Page 2 of 2

Re: REW with Software EQ

Posted: Thu Aug 29, 2019 1:16 pm
by Capoeira
btw: there is Qloud for the simple meassurement part: https://github.com/molke-productions/qloud
lilith wrote:Thanks!!! Just tested it and after some fiddling it works. I added the lines in .asoundrc
Then start Cadence, Jack and the alsa jack bridge.

Image

Then started REW and choose Default Device as output. After this a rawjack device appears in Catia and a EQ can be added in front of it.

Image
if you activate alsa-jack-brige in Cadence it overwrites you .asoundrc with a snd-aloop one
CrocoDuck wrote:
lilith wrote:
CrocoDuck wrote:
Just for curiosity, I am not very familiar with REQ, but what can you export from REW? Can you export the whole impulse response in some way?
You can export EQ settings as a file based on your measurement for various hardware or plugins (e.g. miniDSP). I was just interested in equalizing three frequencies on which I have have strong room modes (~40, 50 and 120 Hz). You can also show the correction settings as a table (frequency, Q-factor, damping) which you can use for any EQ (plugin or hardware) by setting the parameters by hand.
The feature list suggests that it is possible to export the measured impulse response as WAV. If you manage to do that, it would be fun to try to invert it and use the inverse in a convolution reverb. Octave can be used for that pretty easily. This should give you the highest degree of compensation for the transmission line you measured. This means that, if you put your head where the microphone was during the measurement, then the compensation will be not optimal as your Head Related Transfer Functions are not the mic response, but maybe it will work OK anyway, especially at low frequency I think.

If you have time to spare, try to give us your impulse response, if you figure out how to do it. I can try to calculate the inverse for you and we can give it a spin, and compare it with the EQ solution.
I wonder why DRC and sorts use complicated convolution filters and not just invert the IR. there must be a good reason for this (I hope)

Re: REW with Software EQ

Posted: Thu Aug 29, 2019 1:55 pm
by lilith
Thanks, so I don' t have to start the alsa bridge. It still doesn't work. I can choose the rawjack in REW for input and output

Image

But Catia only shows the input channels

Image

Thanks for the tip with the other program, I'll give it a try.

Re: REW with Software EQ

Posted: Thu Aug 29, 2019 3:19 pm
by Capoeira
lilith wrote:Thanks, so I don' t have to start the alsa bridge. It still doesn't work. I can choose the rawjack in REW for input and output

Image

But Catia only shows the input channels
yea. it only seams to work with the bridge.

I had artifacts in REW with this solution though. atm REW is acting all strange here anyways. I prefer using it in Windows.
there should be a jack plugin for java, but there doesn't seam to be

Re: REW with Software EQ

Posted: Thu Aug 29, 2019 3:25 pm
by lilith
Capoeira wrote:
lilith wrote:Thanks, so I don' t have to start the alsa bridge. It still doesn't work. I can choose the rawjack in REW for input and output

Image

But Catia only shows the input channels
yea. it only seams to work with the bridge.

I had artifacts in REW with this solution though. atm REW is acting all strange here anyways. I prefer using it in Windows.
there should be a jack plugin for java, but there doesn't seam to be
I'll try again later, but everytime it behaves different and REW is also much slower (the GUI) compared to using is with Jack turned off.

Re: REW with Software EQ

Posted: Sat Aug 31, 2019 1:29 pm
by CrocoDuck
Capoeira wrote:
I wonder why DRC and sorts use complicated convolution filters and not just invert the IR. there must be a good reason for this (I hope)
Inverting the IR is often dangerous, meaning that can result in not causal filters, i.e. a filter you cannot actually implement in practice unless you add delay. Normally, the best approach is to identify a new IR with the goal of "undoing" another. This way you get a compensation with no additional latency beside that necessary due to the phase response of the compensation filter.

Re: REW with Software EQ

Posted: Sat Aug 31, 2019 1:59 pm
by Capoeira
CrocoDuck wrote:
Capoeira wrote:
I wonder why DRC and sorts use complicated convolution filters and not just invert the IR. there must be a good reason for this (I hope)
Inverting the IR is often dangerous, meaning that can result in not causal filters, i.e. a filter you cannot actually implement in practice unless you add delay. Normally, the best approach is to identify a new IR with the goal of "undoing" another. This way you get a compensation with no additional latency beside that necessary due to the phase response of the compensation filter.
I think the real problem are the artifacts which it would produce.
If you read http://drc-fir.sourceforge.net/doc/drc.html the amount of correction has to be limited in order to prevent them to occure

Re: REW with Software EQ

Posted: Wed Mar 17, 2021 11:01 am
by lilith
CrocoDuck wrote: Sun Jan 21, 2018 5:34 pm
lilith wrote:
CrocoDuck wrote:
Just for curiosity, I am not very familiar with REQ, but what can you export from REW? Can you export the whole impulse response in some way?
You can export EQ settings as a file based on your measurement for various hardware or plugins (e.g. miniDSP). I was just interested in equalizing three frequencies on which I have have strong room modes (~40, 50 and 120 Hz). You can also show the correction settings as a table (frequency, Q-factor, damping) which you can use for any EQ (plugin or hardware) by setting the parameters by hand.
The feature list suggests that it is possible to export the measured impulse response as WAV. If you manage to do that, it would be fun to try to invert it and use the inverse in a convolution reverb. Octave can be used for that pretty easily. This should give you the highest degree of compensation for the transmission line you measured. This means that, if you put your head where the microphone was during the measurement, then the compensation will be not optimal as your Head Related Transfer Functions are not the mic response, but maybe it will work OK anyway, especially at low frequency I think.

If you have time to spare, try to give us your impulse response, if you figure out how to do it. I can try to calculate the inverse for you and we can give it a spin, and compare it with the EQ solution.
Bit late ... I will try this in the next days. REW can export the impulse response of the EQ filters (which is then roughly the inverse room response?). I try to load this into the LSP convolver plugin for room correction. It´s still unclear to me if this approach offers any advantages compared to using the EQ settings with a parametric EQ or is it exactly the same in the end? I also can compare the result to Sonarworks then.

just for me to find again:
https://www.audiosciencereview.com/foru ... 2.5/page-9