M/S Technique?
M/S Technique?
I cant find a basic explaination of MS anywhere. Im having a really hard time wrapping my head around why it works. From reading a few web pages i think they are saying the mid channel is left alot, the side channel is duplicated, panned left and right and one side is inverted. why does this work? when inverted left and right will cancel each other out right? how does mid make this any better? can someone explain this in a simplified manner please.
-
Stephen B.
- gettin' sounds
- Posts: 122
- Joined: Wed Dec 01, 2004 11:39 am
[deleted because it was wrong]
Last edited by Stephen B. on Tue Dec 20, 2005 4:16 pm, edited 1 time in total.
"Badness is only spoiled goodness."
C.S. Lewis
C.S. Lewis
in practice, you would decode the "S" mic like so:
1]have S mic coming into a channel strip. lets say channel 2
2]patch Ch2 into ch3 via post aux or direct out
3]flip the phase of 3
4]pan 2&3 hard left(or right)
5]do all your assignments and whatnot, so that your signals will wind up in the L/R mix
6]bring 2 to -U-
7]bring up 3 until maximum cancellation occurs(will be total on good digital gear). Once that is done, DONT moce the ch 3 fader again or adjust the aux or whatever you used to send it to ch3 from ch2
8]pan 3 hard right, 2hardleft
9]bring up the "M" mic and mix to taste. Adjust sides using ch 2 fader only.
1]have S mic coming into a channel strip. lets say channel 2
2]patch Ch2 into ch3 via post aux or direct out
3]flip the phase of 3
4]pan 2&3 hard left(or right)
5]do all your assignments and whatnot, so that your signals will wind up in the L/R mix
6]bring 2 to -U-
7]bring up 3 until maximum cancellation occurs(will be total on good digital gear). Once that is done, DONT moce the ch 3 fader again or adjust the aux or whatever you used to send it to ch3 from ch2
8]pan 3 hard right, 2hardleft
9]bring up the "M" mic and mix to taste. Adjust sides using ch 2 fader only.
01010100 01100001 01101011 01100101 00100000 01001101 01100101 00100000 01110100 01101111 00100000 01111001 01101111 01110101 01110010 00100000 01101100 01100101 01100001 01100100 01100101 01110010 00100001
this doesnt make sense to me because its a mono signal from the S mic and no matter if the lobes are inverted of each other or not, its one mono track and when u invert it 180 degrees its gonna flip EVERYTHING and if you have one inverted and one not and put the 2 close to each other its gonna be complete cancelation, no left or right lobes left. is that wrong? this is why im wondering exactly why it works. what does the M mic do to bring the left and right out in an otherwise empty recording? i just need like a newb explaination.Stephen B. wrote: When the side mic is duplicated, panned and inverted, the "right lobe" of the left channel and the "left lobe" of the right channel are cancelled out. This leaves only the "left lobe" in the left channel and the "right lobe" in the right channel. Now these are 180 degrees out of phase from one another, so when this recording is collapsed to mono, the side mic is completely eliminated from the action, leaving only the mid mic.
r0t4ry wrote:this doesnt make sense to me because its a mono signal from the S mic and no matter if the lobes are inverted of each other or not, its one mono track and when u invert it 180 degrees its gonna flip EVERYTHING and if you have one inverted and one not and put the 2 close to each other its gonna be complete cancelation, no left or right lobes left. is that wrong? this is why im wondering exactly why it works. what does the M mic do to bring the left and right out in an otherwise empty recording? i just need like a newb explaination.Stephen B. wrote: When the side mic is duplicated, panned and inverted, the "right lobe" of the left channel and the "left lobe" of the right channel are cancelled out. This leaves only the "left lobe" in the left channel and the "right lobe" in the right channel. Now these are 180 degrees out of phase from one another, so when this recording is collapsed to mono, the side mic is completely eliminated from the action, leaving only the mid mic.
you're right that the lobes are inverted, and that it will flip it 180 degrees. however, since the opposite sides of the pattern are picking up different things, you won't see complete cancellation.
what you then do after this is pan the two split sides, and that gives you your stereo field of width.
that devil bastard protools
-
Mark Alan Miller
- dead but not forgotten
- Posts: 2097
- Joined: Wed Apr 14, 2004 6:58 pm
- Location: Western MA
- Contact:
Imagine a compression (positive pressure of air) hitting the left lobe of the S mic. That compression will also hit the M mic. The two sum together (to some degree) on the left side of the image, but the phase inverted right side of the S mic will cancel (to some degree), providing localisation of the sound to the left. The opposite is true for a sound from the right hand side, where the right lobe of the S mic looks like a compression on the phase-inverted right channel when a compression hits it.
