There are so many concert recordings of awesome performances that sound like crap because they are audience tapes.
Often before a performance recorded music is played and captured in the audience recordings.
Would it be possible to train a model on an archive of these concert audience recordings of studio recordings paired with the original studio recordings to develop a system to “clean up” audience recordings?
I'm referring to the pre-concert, when they're just playing a mixtape of whatever through the board. Compare studio recording of whatever and the audience recording.
Often before a performance recorded music is played and captured in the audience recordings.
Would it be possible to train a model on an archive of these concert audience recordings of studio recordings paired with the original studio recordings to develop a system to “clean up” audience recordings?