Doubling and Harmony
Doubling thickens one voice, but a third copy starts fighting itself in phase.
Related
The problem
Doubling records the same line two or more times and stacks the takes, making one voice thicker or making it sound like several. It is cheap, effective and very easy to turn plastic. By the third layer the problem is rarely the singing. It is phase. The harder task is keeping the thickened result sounding like one melody rather than four lines each doing its own thing.
How it works
Two nearly identical waveforms summed with a very small offset cancel each other, producing comb filtering: the top end drops first and the sound turns thin and sharp. Once the offset reaches fifteen to thirty milliseconds, the ear begins to hear two sources. A few cents of detuning also helps, because the formants no longer coincide exactly. Harmony stacking is a different operation. It changes pitch relationships, where thirds and sixths are stable and seconds tend to collide with the melody.
Practice
Keep the lead centred and leave it mono. Place doubles left and right, each detuned five to ten cents and delayed ten to twenty-five milliseconds, sitting three to six decibels lower. Write harmony in thirds and sixths first, check for minor seconds against the melody, and give the lower part the root and third. When a generated take is used as a double, choose a similar but different timbre rather than running the same model with the same seed twice. Check in mono after every added layer. If it collapses, remove a layer instead of adding more to compensate.
What to listen for
In Monteverdi's madrigal, listen to how each part stays independent. In Weelkes's four-part anthem, listen to where the thickness of the block comes from. In the four-part chorale, hear the simplest stack avoid mush. In the slow movement of Beethoven's quartet, listen to four lines making room for one another. Doubling should aim at these things, not just at a higher fader.
Listen in the library
All tracks from midicn-lib — inline badge is the license tier (C1 commercial / C2 non-commercial / C3 study only).
Sources
- 叠录与和声的相位问题,依录音实践与公开技术说明核对,未抄原文
- Doubling and phase cancellation, checked against published recording references, no text copied
Used for fact-checking only; all prose is original.