converting AB to MS

[FALKEN]

I read in a book that any AB signal can be converted to MS, and vice-a-versa. in the same book, however, it said that for MS to work properly, the S mic has to be figure-8 pattern. these two statements seem to be contradictory....

If I recorded a drumset using spaced omnis, is it possible to convert to MS? I guess I could just load up the tracks and find out, but I am travelling, and this is boggling my mind.

 

[baekgaard]

I read in a book that any AB signal can be converted to MS, and vice-a-versa. in the same book, however, it said that for MS to work properly, the S mic has to be figure-8 pattern. these two statements seem to be contradictory....

If I recorded a drumset using spaced omnis, is it possible to convert to MS? I guess I could just load up the tracks and find out, but I am travelling, and this is boggling my mind.

A stereo signal can be 1) recorded in many different ways, and 2) encoded in at least several common ways. How it is recorded and how it is finally encoded is not the same thing, although one way of recording usually natively results in one kind of encoding.

Recording is often done in AB, spaced pair, ORTF, ... you name it, using at least two (or maybe even more mics), panning them to the sides.

It is also possible to record with a figure-of-8 and a cardioid microphone in such a way that the figure-of-8 captures the sides (the S part) and the cardioid captures the center part (the M -- mid). This is what you refer to above.

Without any processing, an AB/spaced pair/ORTF/... will give you a left-right encoded signal.

Similarly, if you just record the output from a typical MS pair, you will end up with an MS encoded signal, i.e. a mid and a sides signal. The Joint Stereo encoding of MP3 files is also a MS encoded signal, and it improve the compression by applying different compressions to the mid and side signals, since most of the musical signal often are in the mid with less information in the side, which can then maybe be compressed even more.

You can convert between these as follows:

M = L + R
S = L - R

and hence

L = (M + S) / 2
R = (M - S) / 2

Thus, if you record with an AB pair, you can sum the signals and subtract the signals as above, and get an MS encoded signal. This can be useful for certain types of manipulation -- if you e.g. reduce the S signal and then convert back to LR, you effectively minimize the stereo spread. If you amplify the S signal, you can widen the stereo perspective.

The "other way", converting the MS signal to a normal LR signal, is is usually a required step when working with a figure-of-8/cardioid setup.


So, to sum up: If you want to natively capture (without any processing) a MS encoded signal, you need a figure-of-8 and "some other" mic. This is usually a cardioid, but can also be other patterns. If you do not have a figure-of-8, you may have luck replacing it with two cardioids (in addition to the center mic, so 3 mics in total), just phase reverted and summed to simulate a figure-of-8 mic.

BUT: If you're looking for a way to record a signal with a mid mic and a side mic, you cannot do this with an AB part and the just convert it to a MS encoded signal. You still don't get the benefits you may be looking for, such as having a well behaved non-phase cancelling mono signal.


-- Per.

 

[tmix]

What baekgaard said!

A very unsophisticated way of saying it is:

What makes the M/S configuration work well is the information the mics are picking up, and how (because different from each other) when added together there is excellent phase coherence.
When using a figure 8 placed properly for a m/s micing, the mic only picks up the side info and not the direct projection from the guitar or drums since the null point is aimed direct at the source.
When you use omnis there is so much shared material picked up from the same sources, switching phase and extreme panning will often give you some horrendous results (could be cool I guess).

I've heard that some have gotten somewhat ok results using cardiod mics instead of omnis or fig 8, but truly to get the full benefit I'd use the proper mic configuration.

 

[FALKEN]

okay! I think I get it! thanks.

one more question though - doesn't using omni's (spaced or otherwise) effectively eliminate all phase problems?

 

[tmix]

No.

Careful spacing in a x/y array or following the 3 to 1 rule of distance apart to distance to source helps in keeping phase issues to a minimum.
To my knowlege (admittedly not THAT great) omni's don't automatically preclude phase issues.

I hope someone else chimes in, I could be wrong on that issue.

Tom

 

[baekgaard]

To my knowlege (admittedly not THAT great) omni's don't automatically preclude phase issues.

I hope someone else chimes in, I could be wrong on that issue.

You're right on.

There are roughly two ways of achieving a stereo effect: 1) By different levels in the channels (i.e. what you often do with panning), and 2) by delaying the sound slightly in one channel versus the other.

The ear uses both of these hints when trying to determine the apparent direction of sound.

1) can be achieved by either using directional mics or by using a shielding of some kind. 2) is usually achieved by spatial separation of the mics.

When recording coincident stereo with a cardioid pair in X/Y configuration, where the mics are at the "same" position, you achieve only 1) above -- and have no phase problems. Same applies for MS recording, where the mics are also at the same position.

When recording a spaced pair of omnis, you achieve almost exclusively 2) above, but since there is a difference in their spatial position, you will get phase cancellation issues. If they are e.g. 1m away from each other, any sound from the side around 680 Hz will cancel out totally when listening in mono. From there on and up in frequencies, it just gets worse and worse with nulls and peaks...

Using a spaced pair of cardioids or any near-coincident techniques with any directional microphone usually gives you both 1) and 2) above, to varying degrees.

Recording a coincident pair of omnis is usually not that interesting (unless you really try to shield the mics to make them a bit directional), since you will get neither 1) nor 2) and thus no stereo effect at all -- it is just the same signal recorded twice...

You might find this link interesting:

http://www.soundonsound.com/sos/1997_articles/feb97/stereomiking.html

Rgds,


-- Per.

 

[FALKEN]

cool, thanks. you guys are really helpful. I think I ought to fire a couple more at you.

1. can you recommend a good figure-8 mic for MS? dynamic or condenser. I already have a bunch of different cardoid mics. I am not so hot on the cheapo mics either, I generally buy time-tested stuff in 200 or less range, but would be willing to go higher (up to 500).

2. whenever I see drums being mic'd in a studio in vidoes or something with an xy pair, they are almost always right above the kit, angled towards the edges of the kit. this seems counter-intuitive, because most people hear a kit from the front, not above. I see how it would be useful if there were other instruments in the room, to reduce bleed, but wouldn't it otherwise make more sense to put the pair out in front of the kit?

 

[Innovations]

cool, thanks. you guys are really helpful. I think I ought to fire a couple more at you.

1. can you recommend a good figure-8 mic for MS? dynamic or condenser. I already have a bunch of different cardoid mics. I am not so hot on the cheapo mics either, I generally buy time-tested stuff in 200 or less range, but would be willing to go higher (up to 500).

2. whenever I see drums being mic'd in a studio in vidoes or something with an xy pair, they are almost always right above the kit, angled towards the edges of the kit. this seems counter-intuitive, because most people hear a kit from the front, not above. I see how it would be useful if there were other instruments in the room, to reduce bleed, but wouldn't it otherwise make more sense to put the pair out in front of the kit?

1. All ribbon mics are figure-8 mics so you might look into that as it would also get you a new sound to add to your kit for other purposes. There are some things that record SO well with a ribbon.

2. If those mics were all that was being used then you would have a point, but usually the overheads are used in addition to a separate kick drum mic and a separate snare drum mic and maybe even separate mics on the toms. Thus the overheads are mostly there to pick up the cymbals and other percussion. Since the snare and tom are dead center in the kit and usually also in the mix getting a stereo image of them is unimportant. It is with the other stuff that you would like to have a sense that one cymbal is to the left and the other is to the right.

 

[FALKEN]

that's a good idea. only, how will a ribbon on the S mic and a condenser or dynamic on the M mic sound?? I think it would be inconsistent, no?