How to isolate phase-correlated signal between Left and Right channels
Hi everyone, here's my problem — and sorry if it sounds a bit basic for you...
I have a number of Stereo recordings which were originally MS, and have been converted to LR by mistake. Original M and S signals are lost. I HAVE to get the M back.
So what I need is to extract the only part of the signal that's strictly phase-correlated between Left and Right...
[ I tried to do the maths back, like the good old sum-difference trick used in mastering for stereo enhancement and so forth, but it won't do; probably because the original S signal is from an actual fig-8 mic, therefore the signal from the negative end is far from being the inverted positive end; in other words: S ≠ - (-S) ]
So my guess is that I need some fft processing that could eliminate everything that's non phase-correlated, maybe like fft—>cartopol—>unknown math ? dunno.
I'm sure most of you signal-processing-experts have done it already, or you can figure it out pretty quick..
Thanks a LOT. Best,
Olivier
if you first properly time aling the tracks, it should be possible to just encode it to M/S again.
not sure what the best way is, but have you tried doing it manually (or to be exact: graphically)? put a tapin/tapout delay on both tracks, fiddle with the settings, and chekc the result in a large display where you can see (at least) single samples.
& you could make a cut and not process and not encode the spectrum over 10khz or so.
Thanks. You're right.
I don't know why it didn't work last time, it does the job just fine.
Sorry for the useless post.