Acoustic fingerprint · audio
Do these recordings have the same audio?
An acoustic fingerprint is a compact code for what a recording sounds like, not what its bytes are, so it survives re-encoding, volume changes and a resave. Drop in one file to fingerprint it, or two to check whether they are the same audio even after the file has been changed.
Runs entirely in your browser. Your files are never uploaded or stored.
Analysing…
Where they line up
matching different / no overlap · . The bar spans recording A from start to finish; green is where recording B carries the same audio (brighter = closer), dark is where it does not. Two unrelated tracks stay dark end to end.
What an acoustic fingerprint is
A file hash like SHA-256 changes completely if a single byte changes, so it only recognises an exact copy. An acoustic fingerprint is the opposite: it listens to the sound itself. This tool uses the method of Haitsma and Kalker, which divides the audio into short overlapping frames, measures the energy in a band of frequencies, and records, for each frame, only whether that energy rose or fell compared with the neighbouring band and the previous frame. The result is a stream of compact codes that stays almost identical when the audio is re-compressed, made louder or quieter, or saved in a different format, because those changes barely affect which bands are rising and falling.
What the match tells you
With two files the tool slides one fingerprint against the other to find the best time alignment, then measures how much of the track actually matches: the share of recording A where the two carry the same audio. A high share means the same recording, re-encoded or resaved; a partial share means they share a section, a sample or a clip; a share near zero means different recordings. It insists on a long overlap before calling anything a match, so a brief coincidental similarity between two unrelated tracks is not mistaken for one. The match ribbon under the result shows this along recording A from start to finish: green where recording B carries the same audio, dark where it does not, so a shared section, a trim or an edit is visible at a glance.
Uses and limits
This is the check for "is this the same recording that was published elsewhere," or "is this so-called original just a re-export of that one." It does not tell you which came first, and it is not built to identify a track against a commercial catalogue. Very short clips give it too little to align, and a cover version or a live re-performance is a different recording and will not match. To look at a single file's history instead, pair it with the spectrogram and the metadata reader.
Sources
- Haitsma, J., Kalker, T. (2002). A Highly Robust Audio Fingerprinting System. International Conference on Music Information Retrieval (ISMIR) 2002.