Topic: minimize effect of shifted stereo image

[Gutbucket]

For some reason - I had the idea that M/S would be more forgiving.

Instead of using the same Mid/Side ratio for both sides as is typically done to maintain Left/Right symmetry, you can instead use a different M/S ratio for each side if you like.  That's going to produce both a different virtual microphone angle and a different virtual microphone pickup pattern between channels.  And will produce somewhat different results than using the same Mid/Side ratio for both sides, which maintains the identical virtual microphone angles and patterns and then adjusting the level of one channel with respect to the other.

Would have to try it to determine if it does what you want, and setting up things to allow adjusting each side independently may be more involved than it's worth.  But it's certainly doable.

[Gutbucket]

The crux seems to be that those discrepancies can be delicious at low frequencies while being confusing and even nasty sounding at midrange and upper-midrange frequencies. This is why some people (I don't know whether there are any on this board, but some people in Europe that I've heard of) record with coincident or closely-spaced pairs of directional microphones placed at the center, while mixing in the signals from a spaced pair of omnis, with "crossover" filtering so that the overall pickup comes increasingly from the spaced pair the farther down in frequency you go.

I began playing around with the combination of spaced omnis and coincident (or closely-spaced) pairs years ago with exactly that intent in mind - limiting the contribution of the spaced omnis to the lower part of the frequency range and crossing over to the center pair for the mid and higher frequency range.  But then found I rather liked some amount of those midrange and upper-midrange phase discrepancies, which in moderation didn't sound to me confused or nasty, but rather contributed to a sound which seemed more even more open, spacious, clear and natural.  Plus it was a lot simpler not to need to mess with crossing them over.  Granted the frequency responses of the microphones I use inherently do a bit of that on their own- the omnis are more sensitive at low frequencies and flatter through the upper-mids and highs, and the directional mics are less extended down low while sounding bit brighter though the mids and up top.  In addition, Ill often EQ each pair differently so they work best in combination. But that's always by ear rather than crossing-over more exactly and completely, and often more about getting a natural sound out of each pair prior to combining them,in addition to rolling off some bottom from the center pair so it doesn't muddy up the clean stereophonic bottom end the omnis are contributing- sort of the opposite concern from avoiding nasty sounding mid and upper-range discrepancies.  Just how I hear it and what has worked for me.

Cheers and regards to everyone still following all the arcane discussion in this thread!

[kuba e]

Life In Rewind, Gutbucket, it is very interesting idea of using different ratio M/S to each side to correct stereo.
I have not fig. 8, but i will try it when doing XY onstage. I will decode XY to M/S (i hope it is possible) and mix it back with different side ratios. I am curious how it will be compared - change level difference vs. change M/S side ratio.

Gutbucket, thank you for very nice explanation of polarity relationship on example with crossed bidirectionals. All is clear for me now.
It is inspiring your explanation of mixing spaced omni and center pair. It makes it easy to understand with theoretical basis from this site.

[Gutbucket]

Glad you found this useful. 

You'll have no problem converting an X/Y recording to Mid/Side and back again.  You can do that with any L/R stereo signal actually, it's just that doing so with a signal which is free of phase-differences like that produced by a coincident pair avoids potential comb filtering issues when converting back to L/R using a different ratio.  That's not to say one shouldn't try the M/S manipulations on recordings which do have phase variations, such as those made with near-spaced or spaced techniques.  Only that one should listen carefully to make sure the M/S correction isn't introducing more problems than it's solving.

To expand a bit more on this, that kind of L/R > M/S (do something) > L/R conversion is useful for other things too.  The do something part doesn't have to be a change of M/S ratio. It allows one to manipulate a two channel stereo signal in terms of "the stuff in the middle verses the stuff at the outside of the stereo image" instead of in terms of "Left versus Right".  So one can for instance, EQ the the "center" to bring out some needed vocal clarity, differently than the ambience and instrumentation which images more widely.  Which is why many EQ plugins now feature a built in Mid/Side mode which does the conversion (and back) for you. Or do things like spectral edit out ambient hall noises from the Side channel which are heard ambiently in both channels but not in the center.  Or whatever.

For some reason - I had the idea that M/S would be more forgiving.

