Using REW to do Room Correction

I added the EQ settings manually from REW (copied the numbers), if you use the convolution features you get the same (or even better) audio result, but you can’t see the changes in the EQ.

But if you see big filters applied in REW (especially if those modified the bass), you should hear a clear difference. The difference for me was mega-huge, but then Im not exactly in the perfect listening room :slight_smile:

You loaded them into the convolution section? That does not generate PEQ filters.

And the convolution section has no visualization.

Not many of us are, sadly…

Where else do you load them? REW generated a WAV file and the convolution section was the only place I could find that accepts a WAV file to import.

I think what Anders is saying, is that you’re talking about convolution, but your screenshot is of a parametric EQ. The two aren’t related, and you can’t visualise the convolution. So if it sounds right with the convolution applied you’re probably ok and it’s working.

That still leaves the mystery of where your PEQ values all with gains at 0 came from, since I don’t think you can import PEQ settings into Roon. Or is it - when I make a new PEQ it’s populated with several bands similar to yours but they’re all integer frequencies @ 60,250,1000,4000,8000 Hz. Maybe you (or someone else) accidentally added one and then a few extra bands?

Here are some suggestions for room correction in REW (these are based on my personal experience, and might not be applicable for all scenarios):

  • Use generic equalizer
  • Use var smoothing before generating filters in the EQ window (this will prevent to much modification of higher frequencies which tends to sound awful).
  • If you have deep holes in frequency, make sure you don’t try to boost them since they are likely destructive nodes and feeding more power will just make it worse. Use low values for Individual/Overall Max Boost to prevent this if needed.
  • Adjust the LF/HF cutoffs and slopes to get the “house curve” you want (the defaults works pretty good though, although I liked a little more bass so I modified mine).
  • Manually set target to match the high frequencies as good as possible, so most adjustments will be on lower frequency.
  • Remember to select speaker and if needed adjust values (personally I use None which works good for full range).
  • After generating filters with “Match response to target”, save impulse response to wav and use in Roon convolution engine, or copy filter values manually to the equalizer in Roon.

I also discovered that for me, I got best result if I did 2 measurements, one with microphone close to left ear and one with microphone close to right ear while sitting in normal listening position, and then averaging them. The difference was quite big compared to the normal way of doing it, which is placing the microphone where your head normally is. Obviously this only works if you have a very fixed listening position, but that way you get the measurements where your body/head sound absorbing is included. For big room I suspect this wont matter though.

So just to be clear - the WAV output from REW should be loaded into the convolution section, am I right? So do you not use the PEQ section with REW?

Thats correct, best is if you generate one wav for each output frequency supported, then zip them and use the zip file from REW. If you do it like that, the EQ will not show anything and you don’t need to make any EQ adjustments.

Yes. REW can do two things - generate PEQs which can be manually fed into Roons PEQ, or export waves for the convolution. So it would usually be one or the other.

Of course, nothing then stops you making additional adjustments with PEQ on top of the convolution - say you wanted more bass, or just felt something needed tweaking and it was easier to do it that way rather than go back and regenerate convolution filters.

It might be interesting to set it up both ways (convolution vs PEQ) and see how the sound output compares.

1 Like

I have done the comparison between Roon PEQ and Roon convolution.
This is the setup: MacMini USB -> Mutec MC3+USB -> Chord 2Qute (optical SP/Dif), Roon volume leveling, sample rate conversion to 192 kHz. For the convolution I used two identical mono (L and R) 192kHz impulse waves with 131k taps generated by rePhase with the PEQ settings in Roon.

Two observations: Roon PEQ dulls the sound quite a bit while convolution does not, even without PEQ correction (PEQ and individual correction “on”, but all PEQ gains at 0 dB).
Also, convolution correction seems to work stronger than Roon PEQ. For instance, applying -1 dB gain on a frequency with convolution sounds to my ears more like -3 dB with Roon PEQ.

I am quite happy with convolution in Roon.

1 Like

If thats correct then the devs at Roon has some work to do. But you say you generate the wav files in rephase, but rephase does adjust the phase also. If you generate the wav from REW, and then apply the same REW EQ to Roon, and switch between those, is there still a difference (there should not be as far as I know).

I could not hear any difference switching on/off an “empty” EQ in Roon.

I used the wavs from rePhase without changing the phase. In rePhase the phase changes caused by the PEQ are visualized but you can choose not to compensate for that (it is a manual action), which is what I did for the Roon PEQ/convolution comparison. I hear minor differences when I compensate for phase changes, but I can not tell if it is better or worse. Just different to a very small extent.

I have to mention that the Mutec MC3+ removes a lot of the jitter “veil” and lowers the background noise a lot. Maybe that is why I am picking up these differences.

I wrote a complete guide about how to do room corretion in Roon, in a way that has given me the best result. Check it out: A guide how to do room correction and use it in Roon

1 Like

When copying the Dirac measurment into roon manually, how can I set different filters for each loudspeaker?
The Dirac measurment shows different curves for each loudspeaker and sets different filters for each one.
At least this is how I think Dirac works, because you can set the target curve for each one individually.

Unfortunately, you can’t. The PEQ settings in Roon apply to both left and right speakers. You can use the average dirac measurement (making sure both L & R channels are linked together) and use that as your guide.

In light of Roon linking the L and R speakers together, how would one use REW to generate PEQ filters for Roon for a surround environment with two subwoofers? Or is Roon sound correction solely for a two channel system? JCR

@Mirco_Strauss @Jeffrey_Robbins Away from my system at the moment, but i think you can use Procedural to set up over-channel PEQ.

I just took a look more deeply and @AndersVinberg is correct. You can add a procedural EQ filter and within that create Parametric EQ operations that apply to certain channels. Hadn’t seen that before.

You can also generate stereo .wav impulse files for use with the convolution engine in Roon.

I noticed most folks on this thread were exporting REW’s measurement sweep to Roon and playing that for measurement.

I exported REW’s pink noise generator and then used the real-time measurement in REW.

Is there any reason to prefer one over the other?

Thanks,
Jeff.