At 1:16 AM -0700 8/22/10, James Shatto wrote:
>I guess it depends on how you do the math. Not
>a real world test by any means, but I
>synthesized some stuff to better clarify my line
>of thinking. Not that you'll have conditions
>where M and S have matching phase + amp +
>frequency (much) in practical applications.
>But...
>
><http://home.earthlink.net/~shadow_7/MS_TEST.sh>http://home.earthlink.net/=
~shadow_7/MS_TEST.sh
incomplete extension on the uploaded file. .sh?
>
>It uses bash and sox to synth some pitches
>across a broad range. Kept the root and 5th of
>a chord unique to M and S. Mixed a common tone
>(the third of a chord) between the two tracks
>and the results were interesting. -S for the
>right channel completely cancels out the 3rd of
>the chord on the right channel. When going from
>M and S to L and R.
>
>Going from the generated L and R back to M and
>S. And +S and -S don't cancel each other out
>anymore. And to make matters worse the 5th
>which was unique to M (and NOT S) is the result
>of trying to cancel S out. Not that it's a real
>world practical test. And I probably didn't get
>the math just right. But it seems a bit
>interesting to me. Assuming that it's not some
>software quirk with sox or my math.
Hi James--
Should work with any two, single channel sound
files. Seems like your patch has issues.
Has anyone done this experiment with a PC
compatible plug-in in that is simple (e.g. having
only mid and side gain adjustments with a unity
position) like "+Matrix" is for Mac?
Maybe try the experiment using the suggested simple MS matrix plug-in? Rob =
D.
>
>I was only expecting to find some minute
>differences in the waveforms of the derived
>versus original tracks when I started out. And
>maybe some loss of frequencies in the upper
>bands. Needless to say that the quirks I found
>were VERY audible. The timing and other
>variances of actual M/S would likely not yeild
>the same results. Especially if M and S are NOT
>mixed at unity gain. But it should lend some
>credibility to recording the M and S tracks and
>not the L and R tracks in the field. If you
>intend on doing any editing IMO.
>
>- James
>
>--- On Fri, 8/20/10, Marinos Koutsomichalis
><<marinos%40agxivatein.com>>
>wrote:
>
>From: Marinos Koutsomichalis
><<marinos%40agxivatein.com>>
>Subject: Re: [Nature Recordists] M/S recording setup
>To:
><naturerecordists%40yahoogroups.com>=
m
>Date: Friday, August 20, 2010, 2:04 AM
>
>
>
>I still think that if the maths are done
>correctly there will be no loss at all.. some
>plugins may have 'strange' algorithms for doing
>this, but if you find the right one, or better
>if you program this yourself in some platform
>then it should be totally transparent..
>
>m
>
>On 20 =C9=FC=C9"=C9=A1 2010, at 7:09 =BC.=C9 ., Dan Dugan wrote:
>
>> > Yes the sample rate remains the same. Yes
>>the file size probably remains the same. And
>>yes the duration of the recording stays the
>>same. But I find it hard to believe that a
>>twice edited digital file (L/R -> M/S -> L/R)
>>is a bit for bit exact copy of the
>>original(even if you exclude the headers from
>>the comparison).
>
>>
>
>> I wouldn't expect it would be, but the
>>"generation loss" would be down in the noise
>>bits and insignificant. MS equalization is
>>quite commonly done in mastering suites, where
>>you find the best monitors and ears in the
>>business.
>
>
>
>
--
|