Cents, semitones and what listeners forgive

Every app that scores your singing has a number hidden inside it: how far off you are allowed to be before it says you missed. Here is where that number comes from, and why there is no single right answer.

What is a cent?

A cent is one hundredth of a semitone. A semitone is the distance between two adjacent keys on a piano, black or white — so 100 cents is exactly one step, and 1200 cents is an octave.

The scale is logarithmic, which is the useful part: 100 cents is the same musical distance whether you are low in your range or high in it, even though the difference in hertz is several times larger at the top. That is why pitch software talks in cents rather than hertz. Fifteen hertz means nothing on its own; fifteen cents means "a fifteenth of a step, nobody will hear it".

Some rough anchors:

• 1200 cents — an octave.
• 100 cents — a semitone, the next note along.
• 50 cents — halfway to the next note. Past this point, which note you meant becomes a matter of opinion.
• 10–20 cents — the region where trained ears start to disagree with each other.

How far off can you be before someone notices?

Further than you would think, and further with a voice than with an instrument.

Hutchins, Roquet and Peretz (2012) named this the vocal generosity effect: listeners judging a sung note tolerate roughly twice the deviation they tolerate from an instrument. Where an instrument is called out of tune at around 20 to 30 cents, a voice gets to about 50 before listeners reach the same verdict.

So the folk wisdom is roughly right. A voice that is a little sharp still sounds like singing; a violin that is a little sharp sounds broken. The tolerance is in the listener, not in the signal.

Then why do apps use 50 cents?

Because 50 cents is where the question changes. Below half a semitone, you are singing the intended note, slightly off. Above it, you are closer to the neighbouring note than to the one you meant, and the listener has to guess what you were aiming at.

That makes 50 cents a natural ceiling for anything involving a melody — but a ceiling, not a target. In NoteNail's own exercises the tolerance narrows as you improve: 50 cents to start with, then 30, then 20, then 10. The last of those is deliberately tighter than any listener needs, because the point at that stage is no longer "does this sound acceptable" but "can you put your voice exactly where you meant to".

Does everyone hear it the same way?

No, and this is the part that makes a single universal threshold impossible.

Larrouy-Maestri and colleagues (2019) built the Mistuning Perception Test precisely to measure this, and the spread between individuals is wide: the deviation one listener finds obvious is inaudible to the next. There is no population-wide line where "in tune" stops.

Context matters as well. A 2018 study in Music & Science with the fitting title "I know it when I hear it" found that the same deviation is judged differently depending on the surrounding music, the instrument and the listener. "Out of tune" is a judgement, not a measurement — which is exactly why an app should show you the measurement and let you keep the judgement.

Why one note is not a song

A tolerance in cents describes a single note in isolation. Singing is not made of isolated notes.

Recall the octave leap in "Happy Birthday": Nichols, Hua and Wang (2023) found that 34 of 37 adults missed it, and that the intervals sung earlier predicted 65 per cent of the variance in how far off they landed. Errors accumulate along a phrase. A scoring rule that grades each note against the score sheet and stops there will tell you that you drifted, but not that the drift started four notes earlier.

That is the argument for comparing a whole phrase against a whole phrase, rather than one note against one number.

The short version

• Cents are a logarithmic ruler: 100 to a semitone, 1200 to an octave.
• Listeners forgive a voice about twice what they forgive an instrument — roughly 50 cents against 20 to 30.
• Individual thresholds vary widely, so a single "in tune" line does not exist.
• Half a semitone is a sensible ceiling for a melody, and a poor target for a trainer.

If you are here because you were told you cannot sing, start with why you sing off-key instead — the short answer is that your ears are probably fine.