Getting back to this ^, usually there is a certain range of ratios which just work and sound best.  Trying to adjust too far either way causes things to not work so well in stereo (unless one wants to extract just the Mono sum or the Side component for some purpose).  So Mid/Side recording (and post manipulation) is best thought of as providing a stereo optimization tweak rather than offering a huge range of potential adjustment.  Remember that the virtual microphone angle and the virtual pickup pattern remain inextricably linked.  Changing the M/S ratio changes both these things together.  One cannot change the virtual angle while keeping the virtual pickup pattern unaffected too, and vice-versa. 

The ultimate extension of Mid/Side which does offer infinite, independent adjustability is ambisonics, which is in many ways the ultimate coincident microphone setup.  In it's basic form, ambisonics extends Mid/Side by recording and manipulating 3 or 4 recorded channels instead of just 2 and allows one to point as many virtual microphones as one wishes in any direction one wishes, choosing the pickup pattern for each independently, as well as selecting any 1st order pickup pattern along the continuum from omni through figure 8.  Recording 3 channels allows for that adjustability around the entire horizontal plane.  Recording 4 channels allows that adjustability for the entire 3-dimensional spherical space around the microphone array, meaning you can adjust the vertical angle of the virtual mics as well.  The caveat is that all virtual microphones remain coincident, one cannot introduce near-spaced or spaced phase-difference stereo without using additional non-coincidently located mics.

Conceptually, and sometimes in practice, a basic 3-channel ambisonic (horizontal plane) microphone setup consists of a pair of bidirectional microphones crossed at right angles and an omni, all arranged coincidently.  It's like the Blumlein technique I described earlier combined with a coincident omni.  You may recall that all microphone pickup patterns can be thought of as the combination of omni and bidirectional components (pressure and pressure-differential components) combined in various ratios. Ambisonics does that literally, which may shed some light on how it allows one to choose whatever pickup pattern one wants after the recording has been made, by clever ratio combinations of the 3 recorded channels which represent omnidirectional pressure, left/right differential, and front/back differential as measured from a single point.  4-channel ambisonics adds a 3rd coincident bidirectional mic oriented in the Z-plane, pointing up/down, to also measure the vertical component, completing the mapping of the entire vector sphere surrounding that single point.

[kuba e]

I'm always learning something new on this site. I will read MS technique in detail and come back.

[kuba e]

Ambisonics is beautiful. I do not know wheter coincident configuration is used in audience (maybe blumlein), but it must be very useful for onstage coincident recording. We can choose mics pattern and angle mics to acoustic center in post, great.

In which cases do you use this technique? Are there any restrictions except coincident configuration?

[Gutbucket]

I have an ambisonic mic and find it most useful on stage or close to performers where the ability to to adjust the angle, elevation and pattern of the virtual microphones after the recording has been made is more advantageous than it would be from further back in the audience.  When recording from very close the sources of sound are more differentiated and emanating from a wider arc, both horizontally and vertically, so slight tweaks left/right, up/down, and so forth can make a bigger difference than when the sound is more distant and ambient.

The main drawback as I see with ambisonic recording, besides requiring 3 or 4 recorded channels for a single stereo microphone position, is that unless combining it with other mics, or a second ambisonic mic, a recording made with a single ambisonic mic is always going to be coincident only, and I find some spacing and good phase interaction are often important elements of my best sounding recordings.   That aspect becomes more important as the recording position is moved further away from the source.  In general, as the recording position is moved further away, the best way to compensate is by using more spacing and less angle between microphones.  With an ambisonic mic you can adjust the microphone angles to be anything you want, but the spacing is always coincident.

[kuba e]

Thank you for the explanation. It must be fun to play with ambisonics mics.

I understand, unfortunately mics spacing can not be simulated or modified in DAW. The same is valid for stereo correction with time difference. Mics spacing remain the same when we shift one of channel.  We can imagine time shift in DAW like we are moving listener closer to one playback speaker (neglecting level difference). The same imagination can be used for level difference.

[Gutbucket]

Are there any restrictions except coincident configuration?

The coincidence and what follows from that are the main thing.   It's best for optimally tweaking good sounding coincident stereo techniques, so it works excellently for Blumlein crossed bi-directionals and wider angled crossed hypercardioids or supercardioids.  Although one is able to derive as many virtual microphones as one likes after the fact from that single ambisonic mic position, for really top quality surround playback one is effectively limited to 4 channels (quadraphonics) at most, while retaining sufficient differentiation between the playback channels.  When trying to derive more than four channels for playback from the same coincident position there ends up being too much correlation between channels.  For playback of five or more channels, spacing between mics is required to sufficiently decorrelate the ambient pickup between channels.