0110.be logo

~ Updates for Panako - an acoustic fingerprinting system

» By Joren on Sunday 11 July 2021

Panako is an acoustic fingerprinting system I developed a couple of years ago. With acoustic fingerprinting systems it is possible to find duplicates in digital music archives and compare meta-data or identify unlabelled audio fragments. In the margins of my post-doc project working with large music archives, I have found the time to update Panako significantly. The updates simplify, improve and speed up Panako.

General content based audio search scheme
Fig. General content based audio search scheme.

The main algorithms are simplified. There is also a reduction of dependencies and a refocus to core functionality. This also simplifies building the software. The retrieval characteristics are improved, mainly thanks to the use of a fine-grained Gabor transform. Also new is the near-exact hashing construct which helps with off-by-one issues when matching time bins. The key-value store used is now LMDB, which speeds up the query performance of Panako significantly. The updates should make Panako stand the test of time somewhat better.


Fig. The top one true positive rate for 20s query fragments. The audio playback is speed modified from 84 to 116% with respect to the indexed reference audio. The original query length is 20s, if it is slowed down by 10% it takes, evidently, 22s. Note the improvement of the 2021 version of Panako (blue) vs the 2014 version (light-gray). As a baseline the standard algorithm (wang 2003) is included as well. For the 2021 Panako algorithm, audio recognition performance suffers (below 80%) when playback speed is changed more than 10%.

A more complete list of updates can be found below and on the Panako GitHub repository:

  • The number of dependencies has been drastically cut by removing support for multiple key-value stores.
  • The key-value store has been changed to a faster and simpler system (from MapDB to LMDB).
  • The SyncSink functionality has been moved to another project (with Panako as dependency).
  • The main algorithms have been replaced with simpler and better working versions:
    • Olaf is a new implementation of the classic Shazam algorithm.
    • The algoritm described in the Panako paper was also replaced. The core ideas are still the same. The main change is the use of a Gabor transform to go from time domain to the spectral domain (previously a constant-q transform was used). The gabor transform is implemented by JGaborator which in turn relies on The Gaborator C++ library via JNI.
  • Folder structure has been simplified.
  • The UI which was mainly used for debugging has been removed.
  • A new set of helper scripts are added in the scripts directory. They help with evaluation, parsing results, checking results, building panako, creating documentation,…
  • Changed the default panako location to ~/.panako, so users can install and use panako more easily (without need for sudo rights)

An interactive CLI session with Panako
Fig: An interactive CLI session with Panako.

Tags Code and UGent