Without the M mic, the imaging doesn't happen. Also, the S mic does completely cancel if the matrix is set up correctly and then summed to mono, regardless of different things hitting the two lobes, it's still one signal. A M/S summed to mono only yeilds the M signal.
(I'm using only the compression half of the compression/rarefaction of actual sound to simplfy the example, with compression indicating a moment of positive-going air pressure and thus, voltage. It's easier to conceptualise if you think of a momentary part of the waveform this way... I think.
)
The matrixing is similar to how stereo LPs and FM radio work. L+R and L-R, or sum-and-difference signals.
Without the M mic, the imaging doesn't happen. Also, the S mic does completely cancel if the matrix is set up correctly and then summed to mono, regardless of different things hitting the two lobes, it's still one signal. A M/S summed to mono only yeilds the M signal.
(I'm using only the compression half of the compression/rarefaction of actual sound to simplfy the example, with compression indicating a moment of positive-going air pressure and thus, voltage. It's easier to conceptualise if you think of a momentary part of the waveform this way... I think.
)The matrixing is similar to how stereo LPs and FM radio work. L+R and L-R, or sum-and-difference signals.
he took a duck in the face at two and hundred fifty knots.
http://www.radio-valkyrie.com/ao/aoindex.htm - download the new record (free is an option!) or get it on CD.
http://www.radio-valkyrie.com/ao/aoindex.htm - download the new record (free is an option!) or get it on CD.
-
Professor
- ghost haunting audio students
- Posts: 3307
- Joined: Wed May 07, 2003 2:11 pm
- Location: I have arrived... but where the hell am I?
Um, I don't want to seem mean or anything (though for some reason I'm up much earlier than I normally like to be today and may be kinda cranky) but those explanations are a little off.
Rotary, you're absolutely right that you can't split the lobes from a fig-8 mic. The mic captures an image of the world around it and produces a single AC voltage at its output, and there ain't nothing you can do to split the sounds from the front vs. the back any more than you can split the front and back of a cardioid or omni mic. It is true that the sounds arriving at the rear of the fig-8 mic are captured 'out of phase' for the microphone, but that simply means that a sound pressure waves 'pushing' at the rear of the mic generates a negative voltage, while a pressure wave 'pushing' at the front generates a positive one (and reverse that for 'pulling'. The signal from that mic is still just a single AC voltage.
But, this is a technique that involves TWO microphones which are capturing sound at the same point in space but doing so in different ways.
First of all, let's skip the 'how to do it' because that's covered everywhere and you need to know 'why it works'. Let's also skip the flip and pan stuff with the second copy of the fig 8 and just focus on one mid and one side.
I suggest drawing youself a picture: with a circle, a line up and down for 0? to 180?, and a line across for 270? to 90? (right = 90). And then draw in a cardioid mic pattern. Roughly, it doesn't need to be perfect.
Then draw another picture with the same circle and this time a fig-8 mic BUT REMEMBER that the fig-8 is pointing to the "side" so the 'front' of the mic should be at 270? and the 'back' at 90?. Oh, and put a big + in the left side of the fig-8 pattern and a - in the right, assuming that's how you like to set up your side mic?
Now these two microphones are occupying the same point in space (in theory) and sound arrives at both mics at the same time, but at different volume levels from different directions. A sound arriving from 0? pushes and pulls the Mid mic at full volume, but is ignored by the side. A sound arriving from 270? pushes and pull the Side mic at full volume, but is down 6dB in the Mid mic. A sound from 180? is ignored by both mics. And a sound from 90? is pushing and pulling the Side mic at full volume, but the voltage generated is "wrong" and is interpreted as a "push" when it should be a "pull" and vice versa, though it still arrives at the Mid mic at -6dB.
Unfortunately, deciBels are a funny kind of scale because if we add two equal sounds together they go up by 3dB, so it makes wrapping your head around the math a little tricky.
But OK, if we stick the Mid and Side together, then at 0? we get only the Mid, and at 270? we get mostly the Side (the mid is -6dB which is half volume, then half again). At about 315? (45 off from 0) the two mics are down only slightly, and add together to deliver more than we have at 0 or 270. At 180, everyone's out. At 90? it looks like we should have the same as 270 except that this side is backwards as far as voltage is concerned and so it cancels the signal from the Mid. So, at 90? we have the lower Mid signal being cancelled by the Side, and at roughly 45? they are almost equal, but opposite and cancel each other.
If we were to very carefully add the numbers at each degree around the circle, we would draw a very neat little Hypercardioid mic turned almost 45? to the left (centered on about 315).
