Next gen stereo image tool teaser aka vocal removal

[Space Boy]

An extraction window border distortion demo. I found an example with an instrument that occupies the full stereo space. There's no way to slice the image without touching the piano:

Center signal (ie common signal)

Stereo sides signal

Notice how the rest of the signal stays artifact free apart from the piano.

Both of these contain 'watery' artifacts and the center signal version sounds like a mono version without any vocal isolation whatsoever.

P.S. I just tried re-combining the two parts and the artifacts remain - which kind of verifies what I suspected in the 'dusty' demo.

[Kingston]

arghgs!

well of course it contains artifacts! hence the name 'extraction window border distortion demo'. There was never an attempt to isolate the vocals. 2-to-3 (unsuccesfull) unmixing is what's happening. Read that post again, please. It's explained there. "sounds like a mono version without any vocal isolation whatsoever. " Yes of course it does. Everything but the piano and strings are panned dead center



By the way, you are in no way at disadvantage when it comes to these examples. Here's my mp3 example summed like it is supposed to be done:

mp3 center and sides summed back to original stereo and encoded to mp3 AGAIN


Ie, There's twice the mp3 encoding artifacts.

Like was mentioned before, precision sample ripping demo to come to satisfy the need to benchmark isolation performance.


Notice that when summed, the center is summed twice as loud (+3dB compared to sides) in traditional mixer environments. Could be you've summed the center at half the volume if there's distortion. in fact I recommend you redo the dusty extraboy demo, since you probably didn't know to take this into account. It might sound better.

remember what the pseudocode says:

Xsidesleft+center=left
Xsidesright+center=right

center is summed to both 'sidesleft' and 'sidesright' signal "separately". If you do this in a mixer environment it just means doubled volume for center.


(edit. the "doubled volume" in this case is actually only 3dB and not 6. I was talking out of my ass)

[Kingston]

Because I think vocal removal is such a tired and limiting concept, I'm going to stop talking about it. It really isn't what's going on here.

And without furher a do, I'd like to demonstrate the real power of this thing.


file1
file2
file3

The clips are sampleaccurately in time, and it's up to you to mix them back together. doesn't matter what used to be left/right/center/mono/sides... It's up to you to decide.

That used to be a normal stereo mix of course, but this dividing of the image broadens up the opportunities for remastering and surround work
"a tad".

I could even divide it to 5 or more slices.

Gotta love the incomprehensible ramblings of marvin in that song. you can finally hear what he's saying.


note to self: make a plugin out of this and let people play with it.