Most Transparent Multiband Splitter

[rafa1981]

WTF is based on IIR subtraction, that's why it doesn't introduce latency. ReaFIR is FIR, so it shouldn't have these problems and almost null with the original signal if latency compensation is setup correctly. Weird.

[midi_transmission]

Please don't think that linear phase filters are better. They are just different. There is no perfect filter. There is no invisible crossover.

The idea of multiband is that the benefit is bigger than the harm you do. What works best depends on the material and processing you do. Even the slops makes a big difference.

I think a x-over filter should offer some different filter types and also some slopes, so that you can try what works best for every case.

For example. I would never use linear phase for lows in a kick (pre-ringing, they do also smear the transients, in this case worse than minimal phase). And would never use minimal phase for parallel processing of the same signal (comb filtering due to phase shift).

[noiseresearch]

A implementation of the Linkwitz Riley filter for splitting as Max4Live device.
Includes an article about how it's build an some math stuff I don't get

https://www.metafunction.co.uk/post/the ... on-live-10

[rafa1981]

Please don't think that linear phase filters are better. They are just different. There is no perfect filter. There is no invisible crossover...

x1000. I couldn't agree more.

What you described is exactly how I did on MixMaxtrix, now there is only the linear phase one left to do.

[rafa1981]

... And would never use minimal phase for parallel processing of the same signal (comb filtering due to phase shift).

Notice that it is possible to recreate the allpass response of a given linkwitz Riley filter, so one sends the dry signal to the allpass chain and the wet signal to the crossover. I left that out of MixMaxtrix, as I found that each band's dry/wet would achieve the same when internal processing and it could probably be not obvious to everyone what that fx would be for.

Alternatively, a more bruteforce solution is to apply the same crossover with no processing to the dry signal.

[legendCNCD]

A implementation of the Linkwitz Riley filter for splitting as Max4Live device.
Includes an article about how it's build an some math stuff I don't get

https://www.metafunction.co.uk/post/the ... on-live-10

Thanks for this! I will surely try on next time when needing to split up stuff.

[midi_transmission]

Alternatively, a more bruteforce solution is to apply the same crossover with no processing to the dry signal.

Yes, thanks for this addition. I was thinking more general, that I would never use an eq on just one channel. But for the case x-over it's worth to mention.

[attention ♥]

DMG Multiplicity claims to have "bit transparent" crossovers. So until the signals are altered, it remains identical to the source audio.