Mono downmixing

Hi all, thanks in advance for any wisdom.

What processing would you think to do for converting a stereo mix to a mono?

Ideally you would want to be able to remix the thing and remove any phasey effects like stereoization, stereo reverbs, etc. but sometimes the world is not ideal… what could be done post-hoc to minimize the effects of interference when you downmix? Low cut one channel? Invert? Some clever spectral phase alignment (looking at you, Sound Radix)?

I’m thinking also maybe a post-process to match the loudness and peak level before and after, since the increase in level will depend on the material…

Any thoughts welcome.

Are you looking for a fancy solution that deals with mixes that have mono-compatibility issues?

A proper mix should translate well enough when simply summed to mono, which shouldn’t come as a surprise.

I am relatively certain that there is no generic “correct” approach that goes beyond summing L+R or discarding one of the channels. But if there are any, I’d be interested in the approaches.

Can you give a concrete example for a “problematic” mix that you want to turn into mono? Are there any assumptions you can make about your input material?

Yes exactly - a fancy solution that tries to compensate for mono compatibility issues indeed.

I work in TV/film and it happens that we are given a stereo mix of a TV show (produced from a 5.1) without access to the original mix and asked to downmix it to mono. On doing so, we can get phase cancellation in the lower frequencies and stacking which kicks the levels out of broadcast spec. It is also unpredictable how much the loudness will change since you do not know how correlated the left and right are.

I’ll try to put together an example without breaking any NDAs.

The only way to avoid mono compatibility problems entirely would be to just take one channel and discard the other. Most tracks these days tend to be “Big Mono” anyways where sounds are stereo but equally loud on both sides

Ahh just saw your reply that it’s for TV/Film. I’d say discarding the low end of one of the channels would work well. Would still take some manual picking of which channel to discard depending on the current sound

Interesting. I always thought, with the strict rules that tv broadcast enforces on deliverables, these things would not happen :smiley: Years ago, an ad soundtrack I made got rejected because I missed the EBU R128 LUFS target by 0.1dB.

Too bad you don’t have the original 6 channels of the 5.1 mix…

If you can constrain the issue to low frequencies, that is helpful. I’d try the following, maybe you can prototype it in any modular system you have available:

Split the signal at an appropriate low crossover frequency. If available, use a linear phase filter.
Sum the upper frequency range normally (L+R)/2.
Discard either the left or the right channel of the low frequency range.
Sum both results.

Depending on the crossover filter slope, you will get a smooth transition from stereo > crossover to mono < crossover.
The level will not rise over the original level with this solution, so this should be safe to use. It will only work if the issues you observed are in low enough frequencies, so it is safe to mono them, since there should not be any meaningful stereo information in the bass frequencies anyway.
If we’re talking about issues up to 300Hz or sth, then this probably won’t cut it…

I’d like to try some stuff on an example :slight_smile:

Adding to this: ozone’s stereo imager has a really good perfect reconstruction crossover filter

1 Like

Thanks guys! I’ll test this out. And I can ask about a short film mix that’s not under NDA…

“missed the EBU R128 LUFS target by 0.1dB.” - this is the point indeed, you set your true peak to meet spec and you need to ensure it meets spec in all formats… but also that the mix sounds true to the 5.1/atmos.

Tried a couple of things, the idea of throwing away the lows in one channel sounded good in general, but try telling a mixer you’re going to throw away any part of their mix…

Waves S1 imager narrowing to mono sounded good. Does anyone have any insight into what it’s doing exactly?

from the screenshot it looks like an ordinary M/S mixer, plus a rotation feature, which I personally find very useful in many cases. Doesn’t look like it’s doing any crazy magic. If you want to see what it’s doing, look at a vectorscope, maybe while running some artificial test signals through it.

If your mix engineer doesn’t want you to destructively fix things, tell them to deliver a mono compatible mix :wink: