Pitch correction is a funny thing.
Sometimes it can improve a vocal recording. Sometimes it can make it worse. For me, the key to this is in understanding the interplay between pitch and emotion.
For many inexperienced vocalists, pitch correction often improves their recordings. Their poor control of pitch results in performance expression that is inconsistent with the creative direction of the music. In other words, notes sound off-pitch in a bad way. So, pitch correction provides an improvement. It makes the notes sound more like what was intended.
For many experienced vocalists, however, pitch correction is either neutral (and a mild waste of time) or even makes the recording worse. Great vocalists with excellent pitch control will deliberately use pitch deviations in ways that support and enhance the creative direction of the music. In other words, they sing off-pitch deliberately, and it sounds good.
The human voice is not robotic. It’s amazingly fluid and expressive. Quantising to the most-common (I.e in tune) pitches makes about as much sense as reducing the dynamic or tonal range of a performance – it might be appropriate for the vocalist or the music, but know that doing so restricts the expressive range of the vocalist’s performance. For vocalists that don’t have the skill to control their performance with sufficient precision, reducing the expressive range of the recorded performance can result in an improvement.
But sufficiently skills vocalists can make effective use of both types of extremes of their (pitch, dynamic and tonal) expressive range – the extreme ends of their physical capabilities and the extreme subtleties of small changes.