Comp Editing

[rodzimguitar68]

New user here, long time Logic Pro X user.

I would like help on comp editing of takes. I watched the short video where the host used a Keyboard Bass, and said let's pretend this was a few vocal takes.

In the demo, the host was using fadeins/outs on the beginnings and endings of regions.

My question is how to construct a take where I might want some specific pieces of a region, but not the whole region.

In other words, I've got a vocal passage attempted 5 times, and there are 20 words, so I can choose between 5 performances of each word. I might want to surgically construct a great performance by jumping back and forth between the 5 takes, until I'm happy with the Frankenstein performance. Yes, the easier way would be to work with better musicians, but this is still my question nonetheless.

In logic, takes stack up in the track and you can highlight the range you want, and logic performs the crossfade which you can then bounce into a cohesive region that is continuous.

From the looks of the Harrison video demo, I would need to clip my 5 takes into 20 little, word long regions, and then select the ones I want to use, and cross fade them manually.

Thanks for the help!

[PBuryk]

Wow.... I've never used Logic, but that sounds amazing...

Working, almost exclusively, with non-pros I am often comping takes.
However, they rarely occur such that I can cross-fade across them; their fixes falling
basically falling into 3 categories:
- Playing/singing wrong notes/words
- Passage execution
- Passage timing (syncing with other performed parts)
While the first two, above, usually occur at the same point in the timeline that might
allow a cross-fade fix, I opt to cut-and-paste them into a separate track. This is
because the clear majority of my comp fixes are of the "Passage timing" category
and they need to be moved forward or backward in the timeline (can't use cross-fade
to correct this). So, my comping method is a simple cut-and-paste of the best
replacement part into a separate track. Sometimes I only need to do this a couple
times per performance, other times I'm truly building a Franken-Track.
Regardless, that's the method that I find works best for me.

Cheers!
Patrick

[bjohnh]

Logic's quick-swipe comping system is excellent, although it can get a little tricky if you didn't perform all your takes to a click track (the music I do never uses click tracks).

In Mixbus, you have a few ways to do comping. None of them require you to split up all your takes into individual phrases! Split out only the specific good phrases you want to use to replace bad phrases in your best overall take. See some workflows below.

1. Record your takes as layers (this is how Logic works as well). Move the best overall take to the top layer, and then move up the bits you want to replace from other layers (lock your track first so side-to-side movement is prevented, only vertical movement is allowed, that way everything stays in alignment). Whatever is in the top layer is your comp. The audition tool in Mixbus makes all of this a lot easier than it used to be: to listen to your other takes, or any portion of a take, click on any part of a take and press the "a" key to audition just that take starting from your mouse or playhead location.

2. Record your takes on separate tracks and assemble a comp by copying/pasting sections into a new comp track.

3. Record your takes as separate playlists on the same track(s), and do the same copy/paste routine as in #2 above.


I think the first approach is most efficient. One of the Harrison videos shows how its done (sounds like you found it). Definitely not as efficient as Logic or even Reaper, but it works.

---

Also: Ben Loftis at Harrison posted previously that he uses this workflow for comps:

"(1) record the first full-length take onto track “A”

(2) create an overdub/retake track “B” (which can be muted)

(3) record retakes on track B (creating a new layer on top each time), and if you get a good piece, cut out a region and drag it up to track A. The region will have a short crossfade on each end, cutting out/in from the original take.

(4) when you are done, disable B. It will be stored in the session if you need to go back to it, but it won’t use up any disk throughput or dsp resources.

The best time to make a decision is right after the take, when you remember what the good parts were. If you make 10 retakes, then try to choose among them, you’ll waste a lot of time."

[JamieLang]

This is a really, really short shortcoming of Mixbus, IMO.

I really don't do any significant comping outside of the lead vocal. The above explanation is basically the way...it's a kludge even compared to a tape machine, honestly, which if I had to do it regularly is how I would do it--record the takes to different tracks...and live rerecord onto a new track through a utility bus like basically we used to do it with a tape machine, albeit hopeful with less generation loss.

But, I also find the Logic tool lame AF. It's so "easy" to do some big swipes, and that looks cool--until you want to go back and change this or that, and there's no way to audition other takes WITHOUT "swiping" them into the comp...which ends up putting extra crossfades it's hard to get gone...Cubase does a little better, but their implementing a "comp tool" was silly. It worked perfectly fine when you cut them apart and then selected them with the same tool you'd adjust fades or drag in time....now there's just a THIRD tool involved that sure let's you quickly toggle two options back and forth...but...

Anyway--I will also point out as a singer that you'll never get the word by word thing. You need to split it by breaths....and then OCCASSIONALLY there's some rare issue where you can find some "word" to fix an otherwise great line with in another take...

If anyone cares what I think on the subject, I think Avid does it best with Playlist comping. I would simply copy what they've done....screw all this neo "swipe to comp" trend. They tend to make for ironically more clicks and more unneeded crossfades to clean up in order to "look slick" in a demo.

[PBuryk]

As this topic is, indeed, posted in the Forum's Techniques area, not Feature Requests,
I will simply agree that bjohnh has posted a good choice of methods by which one
can comp takes.
I prefer to use method #2 as I find it addresses any level of correction that I've
encountered. And, so, if asked, that is what I would recommend.

As far is why one should have to address performance issues by comping at all, or whether
one particular tool can accomplish the job automatically, etc... These are topics for
separate discussion.

Cheers!
Patrick