The Gerbil’s Revenge: Musical Events: The New Yorker:
But pitch correction has also taken on a second life, as an effect. You’ve probably heard it, most recently on the No. 1 song in the country, Lil Wayne’s lazy, mildly naughty rap “Lollipop.” Auto-Tune, properly torqued up, is the rare edit that calls attention to itself. Auto-Tune software detects pitch, and when a vocal is routed through Auto-Tune, and a setting called “retune speed” is set to zero, warbling begins. This, roughly, is what happens: Auto-Tune locates the pitch of a recorded vocal, and moves that recorded information to the nearest “correct” note in a scale, which is selected by the user. With the speed set to zero, unnaturally rapid corrections eliminate portamento, the musical term for the slide between two pitches. Portamento is a natural aspect of speaking and singing, central to making people sound like people. A nonmusical example of portamento would be “up-speak,” a verbal tic common in some people under thirty. (Can you imagine the end of every sentence rising in pitch? Like a question?) Processed at zero speed, Auto-Tune turns the lolling curves of the human voice into a zigzag of right-angled steps. These steps may represent “perfect” pitches, but when sung pitches alternate too quickly the result sounds unnatural, a fluttering that is described by some engineers as “the gerbil” and by others as “robotic.”
The first popular example of Auto-Tune’s distorting effect was Cher’s 1998 hit “Believe,” produced by Mark Taylor and Brian Rawling. During the first verse, Auto-Tune makes the phrase “I can’t break through” wobble so much that it’s hard to discern. More successful is the gentler variation in the following line, “so sad that you’re leaving,” which highlights the software’s strength. Auto-Tune can produce a controlled version of losing control, hinting at various histrionic stations of the human voice—crying, sighing, laughing—without troubling the singer. It is notable that “Believe” ’s big chorus—“Do you believe in life after love?”—is delivered (mostly) in a full, human-sounding voice, with no robotic modifications. You can only feel so bad for a robot.