CitizenDan wrote:Possibly an ignorant question, but:
Isn't today's noise-reduction technology intelligent enough to not only discern noise from music, but also to show the user graphically where each resides in a waveform, so that the noise can be worked on without touching the music? Or do noise and music exist in separate frequencies so seldom that this isn't feasible?
The trouble with noise (more specifically, hiss), is it tends to be random in nature. 1) It actually covers quite a wide frequency range, and 2) there's no mechanical repetition to it, like there is with hum for example. So when a noise reduction algorithm takes a snapshot of a segment of tape hiss, by nature it's only going to be an approximation of what will actually be "reduced".
Take that to the next level then, and compare the range of frequencies of tape hiss with the frequencies from, say, a snare drum rattle. There's not a lot of dfference. Practical example? Take Come Together from the Abbey Road CD, load that into your computer and sync it with the same song from the Blue CD. The Blue CD has a lot of hiss removed, but have a listen to what it has done to the drum kit aswell.
It doesn't matter how "intelligent" an algorithm is - if no two bits of hiss are the same and it goes for something that is "close enough", how is it suposed to avoid hacking into the music? If a bit of hiss has gone, then ...
