Return to EQ index

All information in these pages is copyright (c) 1989-2000 by Roger Nichols. All rights reserved. Permission for personal reference only, and may not be reproduced by any method without written permission.

Vocal Comps
by Roger Nichols

One of the most time consuming operations performed on many recording projects is vocal comping With 48 and more audio tracks available these days, it is common to record 10 or more tracks of a vocal performance and then gather together the best parts to produce the final vocal track.

I am not particularly picking on Pro Tools, as all DAWs require the same operations to combine recorded tracks into a master track. Pro Tools just happens to be the system I have used most of the time for the last 10 years and is therefore easier to use as an example.

As much as I have tried to do comps in Pro Tools, it is still 10 times faster on the Sony 3348 48 track digital machine. First let me explain how it is done on the 48-track, and then I will share my ideas of how to do the same thing in Pro Tools.

3348 Version
Lets say that we have 5 or 6 tracks of vocals that we have been working with. We listen to all of them, pick the best parts of each, and bounce them into a comp track. Lets say that the first time we do this comp; all of the selections are easy, with no tight punches.

On the 48-track, you select the source and destination track, and record on the destination track. When you get to the spot where you need to use a different source track, you select the new source track, punch into record and continue. Pretty straight forward. Here is where it starts getting tricky.

You now have a comp track, and you record another 5 or so tracks of vocals. You listen to them and think that some of the phrases from the new tracks might sound good in the comp track. You can either punch the new stuff into the existing comp track, or you can make a copy of the comp track in case you change your mind or make a mistake and want to get back to where you were.

Since you are now just replacing small parts of the destination comp track, you set up the source and destination and punch in and out of record using REHEARSE mode. You now hear how the new part sounds in the comp. You sometimes change the in or out points and listen to the pre-roll, auto-punch, and post-roll while still in rehearse just to make sure that everything is ok. If the punch is good, then you do it for real with rehearse turned off. The Sony 3348 has a built-in cross-fade during the punch-in and punch-out. It is usually set to 10msec, but there is a front panel switch that you can change easily to try different cross-fade times from .1msec to 250msec. If the timing between the two vocal tracks is off, or you have to punch in the middle of a sustained note, changing the cross-fade time sometimes cleans it right up.

This method of comping works well, and is very fast. Lets say that you have decided on a word that needs to be replaced on the comp track. You want to try replacing it with the vocal on track 1. You set up the auto-punch as described above, and decide that the performance on track 1 is not quite good enough. You already have the auto-punch parameters set, so you just change the source track, and press auto-punch. If you had a lot of tracks to audition, you could put the machine in an roll-with-pre/post-roll loop, and each time the machine rewinds you could assign another source track to be punched in. When you hear the replacement you like, you turn off rehearse and do it for real.

Pro Tools Version
In Pro Tools you have to select regions in a source track, and drag them up to the comp track. Fine, if you are performing a simple comp without lots of trials to get the feel right.

Lets say that the vocal line you are working with is "I love you so much I could puke." The banter in the control room is as follows:

"Lets try replacing 'so much' with track 2."
"Naw, try it from track 5."
"The 'much' is ok, but let me hear the 'so' from track 2."
"The phrasing is early, but I like that 'so much', try the 'you so' from track 2 and the 'much' from track 3."
"No, that doesn't work, try the punch in the middle of 'you' with a longer cross-fade."
"Great, now lets see if we can find a better 'ch' on the end of 'much' cause it sounds like I was spitting out my teeth."
This is the way it goes. With the 48-track you can perform all of the permutations about as fast as you can say them. Just try it in Pro Tools.

If the edit stayed simple, like just trying the "so much" from various tracks, you would make a group out of all the vocal tracks, make a region, then copy the piece from each track into the comp and listen. As soon as you start sliding stuff around and trying to get new regions, you are dead. And besides, no cross-fades.

Possible Solution
Here we have our 5 vocal tracks and the comp track (they would be waveforms in Pro Tools). In the GROUP dialog, we make these tracks a "COMP GROUP". This would assign a DSP for cross-fades. Default cross-fade time 10msec.

COMP: I l o v e y o u s o m u c h I c o u l d p u k e.
TRK # 1: I l o v e y o u s o m u c h I c o u l d p u k e.
TRK # 2: I l o v e y o u s o m u c h I c o u l d p u k e.
TRK # 3: I l o v e y o u s o m u c h I c o u l d p u k e.
TRK # 4: I l o v e y o u s o m u c h I c o u l d p u k e.
TRK # 5: I l o v e y o u s o m u c h I c o u l d p u k e.
If were able to highlight disconnected regions (Shown by bold print) It would look like this. Playback or "LOOP" pointers would be set before and after the line. All tracks are assigned to the same output. Each track would use another voice so they could cross-fade.

When you hit the space bar, playback would start from the play pointer (or would loop between the pointers if LOOP PLAYBACK was enabled) only the highlighted areas in the group would be heard. Sort of what happens when you have multiple tracks assigned to one voice. In this case you would hear "I love you" from the comp track, "so much" from track 2, and "I could puke" from the comp track. The default cross-fades would be performed at each switch.

Basically, each track would be muted except where highlighted. The highlighted section would start fading up at the cross-fade time before the highlighted area, and fade out by the cross-fade time after the highlighted area.

If you moved the highlighted area on track 2 earlier, it would force the highlighted area preceding it on the comp track to move earlier, like the region trimmer. If you used the option key (or whatever) you could drag the highlighted area of one track without changing the highlighted area of another track, thus allowing overlaps or blank spaces between the selections. There would be a key command to make one region snap to the position of the previous region boundary, or the position of the next region boundary, in effect eliminating the space or overlap.

If you drag a highlighted area vertically, only the highlight moves, not the region. The highlight would be restricted from moving in the horizontal domain when being drug vertically, just like region drags. Playing a vocal comp like this would be a simple matter.

When you are done with one section, you simply move the play pointers to a new section. Highlights remain where positioned for the whole song so that you can go back to a previous area and change something. You could end up with thousands of regions in a comp group.

You should be able to add tracks to a comp group without disturbing the existing highlights. The new added tracks would have no highlights until you put them there, so they would be just like tracks that were already part of the comp group that were never used for anything.

You could add an empty audio track to at the top of a comp group. When the comp section is done, or when the whole track is complete, you could "COMP" the highlighted regions to the empty track. Each highlighted area would be copied as a region up to the empty comp track with the appropriate cross-fades, producing a normal audio track. The highlighted areas would remain in the original positions on the comp source tracks in case you wanted to change anything later. When you make a change and COMP the group again, it would erase the regions in the top track and do the whole comp again.

Since the new track is a normal track, it could be used as the source for a subsequent comp group, or you could bounce to disk to get rid of all of the regions if you wanted to.

Now let's see which method would be faster to comp vocals.