Sample Loops and Crossfades

Loop points keep a sustained note alive; crossfades decide whether the seam is audible.

Related

SamplerTimbreSamplers

The problem

How does a two-second sample hold a ten-second chord without an audible seam?

How it works

The sampler repeats the region between loop start and loop end. If those points do not land on whole cycles, every return creates a jump that reads as pulsing or buzzing. Two fixes are common: align to zero crossings or whole periods, or crossfade so the tail overlaps the head. A short fade leaves a seam; a long one sums two out-of-phase waveforms and produces comb filtering, thinning the tone. The attack stays outside the loop, which covers only the steady portion.

Practice and settings

Place the loop in the most stable middle of the waveform and span several periods. Start crossfades at 20 to 80 ms and lengthen them for low notes. Align by eye on the waveform, then confirm on a spectrum so beating is not mistaken for character. Never loop inside a decaying section, or level will rise on every pass. After editing, audition a sustained note for a full minute.

What to listen for

Hear a long string note and listen for regular undulation. Hear an organ chord, where a seam may click at the attack of one voice. Hear a choir sustain, which thins like a scooped midrange when the crossfade is too long. Hear a low drone, where a misaligned loop is most easily heard as vibrato.

Common misconceptions

That a shorter loop always saves memory without cost; that a crossfade hides any seam; that looping affects length but not timbre.

Sources

• Curtis Roads, The Computer Music Tutorial — 循环与交叉淡化的做法核对，未抄原文
• Ken C. Pohlmann, Principles of Digital Audio — 采样回放的接缝与相位问题核对，未抄原文

Used for fact-checking only; all prose is original.