Son of MultiBand Splitter: 4 Band Splitter for Live 7

Nogi · Post by **Nogi** » Fri Apr 18, 2008 8:07 pm

I while back I posted a multiband splitter for Live 6 and more recently rhythminmind posted a link to his own Live 7 versions. Which got me thinking again...

I put off updating my Live 6 version to Live 7 for two reasons. 1 - I was a bit bummed that Live 7 wouldn't import a Live 6 device properly and 2 - something was rubbing me the wrong way and I needed time to investigate further. Namely, the sound of certain bands wasn't what I expected and FX seemed to spill over to other bands after they were all summed back up. Say, a delay added to the MID band would be applied to the entire frequency spectrum. 'Spectrum' backed my ears up and, eventually, I traced it to the process of inverting the audio phase for a given band and then effecting the remainder. Unfortunately, the remainder is more than just that. In this case it would seem 2 - 1 does not equal 1 but in fact equals 2 + (-1). Meaning, it remains full spectrum audio. So if we treat it as any regular freq band and add FX to it we will smear the FX over the entire spectrum. What's more the cancellation we rely on for transparency will have been compromised when they are summed.

What to do? I created a new Live 7 compatible 4 band splitter that removes the phase inversion from the band splitting process and the individual bands themselves and, thus, from the process of adding FX. The 'EQ8' was used as it is (surprisingly) less CPU intensive than 'EQ3' at 48db/8ve. Using 'Spectrum' you can verify that the band behavior is as expected by solo'ing them.

Download: http://www.base2audio.com/abletonstuff/ ... r_v3.0.zip

Your feedback on ways to improve this device is appreciated.

'Utility' is god.

Cheers.

rhythminmind · Post by **rhythminmind** » Fri Apr 18, 2008 9:51 pm

Nice work! As a self contained unit this has beautiful flow.
But there is flaw.
All the chains in your rack are needed to stay unmodified to retain the transparency & have true separation without bleeding.

Example - If you put FX or alter the "HI Band". Frequencies will leak from the "no FX here" chain.

Good stuff tho.

I love the vertical layout...
So So close.. Just needs one more summing stage.. Same problem i had with my rack. Thats why i had to make it two layers deep.

Afro88 · Post by **Afro88** » Fri Apr 18, 2008 10:04 pm

rhythminmind wrote:So So close.. Just needs one more summing stage.. Same problem i had with my rack. Thats why i had to make it two layers deep.

Could you post an updated version of your rack then please? I built yours from your signal flow picture, and it sounds great by itself (totally transparent) but as soon as you start to put fx on the bands as you say you get bleed from the phase inverted channel. I couldn't figure out a way around it while still maintaining transparency...

rhythminmind · Post by **rhythminmind** » Fri Apr 18, 2008 10:09 pm

http://www.rhythminmind.net/presetblog/ ... udio-rack/

You can use the rack itself. But i found i like using it bussed to channels better. You can download the mixer template as well.

Nogi · Post by **Nogi** » Fri Apr 18, 2008 11:01 pm

rhythminmind wrote:But there is flaw.
All the chains in your rack are needed to stay unmodified to retain the transparency & have true separation without bleeding. Example - If you put FX or alter the "HI Band". Frequencies will leak from the "no FX here" chain.

I'm not seeing this. Could you explain further?

The approach I've tried to take here is to have each band take its own exclusive 'bite' out of the overall spectrum without reliance on phase inversion except in final summing. NOTE: The adjacent bands are daisy-chained together. Of course, this puts the frequency band selection and residual bleed at the mercy of the filter slope but 48dB/8ve is pretty good.

For comparison to other devices, check for bleed across bands. Drop a 'Simple Delay' into each band with the feedback turned up to, say, 50%. Turn the 'Simple Delay' on in one of the bands at a time. Put a 'Spectrum' on the master. Play for a few seconds then stop. You can see the decay of the delay and the effected frequencies via Spectrum. All versions of the band selection by phase inversion devices I have tried have very significant FX bleed across the spectrum.

rhythminmind · Post by **rhythminmind** » Fri Apr 18, 2008 11:06 pm

Just do a simple gain change on the hi band or just mute the hi band. Now that it is gone there is nothing to counteract with the highband in the "no fx" chain so the "no fx" chain HI will leak out.

Nogi · Post by **Nogi** » Fri Apr 18, 2008 11:48 pm

rhythminmind wrote:Just do a simple gain change on the hi band or just mute the hi band. Now that it is gone there is nothing to counteract with the highband in the "no fx" chain so the "no fx" chain HI will leak out.

Thanks for the response. The purpose of the HI^-1 as I see it is to make a hole in the CLEAN chain spectrum for HI+FX to then fill and not to directly cancel against the HI chain - it cancels against the part of HI that exists in CLEAN.

Here's a test (besides the one above that uses Simple Delay) that shows why I'm going down this road. Make a comparison rack with two chains, one that has a garden variety LO pass and one that has a LO pass made by putting an inverted HI pass against a clean chain. They are theoretically the same but yield very different results.

I am working to refine the filtering at this point to eliminate the necessity of phase inversion altogether.

rhythminmind · Post by **rhythminmind** » Sat Apr 19, 2008 12:05 am

Yeah but your undoing all the nice separation that you have done so far.
Any fx that you apply are only 50% wet.

Nogi · Post by **Nogi** » Sat Apr 19, 2008 12:20 am

Yeah. It's a compromise and only as good as the filtering (I will bring this quadcore to its knees) but many FX are good at less than 100% wet and I like the increased control over the frequency spread.

My new Live 8 feature request is for a HI/LO pass filter set with variable slope that are perfect inverses for each slope setting.

So, back to the lab for now...