Phase is a dual sample looper for VCV Rack from Signal Function Set, built around an idea borrowed from Steve Reich’s tape pieces: take two copies of the same loop, let one drift slowly against the other, and listen to the patterns that emerge as they move in and out of alignment.
Because the whole effect rides on the relationship between two loops, the source material you load matters more here than in most samplers. A sample that sounds great as a one-shot can turn to mush under phasing, and an unremarkable field recording can become hypnotic. This guide is about which samples work, why, and how to get them ready.
What Phase expects from a file
Phase is forgiving on format. It loads WAV files in mono or stereo at any sample rate, mixes them to mono on load, and resamples to 48 kHz internally. The maximum length is about 10 minutes per sample. File paths are stored with the patch, so a loaded sample reconnects when you reopen your work.
Two things follow from that. You don’t need to pre-convert anything to a specific rate or bit depth, since Phase handles it on load. And stereo information in your source doesn’t survive that load: Phase sums to mono and builds its own stereo image through per-loop panning. There’s no reason to fuss over stereo width in the source file. Pick samples for their content, not their channel count.
What makes a good phasing sample
The effect is only as interesting as the relationship it creates between the two loops. A few categories of source material tend to reward phasing.
Rhythmic and percussive material. Reich’s “Clapping Music” and “Drumming” are the obvious reference points. Clear transients mean the phasing reads as discrete hits sliding against each other, producing cross-rhythms that neither loop contains on its own. Phase’s transient detection, which lets the playhead jump to detected click points, is built for exactly this kind of material. Drum loops, hand percussion, mechanical clicks, and footsteps all work well.
Spoken word. Reich’s earliest tape pieces, “It’s Gonna Rain” and “Come Out,” looped fragments of recorded speech and let them drift apart. Speech carries strong rhythmic and tonal structure, so a short spoken phrase under phasing produces the same churning, uncanny result Reich found in the mid-1960s. A few seconds of clearly-articulated speech is plenty.
Melodic phrases. “Piano Phase” and “Violin Phase” applied the technique to short instrumental figures. A repeating melodic cell a few notes long generates shifting harmonies as the two copies separate, because notes that started in unison end up stacked against each other. The shorter and more distinctive the phrase, the clearer the effect.
Sustained and textural material. Drones, pads, bowed tones, and field recordings phase differently. Instead of discrete hits sliding apart, you get slow timbral beating and comb-filter motion as the two near-identical signals interfere. This is less rhythmic and more atmospheric, and it suits ambient patches where you want movement without a pulse.
What tends not to work is long, busy, through-composed material with no internal repetition. If the source already changes constantly, the drift between two copies gets lost. Phasing wants source material with enough internal structure that the ear can track the two copies pulling apart.
Prepping a sample before you load it
Phase doesn’t require prep, but a little goes a long way.
Trim to the loop you want. Phase has draggable loop-region handles, so you can set the loop inside the module. Starting from a tight, pre-trimmed file still makes that in-module work faster and keeps your library readable.
Find clean loop points. For rhythmic and melodic material, a loop that starts and ends on a musical boundary (a downbeat, the start of a phrase) phases more legibly than one with an arbitrary cut. For textures, loop points matter far less, since there’s no pulse to align to.
Mind the transients. Because Phase can jump the playhead to detected transients, samples with clear, well-separated attacks give you more to play with. If you’re slicing a longer recording into loop candidates, cutting at transients rather than at fixed intervals tends to produce better starting points.
Watch the length. Phase tops out around 10 minutes per sample at 48 kHz, which is far more than most phasing patches need. It’s only worth thinking about if you’re feeding it long-form field recordings.
Sleep and Rotate, and what to feed each
Phase offers two drift mechanisms, and they reward different source material.
Rotate applies a continuous speed offset, so one loop runs slightly faster or slower than the other and the two slide steadily apart. This is the classic Reich tape-phase behavior, and it suits rhythmic and melodic loops where you want to hear the hits or notes walk against each other over time.
Sleep inserts silence gaps between loop repetitions, so the loops fall out of step in discrete jumps rather than a smooth slide. This works well with shorter, more percussive material, and with spoken-word fragments where a staggered, stuttering relationship is more compelling than a smooth drift.
In practice you’ll often load the same sample and audition both, since the right choice depends as much on the sample as on the patch.
Building a phase-ready library
Once you start patching with Phase regularly, it helps to keep a folder of source material you already know phases well, rather than hunting through your whole library each time. A working set might hold a handful of percussion loops, a few spoken-word snippets, some short melodic cells, and a couple of drones or textures.
This is where a sample manager earns its place. SampleStack is a native macOS app for organizing and editing exactly the kind of library you’d draw on here: you can tag and rate source material, trim and set loop points, audition quickly, and keep a curated phase-ready collection separate from the rest of your samples.
Phase runs in VCV Rack and on the 4ms MetaModule, so the same prepared samples work in both places, software and hardware.
If you’re coming from hardware, two Eurorack modules cover related ground. The Make Noise Morphagene approaches tape and splice manipulation from a granular angle, and the Instruō Lúbadh is a dual tape-style looper whose two-reel design sits close to Phase’s two-loop concept. Samples prepped for one are usually a short step from working in the others.
FAQ
Does Phase need samples in a specific format?
No. Phase loads WAV files in mono or stereo at any sample rate, mixes them to mono, and resamples to 48 kHz on load. You don’t need to convert anything in advance.
What sample length works best?
For rhythmic and melodic phasing, short loops (a bar or a phrase) read most clearly, because the two copies stay close enough that the ear can track them drifting apart. Textures and drones can be longer. The hard ceiling is about 10 minutes per sample at 48 kHz.
Can I use stereo samples?
You can load them, but Phase sums to mono on load. Its stereo output comes from panning the two loops independently, not from the source file’s stereo image, so there’s no benefit to preparing wide stereo source material.
Do I need SampleStack to use Phase?
No. Phase loads any WAV file directly. SampleStack helps on the other side of the workflow: organizing, tagging, trimming, and setting loop points on the library of source material you draw from, which is where most of the time goes once you patch with Phase regularly.
Where do I get Phase?
Phase is available for VCV Rack and the 4ms MetaModule from Signal Function Set.