As I deepen my processes and the understanding of the work at hand, I revise concepts that I knew, or that I thought I knew, finding new nuisances and details. One of them lately has been comping and/or overdub, two concepts that for me until now kinda lived in a grey area where I used them pretty much interchangeably.
Now I've come to differentiate these, and these are the definitions I use. It's been a while since I last researched the "academic" definitions of them -and my brain makes "ouch" at the mere idea of going into that again-, so bear in mind that these definitions are only mine, the ones validated in my process by the value of use. Your millage will for sure vary.
The issue came to mind first when I was tracking guitars, and for the first time I tried doing it in these two phases:
1) Song from beginning to end
2) Listen to the song; whenever you find a fragment not quite to your liking, stop and re-record it.
This is what I understand as overdubbing. I hadn't done it in the past because I was afraid of problems of "vibe", the "body" of the track rejecting the new transplants... But hey, with guitar, as long as you don't do something silly in the middle -like changing parameters-, the newcomer is going to be impossible to differentiate from the earlier recording.
I've found this to be not only a swift workflow to generate guitar tracks, but also a very pleasant one, because of the "load leveling" of tasks between playing-listening. (There is, though, an entry toll: make sure you go into this process having the parts well rehearsed.)
Then I moved to tracking vocals, and that's a different beast.
My usual method here (which is also the one I applied to guitars in the past) goes:
1) Record song from beginning to end. Then again another take form beginning to end. Then again another take from beginning to end...
2) Relisten creating chunks of sections that you like/don't like, etc in each
3) Create a "Frankenstein" of the best moments (a take of "supernatural perfection", like the guys of Sound on Sound put it).
This is what I call comping. As in short for "composing", compose something out of different loose parts, like choosing the best flowers to "compose" a bouquet.
The overdub system had worked so well with guitars, that this time I was tempted to try it on vocals too, but in the end I didn't do it because singing requires a lot more of setup for me right now (take the complications of guitar takes and add bottles of water, pop filters, abundant kleenex, and hundreds of "parameters" like body posture, "role" you play in the song...)
I'm like a "one man orchestra", so I was afraid of unavoidable vibe problems if I went "atomic", plus vocal sessions are already very tiring for me as they are to add more instability by now; I prefer to consolidate and iterate what I got a bit more. Additionally, for this song, the usage of proximity effect was very important, so any variation in distance to the mic was going to add up to the instability, etc...
Maybe in the future, as I streamline my process, I'll be able to do it, or maybe for certain songs; I'd love it because it's very "tight", and fixing problems at the source is a lean principle for a reason; it makes infinitely easier to locate and countermeasure any mistake, while with the other method you're generating a lot of takes, inventory that later you have to go through, sift, etc.
Add to that the fact that, with the singer and the listener being the same person, and with human voice being the complex musical instrument it is, subjectivity sometimes plays tricks on you; in more than one occasion I have left vocal sessions not knowing AT ALL on which part of the cool/sucks spectrum lived the takes I had just recorded. Zero judgement capability. Imagine doing that with 1 second fragments.
This limitation probably yields a lot when you stop being the one-man-orchestra; I've seen at least two examples of bands using vocal overdubs; Deicide, whose vocalist punches in sentence-after-sentence to assure intelligibility of the lyrics, and also in Motörhead. In a 2010 documentary I saw there is this scene: Lemmy sings a missing sentence / Producer says "nope" / Lemmy goes again / Producer says... Etc.
I have a guitar and I'm gonna use it