Second, in the example you suggested, you said you were making as many as three cuts per song. OK, that's fine if you are splicing the track that is mixed lower, but in my case the track I need to alter is the "good" one and when i get them aligned, i will mix the "bad" one down to inaudible, so I don't think that method will be as desirable, as the cuts could become much more noticable without the other track to wash them out.
i see your point as well. you're taking a video track as your base sync track (this can't be altered) and you want to sync an alternate audio to it. i've done this as well.
in this case, i choose my splice locations very carefully. i can recommend few places where splices will be masked:
1. very even audience noise. (works extremely well when sliding a splice forward).
2. few samples just before a cymbal crash. (this works particullary well when you're sliding the splice (<<) forward (shortening), because the cymbal will overlap just a few thousanth of a second of music prior to the cymbal). this method works well even when you slide the splice (>>) backwards (lengthening) since you really can't detect a few thousandth second silence.
3. other similar loud/short transient attacks will also work.
also, if i really have no choice but to use time shift, i would apply it to a very short portion of the audience noise and never over the music. it's extremely difficult to tell if a 1/2 second portion of the audience noise has been shortened or lengthened by 1-2%.
This is a great thread, guys.
i hear ya!! +t for the good discussion.
marc