Skip to content
Go to file

Latest commit


Git stats


Failed to load latest commit information.
Latest commit message
Commit time

Finger Smudge

An art project producing 🎼 music for 💻 machines. Firstly exploring the digital fingerprinting of audio music. As used in sites like YouTube & SoundCloud to detect copyright infringment.


Finger smudged


Using an open source audio fingerprinter Chromaprint to investigate forward and reverse generation of fingerprints.

Fingerprint Images

Clojure FFT Example. Example spectro

Chromaprint FFT Image representation Chroma FFT Internal Image

  • Y axis (Energy) -> A A# B C C# D D# E F F# G G#
  • X axis (Time) - Is is fixed or dynamic...


  1. FFT transform of audio (sampling rate 11025 Hz, frame size is 4096 (0.371 s) with 2/3 overlap.)
  2. Frequencies => Musical notes (not octaves). Chroma features.
  3. 16x12 pixel window moving across image one pixel at a time
  4. Apply 16 filteres that capture intensity dirrerences across musical notes && time.
  5. Filters "calculate the sum of specific areas of the grayscale subimage and then compare the two sums."
  6. Quantize the real number with 3 coeeficents (learnt).
  7. 16 filters and each can produce an integer that can be encoded into 2 bits (using the Gray code), so if you combine all the results, you get a 32-bit integer.

Generative Shazaming

Turn Shazam on, run a generative piece of music until a match is found (We screen-scrap to detect a Shazam match notification). Desktop matcher seems more accurate + requires more sampling time than mobile so we only use this:

Avoid any noise by redirecting audio to SoundFlower and set input to SoundFlower. Perfect siganl :)

Turns out its pretty easy to trick Shazam. To make it more interesting we consider not only a match but the match with the highest number of previous Shazam examples (hence more data and Shazam should be more accurate).


  • Chord progressions seem to dominate detection.
  • Unsuprisingly pentatonic scale tends to do well.

Listen to the Audio Experiments



Copyright © 2016-present Joseph Wilk

Distributed under the Eclipse Public License either version 1.0 or (at your option) any later version.


Music for Machines. Exploring the smudges of fingerprinting digital music.





No releases published
You can’t perform that action at this time.