Room correction - Linux

[Lilith]

Hi, I'm on Linux and I'm producing music on Linux. For room correction I used the Sonarworks plugin that was supported by Sonarworks for ~2-3 years now. The measurement part was done with Windows. I bought SW last year and suddenly this year they dropped Linux support. The plugin is still running, but it's not future save. As a workaround I'm using now REW together with the LSP parametric EQ (https://lsp-plug.in/).

When using Sonarworks the phantom center is slightly better and I assume the reason is the lack of phase correction with REW /LSP. Does anyone know what / how Sonarworks is doing with respect to the phase? Are for instance the filters optimized for linear phase and little phase distortion?

This is a measurement from REW with 1/24 smoothing:

And with LSP I only apply two corrections at ~70 Hz and 130 Hz. Sonaworks also corrects the dip at around 1000 Hz which could be due to desk reflections:

The sound of these two methods is similar and much better compared to none correction. The 130 Hz mode makes everything muddy. My room is well treated with bass traps, ceiling cloud, absorbers on the front wall for SBIR and panels on the side. The 130 Hz is very likely the 002 mode. Ringing times are generally fine.

1. Would you correct the dip at around 900 Hz or would you get used to the sound of the speakers instead?
2. Is there any good alternative for room correction on Linux (hardware or software)? I looked into DRC, but was not able to get it running (problem was with generating the file from the IR response). I thought about mini DSP, but are there XLR versions and does this correct the phase or is this in the end the same result like the "simple" EQ correction?
3. Dirac, etc. are all not running on Linux
4. Does Sonarworks works with e.g. WINE? I didn't find anything and I don't believe it will work

I'm also planning to upgrade my speakers (Presonus Eris 5 so far...) to get more bass (down to at least 50 Hz, lower even better). I thought about buying the GENELEC 8330 which would solve partially the room correction issue (I just need a windows PC for configuration), but I wonder if these are useful without a sub. When comparing the distortion plots they seems to be similar to the Adam T5V, so I wonder if it makes sense to get the Genelec when the Adam will (maybe) perform similar. The T7V at least on paper give me more bass compared to the GENELEC, so I wonder if this is not the better option. Other option: Yamaha HS7, or JBL 306, ??

I tried to work with windows the last 3 days, but I really like to stay on Linux. If anyone has some tips or recommendations this would be great .

 

[dasdoing]

I use http://drc-fir.sourceforge.net/
it's not idiot-proof like Sonaworks, quite the oposite, but since you use Linux........
anyways, here is a good GUI for it: https://www.ohl.to/about-audio/audio-softwares/align2
the correction aprouch is kind of comparable to Accourate

 

[MRC01]

On Linux, there is always PulseEffects, a toolbox of a variety of DSP including parametric EQ, and many other things.

 

[iMickey503]

Watching this thread. I have some aspirations coming up with a linux box dsp setup. Want to do a full out idiot proof tutorial that even I can follow.

The most I have used Linux Multimedia studio before, but not under linux. Always windows. Planing to take the plunge this year as I got a few projects I want to get going before the end of the year and Holiday displays.

 

[Daverz]

2. Is there any good alternative for room correction on Linux (hardware or software)? I looked into DRC, but was not able to get it running (problem was with generating the file from the IR response).

I use DRC-FIR with excellent results. I think it would be worth it to work through the issues you are having with it here.

As for software that requires Windows, I would suggest installing it on VirtualBox.

 

[MRC01]

On the subject of Linux audio setup, a while back I posted before a setup guide that some folks found useful. The biggest flaw in PulseEffects is it locks in a single sampling rate. You can set up Pulseaudio to avoid resampling and that works fine, but as soon as you run PulseEffects it disables this. I've seen discussion on the PulseEffects developer site how to avoid this, so they're aware of it even though they don't have a solution yet.

 

[Lilith]

I use http://drc-fir.sourceforge.net/
it's not idiot-proof like Sonaworks, quite the oposite, but since you use Linux........
anyways, here is a good GUI for it: https://www.ohl.to/about-audio/audio-softwares/align2
the correction aprouch is kind of comparable to Accourate

Oh, this is new to me. Thanks! Will look into it and ask you in case of problems

 

[Lilith]

On Linux, there is always PulseEffects, a toolbox of a variety of DSP including parametric EQ, and many other things.

