What are cents in singing?

The unit

A semitone — the gap between two adjacent keys on a piano — is divided into 100 cents. An octave is therefore 1200 cents. The division is logarithmic, so a cent is the same musical distance everywhere in the range even though the frequency gap it represents grows as pitch rises.

This is what makes pitch error comparable. Twenty cents flat on a low note and twenty cents flat on a high note are the same error, even though one is a difference of a couple of hertz and the other of nearly ten.

How many cents a listener can hear

On a sustained note, most listeners begin to hear an error somewhere around 20 to 25 cents, and trained musicians a good deal earlier. On short, fast-moving notes the tolerance is much wider — the ear has less to work with and fills in the rest.

Under about 10 cents is inside the range in which expressive singing legitimately moves. A vibrato swings considerably wider than that and does not sound out of tune at all, because it is centred on the target.

How it is used in a report

Pitch accuracy is scored from the per-note deviation in cents across the take. A separate reading, the mean bias, averages the signed deviation: if it comes out at minus 15, you were 15 cents flat on average, across everything.

The bias is the more actionable of the two. Scattered errors in both directions average to roughly zero and mean something different from a consistent lean, which is why sentivoice opens a flat or sharp problem on a mean bias past 8 cents rather than on the raw error.

Common questions