Don't think of it as my giving him a challenge so much as my giving him what he originally asked for.
And, to be honest, I can understand it being "fun" to do such things; it's a challenge, a problem, a puzzle to try and solve. I personally get some fun out of doing similar "forensic-style" work with photographic images; removing artifacts, old photo repair, analyzing for reality vs. touch-ups, re-touching and re-comping, etc. So I see nothing wrong with getting a kick out of doing similar things with audio. I in fact used to pride myself on my old-school techniques for cleaning up and removing noise that I was forced to develop before most of today's NR tools were available (even though those techniques were labor-intensive and mostly obsolete when compared to the quality of the tools available these days).
And what engineer worth their salt doesn't like the feeling they get when confronted with a sonic challenge in their recording that they manage to solve where others fell short after a modicum of forensic analysis via phase, frequency or spectral analysis (or all three?) So IMHO, I'd cut bobly some slack in that regard.
There is a side issue that I have a hard time buying that he went out and bought a $2500 editor program just for fun. But the issue here is more what can be claimed for the use of that software; we'll worry about the source of that software later.
Bobly may yet prove me wrong, but in the meantime I am confident that regardless of the nature of the instrument's timbre that clean separation of acoustic or electro-acoustic instruments is just not possible.
The question IMHO is whether the unavoidable artifacting can be checked well enough to make such rough separations useful from a music engineering perspective. Honestly, the examples given so far from that Leadbelly clip are far from convincing, and that was just a guitar and vocal, both of limited frequency response, in mono. The extent of artifacting there, IMHO, renders the separation functionally close to useless. It's an interesting exercise and demonstration of capability, but one with no apparent positive real-life musical engineering value that I can hear.
But bobly has thrown his chips into the pot with this thread. We can argue as to whether he's filling an inside straight or holding a a weak pair until our faces turn blue (with or without filter
) I decided instead to call the hand. If he's holding the winning hand, he deserves to take the pot and I deserve to get burned. If, on the other hand, he's really holding only an ace high pair of deuces, then the pot goes the other way. Fair and square.
G.
.
. But I'm not worried about that stuff, I just figured this would be a good test file for bobly to chew on.
. Two weeks is a LOT of labor for a 3-minute file. It's a good thing you're not charging for it, I couldn't afford it 
