Pitch correction should fix obvious errors while preserving the micro-variations that make a voice sound human

Pitch correction (Auto-Tune, Melodyne, Waves Tune) is a near-universal tool in modern mixing, but its misuse creates the robotic ‘Auto-Tune effect’. Key principles: even the best singers are never perfectly on pitch — it is the slight variations that make a voice human. Use the minimum number of corrections, focusing on notes that genuinely jar. Use graphical (manual) mode rather than Auto mode, which causes audible fluctuation. Don’t correct to exact pitch — get within a few cents. Prefer vocal comping and copy-paste of good takes before applying correction. Background vocals tolerate more correction than leads. When a vocal is too far out of tune, detune two copies by ±2-8 cents and blend them — this smooths intonation and fattens the sound simultaneously.

Examples

Ken Scott: ‘Of the four albums I co-produced with David Bowie, 95% of all his vocals were only one take. There are small errors in pitch and timing if you listen closely, but that’s what makes David sound like David.’ The ±cents trick: copy vocal to two tracks, tune one +5 cents, other -5 cents — blends for a natural, thicker sound.

Assessment

A vocal track has three obviously sharp notes in the chorus and several slightly wavering notes throughout. Describe your complete pitch correction workflow: order of operations, which notes to correct and which to leave, the mode to use, and the criterion for ‘enough’ correction.