Sing the Note

About the Sing the Note converter

A note plays. You sing it back. It scores you out of a hundred and tells you whether you were sharp or flat.

It is a game, and it is also the most direct answer available to a question a lot of people quietly want answered: can I actually sing in tune? Most people are considerably better than they assume, and almost everybody improves quickly, because pitch matching is a trained skill rather than something you either have or do not.

What the points mean

A hundred points is dead on the note. Nothing at all is half a semitone out, which is halfway to the next note along — past that you are not singing that note any more, so there is nothing left to score. The scale in between is even, so ninety points is twice as close as eighty.

The scale is musical rather than a count of hertz, and that matters more than it sounds. Ten hertz flat is inaudible at the top of a soprano range and most of a semitone at the bottom of a bass range, so scoring by hertz would hand low voices an easy game and high voices an impossible one. Points are worked out in proportion instead, so the same miss costs the same wherever you sing it. Ninety and above is closer than anyone can hear as a difference; fifty is where an untrained listener starts to wince.

The note has to be held still

This is the part that makes the score mean anything. Sliding slowly up to a pitch passes through the correct frequency somewhere along the way, so any scoring that took your closest moment would give a perfect mark to a scoop that never actually landed. Singers do this constantly and mostly without noticing.

So the pitch has to sit still for about half a second before it counts, and the answer taken is the middle reading of that stretch rather than the best one. A slide that travels steadily through the note is refused outright however slowly it moves, and a single wobbly frame cannot drag an otherwise good note. Vibrato is fine — it swings either side and comes back, which is exactly what a scoop does not do.

How the pitch is found

By the YIN method, which asks at what delay the waveform most closely resembles itself, rather than by looking for the loudest frequency in a spectrum. For voices the distinction is not academic: the strongest component of a sung note is very often a harmonic rather than the fundamental, so peak-picking reports a confidently wrong answer, usually exactly an octave high.

Frames that are not periodic enough to have a pitch at all — silence, breath, most consonants — return nothing rather than a guess, which is why humming works better than singing words.

Everyone sings in their own range

The first thing it asks for is any note that feels comfortable, and everything afterwards is pitched around that. Asking everybody for the same notes would measure the wrong thing entirely: a bass cannot reach C5 and a soprano cannot produce a low E, and neither failure says anything at all about their ear.

The daily challenge is stored as a sequence of intervals rather than as fixed notes, for the same reason. Two people singing in completely different octaves attempt the same run of leaps in the same order, so their scores are comparable. The sequence comes from the date itself, so everyone playing on a given day gets the same eight notes without anything being fetched from a server.

Why the tone always stops first

There is no mode where the reference note keeps sounding while you sing. A microphone in the same room as a speaker hears that tone perfectly clearly, and would happily score it as a flawless answer — you would be marking the speaker, not the singer.

So the tone plays, stops, and only then does the listening start, with anything captured during the tone discarded. You can replay the note as often as you like; you simply cannot sing over it.

Tone deafness is rarer than people think

Genuine congenital amusia — an actual inability to perceive pitch differences — affects something like four per cent of people. Almost everyone who says they are tone deaf is not; they are untrained, or they cannot yet hear their own voice accurately while producing it, which is a different and very trainable problem.

A score separates those two cases usefully. If you can hear that you are wrong but land flat anyway, that is production, and it improves fast. If you genuinely cannot tell, that is worth knowing too.

As ear training goes this is the narrowest possible exercise — one note at a time, with a number attached — and that is deliberate. Broader ear training asks you to name intervals and chords, which is a different skill and a harder one to start with; matching a single pitch is the foundation everything else is built on, and it is the one you can measure honestly.

Frequently asked questions