Now you can do the exact same math only "flip" the fig-8 mic so that the positive is on the right side and the negative is on the left. Here we are simply adjusting for the physical flaw so that a push on the rear now generates a positive voltage and pull generates a negative voltage. If we do the same math for the two mics, we still get the exact same Hypercardioid pattern, and it is still angled 45? off center, but this time it is to the right, and not the left.
That's it. That's all we needed to do to generate a 'right' and a 'left' from our Mid and our Side tracks, and if we route those to the left and right speakers, we'll get a nice stereo image.
And you can decode with three channels by sending your Mid to a channel panned center (equal to both L&R but -3dB to each), and then splitting your side to two channels (and losing 3dB in the process) and panning those hard left and right with the right channel inverted. OR you could decode using four channels (or in a DAW) by splitting your Mid into 2 tracks routed L and R, and splitting your side into 2 tracks routed L and R while flipping the right.
But the real fun is in manipulating the levels of Mid vs. Side in the mix. Because as you increase the Mid volume, it alters the math relationship and both changes the resulting pattern, and the angle of the L to R decoded pair. More Mid means a more cardioid shape (wider than Hyper) with a narrower angle. Less Mid (more Side) means a narrower pattern (approaching fig-8) and a wider spread.
And then just to mess with your head some more, that mid microphone doesn't need to be a cardioid mic! It can be omni which results in a pair of 'wide cardioids' aimed 180? apart and as you adjust the levels, the pattern widens and narrows, but the angle remains a constant 180?... crazy! And if you use a Fig-8 in the middle, your result is a pair of fig-8s angled 90? to each other, and 45? off axis from the Mid. Iit literally just turns the pattern 45?... but if you change which mic is Mid vs. Side, or flip the phase on both mics in the right ways, you can have a Blumlein pair (crossed fig 8s) pointing in any direction in 45? steps all the way around in a circle. Oh, and if you adjust the Mid vs. Side volume this time, you change the angle of the resulting pair, but the pattern stays a consistant Fig-8.
Hopefully that helps you to wrap your brain around the concept a little more, and from there you can just start experimenting. But it is always good to understand the "whys" just as much as the "hows".
-Jeremy
Rotary, you're absolutely right that you can't split the lobes from a fig-8 mic. The mic captures an image of the world around it and produces a single AC voltage at its output, and there ain't nothing you can do to split the sounds from the front vs. the back any more than you can split the front and back of a cardioid or omni mic. It is true that the sounds arriving at the rear of the fig-8 mic are captured 'out of phase' for the microphone, but that simply means that a sound pressure waves 'pushing' at the rear of the mic generates a negative voltage, while a pressure wave 'pushing' at the front generates a positive one (and reverse that for 'pulling'. The signal from that mic is still just a single AC voltage.
But, this is a technique that involves TWO microphones which are capturing sound at the same point in space but doing so in different ways.
First of all, let's skip the 'how to do it' because that's covered everywhere and you need to know 'why it works'. Let's also skip the flip and pan stuff with the second copy of the fig 8 and just focus on one mid and one side.
I suggest drawing youself a picture: with a circle, a line up and down for 0? to 180?, and a line across for 270? to 90? (right = 90). And then draw in a cardioid mic pattern. Roughly, it doesn't need to be perfect.
Then draw another picture with the same circle and this time a fig-8 mic BUT REMEMBER that the fig-8 is pointing to the "side" so the 'front' of the mic should be at 270? and the 'back' at 90?. Oh, and put a big + in the left side of the fig-8 pattern and a - in the right, assuming that's how you like to set up your side mic?
Now these two microphones are occupying the same point in space (in theory) and sound arrives at both mics at the same time, but at different volume levels from different directions. A sound arriving from 0? pushes and pulls the Mid mic at full volume, but is ignored by the side. A sound arriving from 270? pushes and pull the Side mic at full volume, but is down 6dB in the Mid mic. A sound from 180? is ignored by both mics. And a sound from 90? is pushing and pulling the Side mic at full volume, but the voltage generated is "wrong" and is interpreted as a "push" when it should be a "pull" and vice versa, though it still arrives at the Mid mic at -6dB.
Unfortunately, deciBels are a funny kind of scale because if we add two equal sounds together they go up by 3dB, so it makes wrapping your head around the math a little tricky.
But OK, if we stick the Mid and Side together, then at 0? we get only the Mid, and at 270? we get mostly the Side (the mid is -6dB which is half volume, then half again). At about 315? (45 off from 0) the two mics are down only slightly, and add together to deliver more than we have at 0 or 270. At 180, everyone's out. At 90? it looks like we should have the same as 270 except that this side is backwards as far as voltage is concerned and so it cancels the signal from the Mid. So, at 90? we have the lower Mid signal being cancelled by the Side, and at roughly 45? they are almost equal, but opposite and cancel each other.
