FIG. 1: The top (yellow) loop is one bar long whereas the bottom (green) loop is a 128th-note shorter. They slowly drift out of sync as they loop until, after eight bars, they are a 16th-note out of sync (red marker). They come full cycle and are back in sync after 128 bars.
Layering a phrase over itself with a time offset is as old as the canon, but the technique took a giant leap forward in the ''60s with the use of tape-loop techniques pioneered most notably by American composer Steve Reich. In addition to being a fascinating and well-explored process for composition in many genres, phasing is a powerful sound-design tool that you can use with material as diverse as percussion, speech, pads and ambient sounds, and rhythm tracks (guitar, keyboards, etc.).
The most basic tape process amounts to using two loops of slightly different lengths or two loops of the same length running at slightly different speeds. The results, when reproduced digitally, can be manipulated in different ways. I''ll start with the loop-length approach, which is a bit simpler and easier to reproduce. (In the days of tape-loop splicing and analog tape machines, neither length nor speed was absolutely precise, so both processes were always in play.)
LOOP AND COUNTER-LOOP
Start with a fairly active 1-bar electronic-percussion loop and copy it to two tracks of your DAW. You can do this with audio or MIDI, but choose an audio loop so that you can visually compare the waveforms. Set your DAW''s tempo to match the loop''s, shorten one of the loops by a 16th-note, and then repeat both loops for 16 measures. Notice that at each measure, the shorter loop falls behind by a 16th-note until, at measure 16, the loops are back in sync (see Web Clip 1). Among the 16 1-bar segments you''ll find some interesting variations on the rhythm, which you can then render as new loops.
Repeat the process, but this time shorten the loop by a 16th-note triplet. Now the process will cycle after 24 measures but, depending on your material, fewer of the 1-bar segments will be usable. The ones that are useful will often introduce some syncopation and swing into the pattern (see Web Clip 2).
Large offsets in loop length, like a 16th-note or 16th-note triplet, tend to preserve rhythm, whereas much smaller offsets bring out interesting aspects of pads, ambient sounds, and speech. Start with two 1-bar sound effects loops (they don''t need to be the same) and shorten one by a 128th-note (30 ticks in DAWs that use a 960-tick-per-quarter-note clock). The loops will now come back into sync after 128 measures. Unlike the previous examples, any rhythmic elements will drift in and out of sync rather than presenting cohesive rhythmic shifts. Listen to the whole cycle (roughly 4:15 at 120bpm of 4/4) and drop markers at interesting points to render as individual elements (see Fig. 1 and Web Clip 3).
You can use a delay send-effect to simulate loops playing at different tape speeds. The key is that the delay time, which should not be synched to tempo, needs to increase linearly over the entire process. Unlike using different length loops, using different speeds causes the loops to continually drift further apart rather than jump at each iteration of the loop. Automate the delay time with a straight line starting at 0ms and ending with a delay time equal to the length of the loop (see Web Clip 4). Linear automation emulates the classic tape-loop process, but there''s no need to limit yourself to that. Stopping the automation for a while, introducing slight up and down drifts, and using nonlinear curves will produce interesting results.
Len Sasso is a freelance writer and frequent EM contributor.