How The Heck Does Shazam Work? (An Interactive Exploration)
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
An interactive deep-dive into how Shazam identifies songs in seconds. Starting from raw audio waveforms, it explains how the Fast Fourier Transform converts sound into spectrograms, how peak landmarks are extracted to form a sparse 'constellation map', and how pairs of peaks generate compact hash fingerprints. The matching process uses an inverted index — looking up hashes rather than scanning songs — enabling near-instant identification across millions of tracks. The piece also covers why Shazam struggles with humming, and contrasts server-side matching with on-device approaches like Google's Pixel 'Now Playing' feature. Based on Avery Wang's original 2003 paper.
Table of contents
Reverse Engineering SoundLess Is MoreConnecting the DotsFinding The Perfect MatchMore Modern ApproachesThank you!Sort: