Ever play the perfect solo or riff except that one note was just a little bit early, or a little bit late? If you never have, then stop reading this article immediately, move to Los Angeles, New York, or Nashville, and start doing session work. Otherwise, keep reading.
Digital Audio Quantizing Tools
The first time I saw audio quantization was in Opcode’s late, great StudioVision program. The process required signals with strong transients and fast attacks, like drums, guitar, bass, piano, etc.
Quantization was a two-step process: First, you applied the equivalent of a noise gate to strip out all the silences in a digital audio track. Thus, all the notes with sharp transients started a new “block” of audio. The beginning of each block could then snap to a grid, just like MIDI notes, to quantize the parts.
Separating audio into smaller blocks allows quantizing them to a grid.
Beat Detective for Pro Tools takes a more sophisticated approach. In a nutshell, it too looks for transients, but works pretty much automatically. Once it finds the transients, it moves them so they fall on a grid, and extends or shortens the audio in front of the transient to make up for any timing inconsistencies.
There are several variations on these basic themes, but let’s look at a process that works with just about any DAW software. It’s definitely not an automatic process; you need to slice audio manually, and move the pieces around yourself. While the downside is that this is less convenient, the advantage is that you can be selective in what you do or don’t quantize.
This is important, because it’s okay if notes aren’t quantized exactly to the beat, as long as they sound right. Those little tempo variations that lead or lag the beat are a crucial part of providing emotional impact. I’ve analyzed guitar solos where notes would come in 30 or 50 ms late compared to the beat, but that added a cool feel—quantizing them to the beat took all the life out of the solo. So, always use quantization to fix mistakes based on what you hear with your ears, not what yousee with your eyes.
Another issue is that quantizing in isolation can lead to problems. For example, suppose the drummer played the snare a bit late compared to the beat, and all the players followed right along. If you quantize just one part, then its timing will sound “off, ” so you’ll have to quantize the others too . . . and it all might be for nothing if you then decide that the tune sounded better with the snare lagging anyway. If it ain’t broke, don’t fix it.
But This One’s Broke!
What got me into writing this article was working on a new library of looped guitar riffs called “Adrenalinn Guitars, ” using effects primarily based on Roger Linn’s Adrenalinn processor. Because loops like these are generally used in conjunction with quantized drums and other loops, the notes need to be right on the beat so they fit in properly. If someone using the loops wants to add a timing variation, it’s usually done by inserting a small tempo change to lead or lag, not moving bits and pieces of audio around.
The upper part of the following illustration shows the original guitar riff (only one channel is shown to save space). The four lines (three white, one red) indicate eighth note divisions. Note that the first note is right on an eighth note division, the second one is just a tiny bit ahead, the third note lags by a bit, and so does the fourth.
A guitar riff before and after quantizing.
The lower part shows the same riff after quantizing, using the technique we’re about to describe. The timing has been perfected, but without any glitching, pops, or other indications that quantizing took place.
Let’s first show how to quantize a note that lags behind the beat; as an example, we’ll fix the last note in the above illustration (the one that lags the red line). Here’s the step-by-step:
- Zoom in on the waveform.
- Turn off any snap-to-grid function.
- Split the clip at the precise beginning of the lagging note. A yellow line indicates the split in the top third of the following figure. Also split at the end of the note, or before the next major transient. Basically, we want to isolate the section that needs to move.
Three steps involved in quantizing a note that lags the beat.
4. Turn on the snap-to-grid function (in this case, it’s set to eighth notes).
5. Slip-edit the section of audio just before the lagging note, and move it back to the point where the note should start. Because snap is on, the end of the clip should snap right into position on the eighth note. This opens up a space between where the note should start and its current location (as shown in the middle third of the illustration).
6. Move the lagging note left to the proper start point, in the direction indicated by the arrow in the middle third of the above screen shot. Note that this will also open up a space at the end of the clip, which is why we added a split there in step 3. Otherwise, when you moved the clip to the left, it would move the entire rest of the track to the left. However, you will likely need to go back and close up this space, probably by crossfading in a manner similar to what’s described in the second step-by-step example.
7. Crossfade over the split point to eliminate any clicks or pops. Now the audio looks the lower third of the screen shot. Roll the crossfade out starting from the note’s attack, and move left so that it crossfades with the existing decay.
8. Audition what you’ve done to make sure all is well. Adjust the crossfade for the minimum amount of crossfade needed for a natural sound.
If the note is ahead of the beat instead of behind, the process is somewhat similar.
- Zoom in on the waveform.
- Turn off any snap-to-grid function.
- Split the clip at the precise beginning of the leading note. A yellow line indicates the split in the following illustration. Also split at the end of the note, or the next note transient.
Three steps involved in quantizing a note that leads the beat.
4. Move the leading note to the right so that it snaps into position on the eighth note (middle third of the above screenshot). Do not slip-edit the section of audio prior to the note attack to cover up the space, as that will cause a double attack. (Note that because we added a split at the end of the leading note, moving the clip to the right will cause its end to overlap the start of the section after the split. You will need to revisit this later and fix it, probably by slip-editing the clip so it no longer overlaps the beginning of the next one, then crossfading to smooth over the splice.
5. Crossfade over the space and into the previous decay to eliminate any clicks or pops (the ower third of the above screen shot). Roll the crossfade out starting from the note’s attack, and move left so that it crossfades with the existing decay. Note: If the gap that needs to be covered is large, prior to crossfading you may need to add a hard fadeout to the end of the clip. Otherwise, even with crossfading, there may be an abrupt cutoff.
6. Audition what you’ve done to make sure all is well. As before, use the minimum amount of crossfading needed for a natural sound.
This may seem time-consuming (and if you’re trying to fix all the glitches for a player with a timing problem, it is). But to fix the odd note here and there, this technique provides a simple, transparent-sounding solution.
Originally published on Harmony Central. Reprinted with permission.