Hi,
I'm trying to figure out how to convert stereo track to M/S, if that's even possible.
Various guides found on internet use L+R as mid and L-R as side. This is not quite correct procedure.
Side IS L-R, but mid (center channel) is NOT L+R.
Mid should be something like (L+R)-(L-R), or not?

There are only handful of tools which can extract center channel, and it's never lossless, I wonder why is that?
Anyway, does anyone know how to do it?

 

You can use Voxengo's MSED.
I'm not sure what you mean with "lossless". There is no center channel in a stereo signal. You would have to encode mid/side. Different algorithms lead to (slight) different results, but basically it's M=(L+R)/2 and S=(L–R)/2.

 

Have you looked into using Waves S1 MS Matrix?

 

Bitwig has a tool (Mid-Side Split Container) to split the signal...but you don't use it, of course.

 

There's also iZotope RX3's Extract Center in Channel Operations.
I thought this could be done with mathematical precision, just as cancelling the center.

 

No, mid signal is L+R.
Think about it like this:
You have some random song which is a left and a right channel. Those individual channels are both mono.
Now you take those channels and convert them into a single mono file. That file is your mid channel, but it lost some parts of the signal because of phase cancellation. This "lost" part is the side signal.
To get the side signal you take the one side of L/R and subtract it from the other. The simplest way of doing that is to do the exact same you did with the mid signal, just swap the phase of one signal by 180° before you add them together.
Now you got your two signals, mid and side and you can edit them separately.
To get your original file back you need to reverse the process. (M+S = left and M-S = right)

I'm curious where you found this information.

 

Center by Waves plug-in
Not shure if its really M/S on the high end
But its defenitly put the low frequencies to Mono
3 parameters knob (Low-High-Punch) 2 Faders ( High-Low)

 

Have you checked with this

http://www.brainworx-music.de/en/plugins/bx_control_v2

It can't be lossless if just m or s is playing, because information has been removed.

Check that tool on a mono source with a spectrogram and see what you find, that's the most accurate tool for sound.

 

I've also heard that encoding and decoding in digital has some issues, if I remember correctly, the phase can be affected, But I've never experienced it. Maybe it happens if you try to encode and decode manually (ie with duplicates and phase inversion), but I've always used Pro-Q 2 and Brainworx plugins for MS.

Lossy/lossless aren't the right terms for the issues the OP is talking about, and anyways information isn't being removed, rather its being separated.

 

I think Andrew is talking about the CORRECT EQUATION and COEFFICIENTs, not other crap.

Here it is:

M=0.5*(L+R)
S=0.5*(L-R)

Then you can go

L=M+S &
R=M-S

Regarding lossless, most audio operations may be lossless depending on the datatype and range of the numbers.

"There are only handful of tools which can extract center channel, and it's never lossless, I wonder why is that?" < It may not be lossless due to the complex maths behind it (FFT). For FFT CCX to work, you NEED (typically), a few thousand vaariables, you generate a few thousand more, and you output a few thousand, and it's not a simple transformation on the cursory like +-*/. This is in contrast with MS, when you input 2, generate 4 (at least), and output 2 with simple +-*/ maths on the cursory.

If you're referring to the "audiable FFT artefacts", it is due to the way that FFT works and the parameters & additional DSP that's done in the FX implementation.

 

----

 

Last edited: Mar 23, 2016

To be more specific you must multiply one channel by -1, it's not the same thing, i. e. http://www.churchsoundcheck.com/are-you-confusing-polarity-with-phase.html .
You should add different effects to M and S for obtaining something that is not simple panning.

 

Yep you simply need an encoder and then also a decoder to utilize mid/side or sum/difference processing on any stereo track, The plug-in's already mentioned do a great job (Brainworx bx_solo, Voxengo MSED...etc.) It's pretty simple to setup/do without any plugins but it is alot faster to drop a plugin onto a track than it is to create a MS matrix via three channels set up right to get the job done. In fact most DAWs come with a built-in plugin tool or utility which enables quick and easy MS encoding and decoding these days.

If you use microphones to capture your source material then it is a great alternative way to capture signals that you'd use a stereo pair to capture:

http://www.uaudio.com/blog/mid-side-mic-recording/
http://ehomerecordingstudio.com/stereo-microphone-techniques/
http://www.recordingmag.com/videos/videoDetail/18.html

Some decent stuff on the topic instead of just repeating bits from good articles:

http://www.soundonsound.com/sos/jul12/articles/studio-one-notes-0712.htm
https://www.soundonsound.com/sos/nov10/articles/stereoprocessing.htm
https://www.soundonsound.com/sos/apr12/articles/reaper-0412.htm
http://www.soundonsound.com/sos/jan14/articles/pro-tools-0114.htm
http://tapeop.com/reviews/gear/80/drms-v3-midside-processor-plug-in/
http://audiogeekzine.com/2012/01/3-mid-side-processing-tricks/
https://www.plugin-alliance.com/en/products/bx_solo.html
https://goodhertz.co/midside
http://www.waves.com/mid-side-eq-improve-stereo-image
http://www.about-audio-mastering-software.com/mid-side-vst-plugins.html
and just a quick google search and youtube search for mid side plugins, recording and processing will give you all you could ever need know and want to know about the topic.

One tool that I find extremely useful for certain occasions is this one by Brainworx as it is the only plugin which I am aware of that is able to create a full stereo signal from mono source material and preserve it so that it doesn't sound horrific when folded back down to mono/summed: https://www.plugin-alliance.com/en/products/bx_stereomaker.html
If anyone knows of others which can do that I'd really like to hear about them

There are plenty of plugins now with a built in M/S encoder and decoder allowing for instant processing and the more the better in my very humble honest opinion.

