Most Transparent Multiband Splitter

VST, AU, etc. plug-in Virtual Effects discussion
User avatar
KVRian
1042 posts since 14 Apr, 2008 from /* whitenoise */

Post Tue Sep 28, 2021 7:21 am

NAD wrote:
Mon Sep 27, 2021 12:36 pm
There's this trick for creating a 3-band crossover you can do with some clever routing if your DAW happens to support it:
  • Send the signal through a low-pass filter to one channel for processing the low frequencies
  • Send the signal through a high-pass filter to another channel for processing the high frequencies
  • To a third channel send the full range signal and also send the high- and low-passed signals but invert them so you end up with only the mid frequencies
From there you just buss those three channels together or group them in a folder track and obviously don't forget to disable the master send on the original track.
I got this idea from a fantastic Dan Worrall video in which he uses linear phase filters although even if you use any other type of EQ this still works and will null with the original signal.
Autobot wrote:
Mon Sep 27, 2021 4:54 am
That said I use EQ8 in Live to split the signal. If I measured it right there is no harm to the signal at all. Nulls perfectly and no phase issues. What you see is EQ8 Live (no oversampling) with a 48db four band split. SPAN and Bertom EQ Curve Analyzer used.
Aesaire wrote:
Mon Sep 27, 2021 10:37 am
EQ8 is just the stock equalizer in Ableton though right? Could you detail the steps of your test or provide a project file? I'm extremely skeptical that a 48db four band split would perfectly null. No crossover filter I've tested comes remotely close to this from linear phase to minimum phase.

Only thing remotely similar when using ISOL8, where if the multichannel mode or mute/solo band buttons wasn't engaged it didn't actually start the crossover filter. Thought it was perfectly transparent before that.
Aesaire wrote:
Tue Sep 28, 2021 3:19 am
Wow, this works. I wonder why no one has tried turning this into a plugin.
Yes it works. No idea why it's not there as a plugin. What Autobot describes is also explained here in 2015 : http://blog.dubspot.com/ableton-live-tu ... ion-racks/

Did not made any testing so can't say how *clean* the splitting process is.
/* whitenoise */ /* abandon */ /* reincarnated */

KVRian
566 posts since 4 Jan, 2007

Post Tue Sep 28, 2021 7:57 am

noiseresearch wrote:
Tue Sep 28, 2021 7:21 am
Yes it works. No idea why it's not there as a plugin. What Autobot describes is also explained here in 2015 : http://blog.dubspot.com/ableton-live-tu ... ion-racks/

Did not made any testing so can't say how *clean* the splitting process is.
Subtracting IIR filters has issues. Yes, the signal will reconstruct perfectly, but as soon as one band is processed things break spectacularly because of phasing issues.

Just change the gain on the central band and look at the spectrum of the reconstructed signal. It won't be a step, it will have ripples. The more poles/steepness the worse it becomes, being 1 pole pretty usable.

In any case, by coincidence on MixMaxtrix 1.1 I added a crossover based on this, but just for creative purposes/destruction (called WTF-crossover). I guess this build here will be pretty similar to the final 1.1 one:
https://github.com/RafaGago/artv-audio/ ... 1283307981

EDIT: I removed the binaries because the waveshaper bug is still present and causes some feedback tones.

KVRian
566 posts since 4 Jan, 2007

Post Wed Sep 29, 2021 11:48 am

Ok now got the bug fixed for good, so I released. There you have it, yet another crossover.

https://www.kvraudio.com/product/mixmaxtrix-by-artv

This one has one or two distinguishing features, it allows offsetting the cutoff on the L/R channel and the slope is a per-band setting, not a global one. Maximum 4 bands.

KVRist

Topic Starter

58 posts since 13 Sep, 2020

Post Wed Sep 29, 2021 11:02 pm

rafa1981 wrote:
Tue Sep 28, 2021 7:57 am
noiseresearch wrote:
Tue Sep 28, 2021 7:21 am
Yes it works. No idea why it's not there as a plugin. What Autobot describes is also explained here in 2015 : http://blog.dubspot.com/ableton-live-tu ... ion-racks/

Did not made any testing so can't say how *clean* the splitting process is.
Subtracting IIR filters has issues. Yes, the signal will reconstruct perfectly, but as soon as one band is processed things break spectacularly because of phasing issues.
Can't you get around this problem by sending the inverted signal pre-FX, then applying the effects on each band afterwards. Can set up an unprocessed inverted signal in in the low and high bands.
rafa1981 wrote:
Wed Sep 29, 2021 11:48 am
Ok now got the bug fixed for good, so I released. There you have it, yet another crossover.

https://www.kvraudio.com/product/mixmaxtrix-by-artv

This one has one or two distinguishing features, it allows offsetting the cutoff on the L/R channel and the slope is a per-band setting, not a global one. Maximum 4 bands.
By default the preset WTF-Crossover4 sends to bus 3,4 and 5 instead of 2,3 and 5. Otherwise it works fine.

KVRian
566 posts since 4 Jan, 2007

Post Thu Sep 30, 2021 1:16 am

It is deliberate. I did set the default for using MixMaxTrix as a multiband FX, so to process the low band, it has to be sent to the neighboring mixer with the button that is between mixer limits. Easy to make a preset correcting it.

