Sorry Clio, although I respect your judgments on most audio matters, I think Rrog's analogy is inept. The issue of remaining in the analog domain ignores the major energy transductions from acoustic to electrical then back to acoustic energy. IMO these transductions are far more difficult than sampling a waveform. Mathematical representation of the even high frequency energy in the waveform is entirely possible for digital processing. In either case, the waveform is most likely to have been originally recorded to tape with vagaries like head alignment and print-through if an analog tape recorder was used.
db
db