I heard of it, but as I need low latency and it should work via Jack I don't want to use pulse.

 

[Lilith]

I use DRC-FIR with excellent results. I think it would be worth it to work through the issues you are having with it here.

As for software that requires Windows, I would suggest installing it on VirtualBox.

Can one see what DRC-FIR is doing or is it a kind of black box?

 

[digitalfrost]

Another vote for DRC-FIR. I do my measurements with REW and then use DRC-FIR for the correction. You cannot simulate the result like with Acourate, but you could of course re-measure if you don't move the microphone in between.

 

[Lilith]

Another vote for DRC-FIR. I do my measurements with REW and then use DRC-FIR for the correction. You cannot simulate the result like with Acourate, but you could of course re-measure if you don't move the microphone in between.

How does this work as REW can't use Jack?

 

[digitalfrost]

How does this work as REW can't use Jack?

I use it on windows, so no idea about jack. Measuring with DRC is not so easy imho, with REW it's easy. I measure with REW, then export the left and right measurements without any modification. You could add microphone compensation before exporting I think, but I do it in DRC. To convert the files:

sox.exe l48.wav -t f32 LeftSpeakerImpulseResponse48000.pcm
sox.exe r48.wav -t f32 RightSpeakerImpulseResponse48000.pcm
copy /Y LeftSpeakerImpulseResponse48000.pcm  ..\drc-3.2.0\sample
copy /Y RightSpeakerImpulseResponse48000.pcm  ..\drc-3.2.0\sample

 

[dasdoing]

Oh, this is new to me. Thanks! Will look into it and ask you in case of problems

prepare a month or two if you give it a shot lol
most complex software I've ever used

 

[Lilith]

I use it on windows, so no idea about jack. Measuring with DRC is not so easy imho, with REW it's easy. I measure with REW, then export the left and right measurements without any modification. You could add microphone compensation before exporting I think, but I do it in DRC. To convert the files:

sox.exe l48.wav -t f32 LeftSpeakerImpulseResponse48000.pcm
sox.exe r48.wav -t f32 RightSpeakerImpulseResponse48000.pcm
copy /Y LeftSpeakerImpulseResponse48000.pcm  ..\drc-3.2.0\sample
copy /Y RightSpeakerImpulseResponse48000.pcm  ..\drc-3.2.0\sample

let me try this quickly, I remember there was an issue....

 

[Lilith]

let me try this quickly, I remember there was an issue....

quickly ... have to understand what I did some months ago... completely forgot everything again ..

 

[Lilith]

I start here:

 

[Daverz]

Another vote for DRC-FIR. I do my measurements with REW and then use DRC-FIR for the correction. You cannot simulate the result like with Acourate, but you could of course re-measure if you don't move the microphone in between.

You can have DRC-FIR output a test convolution. Pretty accurate IME.

 

[digitalfrost]

Oooh I didn't know. How would I display what it looks like? Like let's say I have the measurement I took in REW. Can I somehow apply the convolution there and see the results? That would really help with fine tuning excess phase...

 

[Daverz]

Can one see what DRC-FIR is doing or is it a kind of black box?

The text output is very "chatty", and options are provided for producing intermediate output for just about every step if you are interested.

That and the source code is very clean and readable if you know C.

What makes it a bit opaque is that knowing what to tweak in the configuration file is difficult if you don't know the theory of the computations it's using. The options are not expressed in layman's terms.

I've never gotten beyond using the provided configurations, and I usually just use the "normal" one.

More on tweaking the parameters here:

http://drc-fir.sourceforge.net/doc/drc.html#sec49

 

[Daverz]

Oooh I didn't know. How would I display what it looks like? Like let's say I have the measurement I took in REW. Can I somehow apply the convolution there and see the results? That would really help with fine tuning excess phase...

I convert it to wav and import it in REW as an impulse response.