Be sure to try using it on stereo busses/auxes/returns especially say equalization and dynamics processing after for example a reverb (but it can give awesome results after most effects processing). Reason being it gives you alot more control over the results (especially when you factor in automation so you can thin out and widen the reverb in more dense areas of a track such as a chorus section but keep the reverb more prominent whilst on verse sections where a track is often more sparse). I would be happy to see Mid/Side processing built-in to every plugin, Especially modulation, delay and reverb plugins.

Have fun with it and all the best to all as always

Dean

Edit, I would highly recommend these bits for the more technical parts, Especially the masters thesis which is linked to in the Variety Of Sound Blog piece, It's in PDF format and I've put it a link below for speeds sake:

https://varietyofsound.wordpress.com/2011/09/25/the-mid-side-technique/
http://www.midside.com/pdf/nyu/masters_thesis.pdf
http://sepulchra.com/blog/?p=2061

(Courtesy of Dan Worrel, A great guy and an endless supply of knowledge. An engineer I admire alot and am proud to know)
http://www.masteringmansion.com/common/docs_d/doc_1500.pdf
http://www.aes.org/technical/documents/AESTD1003V1.pdf
https://www.ntnu.edu/documents/1001...on_9.pdf/471420df-fe0c-4a78-86e9-05710684cc69
http://www.rs-met.com/tutorials.html
Without buying white papers or text books I've not got anything else off the top of my head.

I hope it helps or is of some use and honestly I wouldn't sweat it too much. We all have to trust our ear's along with utilizing good metering tools as someone mentioned, If it sounds good then it
is good as the saying goes, I like it and stick to it but YMMV of course

 

Last edited: Mar 23, 2016

I bet it takes years to learn all this mathematical stuff. I'm envious. All I do in Ableton is stick a (mono) utility on the channel but I have no idea how it does it.

 

Don't be lazy! It's elementary mathematics! only sums and multiplications like when you buy something in a shop and you must calculate the change

 

Lazy? I can take you from anywhere in London, to any other point in London, in the shortest time using the quickest route without the aid of a map or sat nav device. I can also name every road or street, cinema, hotel, public house, nightclub, theatre along the route that I have used. Should I refer to you as lazy because you cannot do that?
My point is, I have never been taught it, and not because I wasn't interested. When I left school I tried to get on a Sound Engineering Course, but I couldn't afford £5,000. So I became A London Taxi Driver (Licensed Black Cab, not uber or mini cab). Next time you accuse someone of being lazy? Get your facts straight first.

 

It really is that simple, but it'll take you about half an hour to figure it out. Much shorter than memorizing the london streets.

From a maths point of view, it is precise because you deal with lots of decimal numbers & angles. From a human POV, not very cuz artefacts. As olaf said, there is no 'center' channel, and it cannot be derived from simple maths.

But what is the center? There can be 2 definitions depening on whether you're upscaling to 3ch or 5ch, but both have in common "the part of the sterro signal that is panned center". This also means that a sound can be in L' as well as C', but not in R' if it's panned half-left. This requires audio analysis.


Take a look here: https://github.com/extemporelang/kiss_fft/blob/master/kiss_fft.c < does this look simple? FFT is one of the things you need to do to make a center channel.

WHat is FFT?

FFT is the frequency decomposition (or transformation ) of a time-doman signal, such as audio/music.



img src: http://www.eetimes.com/author.asp?section_id=36&doc_id=1325667

See also https://en.wikipedia.org/wiki/Fast_Fourier_transform

HOW is this relevant? You FFT the signal to get it in the Freq Domain. You calculate/analyze the pan of all the different frequencies. FROM the pan, you can make a center (knock out L/R, whatever). And then iFFT to get an audiable signal back.

I've also got some code for extracting the center via phase (NOT PAN), but it's imperfect. Analysis L C R parts about ~50 lines. EXCLUDES the ACTUAL FFT and antiartefacting bits.

Waves Center May not be a true center channel extractor because I've heard from someone that it uses fancy enveloping (instead of FFT). Not surprising, it has the worst isolation of all the FFT stuffs I've tested. It does

• EQ
• Transient designing
• Center channel manipultion
• With 48sa PDC only

Also not surprising (at least from a phase persepective), it really screws with the phase and is not the original with null change at the output. QuikQuak mashtactic (FFT imager) does all of the above with a more beleiveable 4096sa PDC, and AOM Stereo Imager D is a Center CHannel manipulator ONLY (which is what I suggest if you want a good, plain-janes VST). Both of the latter are linear-ish phase.

If you use IIR EQs, yes.

If you use FIR EQs, no. (FIR = linear phase). Screw ringing. Increase your FFT size or deal with it (tm).

I like Ozone 5 (or ozone 4's) EQ for this. I don't like Fabfilter stuff 'cuz it's overhyped and underperforms.

"separated" aka transformed
---

I invite all interested readers to read these 2 articles written by me:

M/S is not FFT
https://docs.google.com/document/d/1rGFjd8bc3WmuUsEQG640qH6Y7l_jzPJF1wi4bkGomVk/edit#

Evaluation of different FFT-based imagers
https://docs.google.com/document/d/1ifJ7GHN0yBvrGwuaNnJT6LWYxjGsNJHfHReEhSxK3Q4/edit#

 

Last edited: Mar 23, 2016

Cheers junh1024, I will definately give these a read and You taking the time to putting them together is appreciated.
I always enjoy getting some fresh stuff to get immersed it as it switches up on whatever area/areas that I'm reading into
99% of the time.

All the best your way and to all as always

Dean

 

I didn't want to be offensive... I remember the times I didn't want to investigate M S believing that the math behind that were difficult..

 

Last edited: Mar 23, 2016

I'm sorry too Angie. Normally I would have laughed it off, you just caught me on a bad day.