tweaking (hard to resist I know), then go back to the original sound and see if you prefer it.

(Magnus) #132

I got my new speakers today, Dynaudio X14A which sounds amazing. This time I thought I would do room correction but still keep the Dynaudio signature sound as much as possible, so I did the measurements, placed the target level on the 500 - 1000 range, and then generated the filters from 20 to 500hz (and skipped step 16a). This way I still did the necessary lower frequency room correction, while allowing the speakers to do their thing at higher frequencies.

The reason I choose to place the target at 500 - 1000 Hz is because of the measurements, for me that area was pretty flat and similar in both speakers so should represent a good target level.

The result: very good, if this is better or worse than a full 20 - 20khz room correction is hard to tell and is up to the listener (I imagine a studio musician would prefer the full room correction). But female voices, guitars etc now is mostly untouched, and the sound is slightly less “clinical”.

(Mark Allen) #133

Congrats on your new Speakers!

(Magnus) #134

Cheers, I am very happy with then, and they sound even better with room correction done. They where also easy to integrate with the subwoofer since they have a 60/80 Hz high-pass filter setting.

(Magnus) #135

Updated the guide with an alternative “Lite-guide”, which aims to correct the most crucial room correction issues without going all the way. Should be a good guide to try if you did the complete guide and didn’t like the result, or if you want the sound to be more like what you had before room correction.

The lite-guide should also be a little less resource demanding, for those who run Roon on a limited host.


So I had previously attempted some DRC via REW with no so great results. I did it again, this time using some key findings from @Magnus and his great guide. For me, the key things were:

  • using RTA instead of a log sweep. This is so much easier and comprehensive than doing log sweeps from multiple positions (eg. emulating Dirac). If you watch the RTA average window and don’t stop moving the mic you will quickly settle on an average response that doesn’t change much. i.e you aren’t missing anything.
  • taking separate measurements and creating filters for each speaker. I was surprised, though shouldn’t have been in hindsight, at the differences in bass response between Left and Right.
  • getting the Target Level set correctly is of extreme importance, particularly if your correction range is low.

I think that step 16b is crucial, not optional - I too have noticed how REW won’t correct anything once the bottom drops off. This can result in less bass at the bottom after filters are applied. Correcting for this is very important. And I noticed that the issue was much worse for one speaker than the other, so I actually manually applied an extra EQ filter in the REW EQ window for each speaker and incorporated them into the final filters. Instead of doing a single correction to both speakers via Roon PEQ after the fact. This way the filter response is accurate and flat on both speakers down to their real bottom.

I was able to easily generate the filters, zip up and install into Roon with ease. Much better than my first attempt!

I use the standard minimal target curve that is the default in REW and then I have a single Roon PEQ for quickly managing my personal house curve i.e. bass boost, more hf rolloff. Having your personal curve separate from any room correction makes it easy to adjust the flavor depending upon content or increase bass at low listening levels, etc. Basically, a digital tone control.

(Steve) #137

So many happy campers here. Really hoping I can find some time tomorrow to get some measurements.

Did you point the mic at the ceiling or the speakers? When I used Dirac I pointed at ceiling as I was sitting quite close. The advantage is it’s a lot easier to keep it uniform with all the measuring positions.

I have a stand/arm for the mic but I’m trying to think how I could keep the mic direction consistent if I’m using spirals or whatever. Or maybe it doesn’t matter as much with this approach?


From my brief time using RTA I am not sure it matters much. That’s the great thing with RTA- as long as you move the mic around at a steady pace and cover all of the space you are interested in (close in for single spot, arm’s length for sofa) you will get a good average of the response peaks/drops in this space. You will see the average coalesce fairly quickly and as long as you don’t keep the mic still it won’t change at some point.

As for mic direction I think it matters what kind of calibration file you have. I have an XTZ pro with only a standard calibaration meant for upright positioning. So that’s how I hold it. Umik can have both 90 degree and 0 degree calibration files depending on where you got it from. But again, given that RTA is capturing a moving average I am not sure how much it will matter.

Whereas with log sweeps you are capturing a small set of very accurate measurements and so precision is very important.

That’s my 2 cents …

(Magnus) #139

I have tried both, and it does not matter much as long as you have the correct calibration file. The one I use now I pointed towards the speakers though (which is the calibration file without _90 text). There is some difference though at higher frequencies, at least for my UMIK-1, so if you only have one calibration file point the microphone according to that one,


Dirac recommends upright/diffuse orientation as they seem to work around averages. Acourate, with it’s single high precision measurement or beam-forming, suggests horizontal pointed at the speakers (90 degree for umik). I would think for RTA measurements in REW you would just go with whatever calibration/orientation your mic has.