Mid side

[Standanddeliver]

Ive heard about mid side processing , can someone explain it

[mixtur.se]

It’s just another way of encoding a stereo signal. In stead of left and right you decode the signal to mid (what’s same in both L and R) and Side (the difference). This means you can now process the mid and sides independently. You can actually do this manually with clever routing and phase inversion but it’s rather complicated so it’s great when audio software provides this feature.
https://www.soundonsound.com/sound-advi ... ually-work

[billinder33]

Traditional stereo processing is two channels - left and right. As mixtur.se said, you can take a stereo signal and 'redo' those two channels into a mid channel (the left+right similarities) and the side channel (the left and right differences).

On a typical hardware unit or plugin with a mid-side option engaged, generally the mid will be managed with the left channel, and the side with the right channel. On a modern, graphical frequency manipulation plugin like Pro-Q or Spectre, you can assign a frequency point to either the mid or side.

In practice, I rarely find mid-side processing valuable on individual tracks. But it's highly valuable in 2-buss or mastering applications.

Most common use cases are if a track has weak, poorly defined low end, sometimes boosting the sub frequencies in the mid will help solidify and focus the lows. Or if a track has weak highs or poor stereo imaging, sometimes boosting high frequencies in the sides can make the track feel wider and 'livelier'.

You can also reduce the lows in the sides if there's excessive 'mud' due to too much difference between the left and right channels in the low frequencies, though I would use this particular technique gently and with great caution because you can really screw up a mix if you don't know what you're doing. For this particular problem, a tool like Tone Projects' Basslane might be a safer option than DIY'ing with an EQ. Regardless, always use linear phase when EQ'ing in mid-side, because it can create serious phase problems if you don't. Most dedicated bass mono'ing tools like Basslane have linear phase and oversampling by default.

My favorite mid-side technique is compressing a song's mids and sides differently. By using different attack and release settings on the mids and sides, you can create a sense of separation between the low and high frequency instruments. Typically slower attack and release on the mids, so the mids meter is bouncing along with the kick and bass. Faster attack and release on the sides, so the sides meter is dancing in time with the hats. Then you can boost the sides 1 or 2 db to give the track some extra width. This can really add some 'depth' and 'motion' to a lifeless track, but even a well-mixed track can benefit from some light MS compression.

Hope that helps.

[mjolnir]

a good way to learn is to experiment with Mid-Side EQ.
you can get some really interesting or even wacky results.
but if you need something subtle and yet effective, it can be exactly what you need sometimes for a submix.
but it's hard to do consistently and if you overdo it, you can spoil your entire panning and mix decisions even if it sounds "good"
just my wierd 2 cents on this topic.

[foosnark]

In practice, I rarely find mid-side processing valuable on individual tracks.

I use it a lot in a modular synthesis context. Two oscillators as mid and side (in different octaves, fifths/fourths or just slight detuning) is generally a lot more pleasant than hard-panning them. Or run a mono voice through rwo different filters or LPGs that respond slightly differently.

Or if you patch feedback loops with multiple chained processors, try getting your mid and side signals from two different points in the loop.

Generally speaking, with techniques like this if you keep the side signal quieter than the mid, you'll have good phase coherence.

My favorite mid-side technique is compressing a song's mids and sides differently.

Yeah, that works nicely. Also when saturation plugins or dynamic EQs have a mid/side option it can be nice, if you keep it fairly subtle.

[billinder33]

I use it a lot in a modular synthesis context. Two oscillators as mid and side (in different octaves, fifths/fourths or just slight detuning) is generally a lot more pleasant than hard-panning them.

I just tested this with Bitwig grid. Bounced a M/S to audio and the same version in dual mono. Really interesting. Not a subtle difference at all. Waveforms look completely different. The M/S version is more subdued, the dual mono version is more in-your-face.

Or if you patch feedback loops with multiple chained processors, try getting your mid and side signals from two different points in the loop.

Not sure I understand this one at all. I'm pretty good with configuring semi-modular software like PhasePlant and can do some basic things in grid, but full modular is a little out of my depth.

How do I set up the feedback loop? And how would I route different points in the loop to M and S?