Media Fingerprinting
< prev next >

Fingerprinting music, as popularized by Shazam and other services, helps people locate a specific song from a big library using just some seconds of the audio of that song.

Fingerprinting works by extracting meaningful features out of the music (average pitch, tonality, tempo, etc) for every second of the song, and storing this in a database. Then, when the unknown song is tested, the same kinds of features are extracted and a fuzzy search is performed in the database.

Base 10 Labs’ has implemented several fingerprinting solutions for their costumers to work in real-time, distributed systems, and even to help authenticate the legal status of media items.