Is my track uploaded?

No. The file is decoded and analysed by your browser inside the page, so nothing leaves your device and nothing is stored.

How does it decide there is a vocal?

It tracks one pitch line and measures four things: how much of it sits in the 100–1050 Hz vocal band, how often it glides between notes at a sung rate, whether it modulates 3.5–8.5 times a second while held, and how far it drifts from equal temperament. Each measurement and its weight is shown with the result.

Why does it say 'not clear either way'?

Because that is often the honest answer. When too few frames carry a readable pitch, or the measurements do not lean far enough, the tool reports uncertainty instead of guessing. Uncertain does not mean instrumental.

Will it detect rapped or spoken vocals?

Usually not. Rap, spoken word and whispered parts carry little sung pitch contour, so they commonly measure as instrumental. This tool looks for singing, not for the presence of a human voice.

Can an instrument be mistaken for a voice?

Yes. Expressive lead synths, bowed strings, slide guitar and theremin all glide, sit in the vocal range and use vibrato, so they can score highly. That is a limit of pitch-contour analysis, not a bug.

Does it work on a full mix?

Partly. A prominent lead vocal usually reads, but a vocal buried under dense production often does not, because the pitch tracker follows whichever line dominates. Separated stems or exposed sections give a far more reliable read.

Experimental tool

Vocal vs Instrumental Detector

A heuristic read on whether a track carries a sung line, with the pitch measurements behind the answer shown in full.

Your file is decoded and analysed in this browser tab. Nothing is uploaded, nothing is stored.

Your audio

Analysed in this tab. Nothing is uploaded or stored.

Choose an audio file

WAV, MP3, FLAC, AIFF, M4A, OGG — up to 250 MB. Two seconds or longer.

Detail

How this works

Sorting a folder of bounces into vocal takes and instrumentals is dull work, and the filename usually lies. This tool makes a quick estimate from the pitch contour alone: a voice tends to live in a narrow frequency band, slide between notes rather than jump, wobble a few times a second on held notes, and sit slightly off the tuning grid.

Those four measurements are combined at 35%, 30%, 20% and 15%, and all of them are shown with the result, because a verdict you cannot inspect is not worth much. When the evidence is thin, the tool says so instead of picking a side.

It is a heuristic and it has real limits: rapped and spoken parts read as instrumental, expressive lead instruments read as vocal, and a vocal buried in a dense mix is easy to miss. Use it to triage a library quickly, then trust your ears on anything that matters.

FAQ

Frequently asked questions

Toolbox

See all free ANYANO tools

Train your sound. Generate your song.

ANYANO is the bigger version of this: prepare music you own, train a Sound that stays private to your account, then generate from it.