Unrelated. You can do this experiment. Set the WTF-Crossover bands to e.g. 30dB/Oct. Add a 4/6 dB gain to one of them and look at the resulting spectrum of the summed bands a spectrum analyzer or with Bertom Audio EQ analyzer. Then compare with the Linkwitz-Riley one. You will clearly see why IIR crossovers based on subtraction are flawed.

https://www.bertomaudio.com/eqca.html

Or looking it from another perspective, why would DSP people bother with Linkwitz-Riley if a much simpler and light on CPU solution worked better?
Last edited by rafa1981 on Thu Sep 30, 2021 1:20 am, edited 1 time in total.

KVRian
566 posts since 4 Jan, 2007

Post Thu Sep 30, 2021 1:20 am

duplicate

KVRist

Topic Starter

58 posts since 13 Sep, 2020

Post Thu Sep 30, 2021 4:10 am

rafa1981 wrote:
Thu Sep 30, 2021 1:16 am
It is deliberate. I did set the default for using MixMaxTrix as a multiband FX, so to process the low band, it has to be sent to the neighboring mixer with the button that is between mixer limits. Easy to make a preset correcting it.

Unrelated. You can do this experiment. Set the WTF-Crossover bands to e.g. 30dB/Oct. Add a 4/6 dB gain to one of them and look at the resulting spectrum of the summed bands a spectrum analyzer or with Bertom Audio EQ analyzer. Then compare with the Linkwitz-Riley one. You will clearly see why IIR crossovers based on subtraction are flawed.

https://www.bertomaudio.com/eqca.html

Or looking it from another perspective, why would DSP people bother with Linkwitz-Riley if a much simpler and light on CPU solution worked better?
Ah okay, I see the problem now. Results if anyone is interested. The Linkwitz-Riley filter is much more transparent once there are any gain changes or processing. The WTF-crossover is from the MixMaxTrix plugin, which emulates the crossover technique in this video.

Here is the comparison for+5db of gain on 1 band.
Crossover Filter comparison 5db gain.jpg
This is the comparison for some softclipping applied to 1 band.
Crossover Filter comparison softclipping.jpg
You do not have the required permissions to view the files attached to this post.

KVRist

Topic Starter

58 posts since 13 Sep, 2020

Post Thu Sep 30, 2021 4:10 am

duplicate

KVRian
566 posts since 4 Jan, 2007

Post Thu Sep 30, 2021 5:02 am

Notice that what Dan Worall does in his video is right, he is using ReaFIR, which is a linear-phase filter (introduces latency). Those can be subtracted.

Doing it with Ableton's EQ8 is wrong, as those filters are IIR (minimum-phase).

User avatar
KVRian
1042 posts since 14 Apr, 2008 from /* whitenoise */

Post Thu Sep 30, 2021 5:19 am

rafa1981 wrote:
Thu Sep 30, 2021 5:02 am
Notice that what Dan Worall does in his video is right, he is using ReaFIR, which is a linear-phase filter (introduces latency). Those can be subtracted.

Doing it with Ableton's EQ8 is wrong, as those filters are IIR (minimum-phase).
Thank you for clarifying this. Which splitter would you recommend? Is Crossover by RS-MET a good option?
Image
https://www.kvraudio.com/product/crossover_by_rs_met
/* whitenoise */ /* abandon */ /* reincarnated */

User avatar
KVRian
1042 posts since 14 Apr, 2008 from /* whitenoise */

Post Thu Sep 30, 2021 5:44 am

I just see that Crossover is 32bit only. Klevgränd Gaffel seems decent and affordable (EDU price). I have Triad by Unfiltered Audio but it mess up the signal when split in bands
/* whitenoise */ /* abandon */ /* reincarnated */

KVRian
566 posts since 4 Jan, 2007

Post Thu Sep 30, 2021 5:45 am

I'm biased, but as the author of MixMaxtrix I would go for that for Linkwitz-Riley unless you are on Mac or need more than 4 bands, as they allow offsetting the cutoff between the left and right channels and have variable slope per band.

I seem to recall that RS-METs Crossover was Linkwitz-Riley too, so they should be soundwise equivalent. I don't think that mine reaches so high slopes, I stopped at 48dB.

I don't know, maybe you can measure CPU and pick the lightest.

KVRian
566 posts since 4 Jan, 2007

Post Thu Sep 30, 2021 5:48 am

Gaffel is Linkwitz-Riley too.

User avatar
KVRian
1042 posts since 14 Apr, 2008 from /* whitenoise */

Post Thu Sep 30, 2021 6:01 am

rafa1981 wrote:
Thu Sep 30, 2021 5:45 am
I don't know, maybe you can measure CPU and pick the lightest.
I think I'll go with that :D Do you know what kind of tech TRIAD by UA use? Seems a bit of to my ears often. I'll give Gaffel and Multipass a try.
/* whitenoise */ /* abandon */ /* reincarnated */

KVRist

Topic Starter

58 posts since 13 Sep, 2020

Post Thu Sep 30, 2021 9:07 am

rafa1981 wrote:
Thu Sep 30, 2021 5:02 am
Notice that what Dan Worall does in his video is right, he is using ReaFIR, which is a linear-phase filter (introduces latency). Those can be subtracted.

Doing it with Ableton's EQ8 is wrong, as those filters are IIR (minimum-phase).
Oh I was under the impression that your WTF-Crossover was linear phase. Trying processing on IIR subtraction with ReaFIR seems to give slightly worse but similar results to Linkwitz-Riley filters.

Return to “Effects”