If we were to very carefully add the numbers at each degree around the circle, we would draw a very neat little Hypercardioid mic turned almost 45? to the left (centered on about 315).
Now you can do the exact same math only "flip" the fig-8 mic so that the positive is on the right side and the negative is on the left. Here we are simply adjusting for the physical flaw so that a push on the rear now generates a positive voltage and pull generates a negative voltage. If we do the same math for the two mics, we still get the exact same Hypercardioid pattern, and it is still angled 45? off center, but this time it is to the right, and not the left.
That's it. That's all we needed to do to generate a 'right' and a 'left' from our Mid and our Side tracks, and if we route those to the left and right speakers, we'll get a nice stereo image.
And you can decode with three channels by sending your Mid to a channel panned center (equal to both L&R but -3dB to each), and then splitting your side to two channels (and losing 3dB in the process) and panning those hard left and right with the right channel inverted. OR you could decode using four channels (or in a DAW) by splitting your Mid into 2 tracks routed L and R, and splitting your side into 2 tracks routed L and R while flipping the right.
But the real fun is in manipulating the levels of Mid vs. Side in the mix. Because as you increase the Mid volume, it alters the math relationship and both changes the resulting pattern, and the angle of the L to R decoded pair. More Mid means a more cardioid shape (wider than Hyper) with a narrower angle. Less Mid (more Side) means a narrower pattern (approaching fig-8) and a wider spread.
And then just to mess with your head some more, that mid microphone doesn't need to be a cardioid mic! It can be omni which results in a pair of 'wide cardioids' aimed 180? apart and as you adjust the levels, the pattern widens and narrows, but the angle remains a constant 180?... crazy! And if you use a Fig-8 in the middle, your result is a pair of fig-8s angled 90? to each other, and 45? off axis from the Mid. Iit literally just turns the pattern 45?... but if you change which mic is Mid vs. Side, or flip the phase on both mics in the right ways, you can have a Blumlein pair (crossed fig 8s) pointing in any direction in 45? steps all the way around in a circle. Oh, and if you adjust the Mid vs. Side volume this time, you change the angle of the resulting pair, but the pattern stays a consistant Fig-8.
Hopefully that helps you to wrap your brain around the concept a little more, and from there you can just start experimenting. But it is always good to understand the "whys" just as much as the "hows".
-Jeremy
-
Mark Alan Miller
- dead but not forgotten
- Posts: 2097
- Joined: Wed Apr 14, 2004 6:58 pm
- Location: Western MA
- Contact:
Professor- super, as always.
However, I don't believe I was off in my explaination, but over-simplified and lacking detail. I just don't have the typing stamina (or the time) that you have. Wow.
However, I don't believe I was off in my explaination, but over-simplified and lacking detail. I just don't have the typing stamina (or the time) that you have. Wow.
he took a duck in the face at two and hundred fifty knots.
http://www.radio-valkyrie.com/ao/aoindex.htm - download the new record (free is an option!) or get it on CD.
http://www.radio-valkyrie.com/ao/aoindex.htm - download the new record (free is an option!) or get it on CD.
-
Professor
- ghost haunting audio students
- Posts: 3307
- Joined: Wed May 07, 2003 2:11 pm
- Location: I have arrived... but where the hell am I?
Indeed, you were pretty accurate and I was referring to the other posts more. And I was kinda going back and forth between reading and cooking breakfast. No intention to put anyone down.Mark Alan Miller wrote:I don't believe I was off in my explaination, but over-simplified and lacking detail. I just don't have the typing stamina (or the time) that you have. Wow.
-J
-
Professor
- ghost haunting audio students
- Posts: 3307
- Joined: Wed May 07, 2003 2:11 pm
- Location: I have arrived... but where the hell am I?
Lots of guys are afraid of commitment and who can blame them. I always thought of the MS technique as being useful in a fixed mic position setup for a varying musical performance, like recording a live variety show, and also for making the most out of a limited number of tracks. It's also handy in broadcast situations. If you only have 2-tracks on a location recorder or 8-tracks in a studio, it can be really useful to have a mic array that is more adjustable in mixdown.mjau wrote:Mid-Side is great for someone like me who doesn't like to commit to anything while I'm tracking. It leaves me with a lot of possibilities later.
-Jeremy
Well, the thing is, there's tons of phase issues, yet it still works in mono because of them.r0t4ry wrote:lol, prof thats so awesome, thats the exact explaination i was looking for. you explain stuff so well! thanks alot man! why dont more people use this? it seems very versatile and without phase issues right? i gotta get me a figure 8 mic
But I'm scared, man, I'm scared.Have I taught you nothing? Commit my friend commit.
Who is online
Users browsing this forum: No registered users and 57 guests