Types of low-pass filters

Stemcell
#1

Hello.

I have a question about low-pass filters.
From the documentation, I understand the design philosophy of using Linkwitz-Riley filters to avoid dips and bumps.
However, even with this approach, wouldn’t a dip occur near the crossover point when adding the same source to the LFE?

This may not be much of an issue with traditional movie LFE sounds, but when working on Dolby Atmos Music, it can create a significant difference.

How are people typically handling low-pass filters for the LFE in practice?
Personally, I think having a linear phase option here would be great, but is it something others don’t find necessary?

#2

You mean you have the same signal being sent to Main channels and LFE?
You should never do that.

Music or Post, all the same theory applies. the LFE signal should be something unique, or if you must redirect LF signal to the LFE from the mains, it should use a flat crossover filter type like linkwitz-Riley.

The crossover-style filters are used in REDIRECT mode only.

Linear phase filters designed to work well for low frequencies would require a massive latency value, which would make realtime mixing extremely awkward (faders out of sync). So those filters are really not an option for filtering LF.

#3

Thank you for your response.

As you mentioned, using a linear phase EQ for low-pass filtering does indeed introduce significant latency, which can interfere with mixing.

However, filters in applications like the Dolby Atmos Album Assembler, used for Dolby Atmos Music, employ linear phase processing.

In genres like Atmos Music, sending the same source to the LFE is not uncommon.

So, having a linear phase mode on the low-pass filter in Stemcell would be a welcome addition.

What do you think about this idea?

#4

That’s just a bad idea. I don’t even recommend Stemcell’s Redirect feature for this purpose - it’s intended for the monitor chain. Film/TV industry has worked this out over decades and it’s advise worth heeding.

If you want to add LFE, use something like Subquake to generate entirely new, decorrelated sub signal, keyed of whichever instrument needs the extra impact.

We cannot add a linear phase mode as that would increase the latency massively. Also, we will never add a mode which changes the latency at runtime - this causes nasty snaps and pops whenever the delay compensation engine has to recalculate. There are popular plugin on the market which do this and I’ve seen it waste hours of expensive mix-stage time chasing ghost-clicks.

I’d like to know if you can actually hear a qualitative improvement using a linear phase LF filter in a real world setting. It will sound different, for sure - but given all the phase variable in the speaker and sub placement, the room acoustics and everything else, I don’t think linear phase is necessarily a better option.

And it’s a PITA.

#5

not to mention the massive pre-ring smearing which you would hear on the entire mix…