naturerecordists
[Top] [All Lists]

Re: [Nature Recordists] X-Y -> M-S & Perfect Reversals

Subject: Re: [Nature Recordists] X-Y -> M-S & Perfect Reversals
From: Rob Danielson on Monday, March 12, 2007 6:10 PM
.................. I made a test with +matrix-- code written by Tom Erbe.

http://tinyurl.com/yo5duf

Maybe someone can figure out something I'm doing wrong, or maybe
someone will find ways to do a "perfect reversal" when non default
plug settings are used. I hope so, because there are many useful
applications. Rob D.


JL-    I recon (not 100% sure) if you set the decode with the same gain on=

both M&S, unity say. Then boosted / attenuated as needed the L&R output
channels by the inverse of the amount used on the En-Code, you may (poss)
get back where you started.  Or as likely the only way you can change the
relative gains of the LR outputs is via a panpot just leave it centrally
panned and see if you've created 2 signals/mono files that are hopefully
identical to the original in all bar level. A null test with the noted
boost/cut you originally used should work.

BTW your original file is odd in that there is more S than M, IE unless you=

reduced the amount of S in the de-code your LR stereo will contain quite
some amount of out of phase info - but that's odd and not relevant to why
your not getting perfect reversal.

It's not clear to me why you chose to use 6dB of gain ** on the decode mix=

bus? It's just possible there could be some rounding or other mathematical=

errors [stuff like plugin being 24bit & I assume Logic uses 32float
internally]

Rob, you could you try the following
Take 2 mono files of, I suggest, Noise (any colour) & sine tone
Use these as your MS inputs & encode to MS (sum & difference) note setting=

of non unity gain gain controls
Decode - you need to be able to listen to just the L or the right output
Whilst listenining to one speaker (as it were) adjust level of either chan=

till you get just noise or tone. Note gain setting & tell us.


** I recon it may have to do with Max's point
 "MS to LR" is exactly the same as "LR to MS", except that you loose 6 dB o=
f
gain in the latter.
Well this is not some universal truth/rule but has some sense to it and is=

for example the way CEP/Audition does things. They quite reasonably decided=

that when one adds the L to R in a worst case if the mix were in fact doubl=
e
mono the sum would be 6dB louder - so to be safe reduce the level by 6dB so=

it doesn't clip. Unfortunately their clear thinking didn't extend to their=

de-code preset but if you work in 32bit Float as I do you can normalise any=

greater than 100% modulation back to what regular integer PCM can handle.

best
John L






<Prev in Thread] Current Thread [Next in Thread>
  • Re: [Nature Recordists] X-Y -> M-S & Perfect Reversals, Rob Danielson on Monday , March 12 , 2007 6 : 10 PM <=
Admin

The University of NSW School of Computer and Engineering takes no responsibility for the contents of this archive. It is purely a compilation of material sent by many people to the naturerecordists mailing list. It has not been checked for accuracy nor its content verified in any way. If you wish to get material removed from the archive or have other queries about the archive e-mail Andrew Taylor at this address: andrewt@cse.unsw.EDU.AU