Spotify, Pandora, Apple and Google Bots are selecting from 100 million songs to create over 7 billion listener playlists automatically. Traktomizer helps artists understand how to make their music algorithmically choice-worthy and therefore organically profitable.

Why Traktomizer

If a song is regularly recommended by an Apple Music, Spotify, Deezer, etc, Playlist-Generator-Algorithm, or simply returned with the majority of Search results, one couldn’t argue that the ‘Music Recommendation-bots’ have taken a shining to it.

 

The way recommendation music platform algorithms listen can be expressed as: comparative audio analysis of a pre-release song with the audio properties of top recommended songs (e.g. the top 50 in the same Genre), tallies up to standard ranges of deviations from the calibrations of those algorithms. Therefore if the pre-release song’s audio properties were optimized for minimum deviation before the song is published, it is predisposed to being chosen for recommendation and play-listing for ever. Whereas failure to make these checks and tweaks, in the extreme case, would have the effect of rendering the song invisible to the recommendation-bots for ever.

Traktomizer’s significance as the pre-release audio litmus test comes from its ability to precisely report every important deviation and individually rank and suggest the crucial audio tweaks requiring the least amount of adjustment for the greatest overall gain.

With a GUI that lights up one’s day; this program is for producers, sound engineers, indie producers, home studio enthusiasts, promoters and label managers.


Traktomizer converts stream-meta-data into comprehension and suggestions. When these optimization actions have been taken, increased playlist inclusion, increased recommendations, increased plays, wider audience reach and maximized song longevity, are logical and reasonable expectations.  

ANECDOTE: during AUDIO Analysis of 23 songs, Traktomizer:

  • used all of its 200 000 lines of code
  • received over 270 000 words altogether
  • received nearly 3 000 000 characters altogether
  • decided which songs to benchmark from a total of between 180 to 340 songs
  • retrieved over 12800 Terms, Tags, Keywords and Genre Names from around 50 additional songs
  • made around 6700 calculations
  • generated over 50 charts (Audio Analysis and Social Media Metrics)
  • produced nearly 1 000 000 xml-words – the constructs of the SVG charts
  • created and persisted 17 byte-arrays for instant data call-back (e.g. for OpenGL plotting)
  • run up to 36 concurrent threads (ave. 24)
  • made over 5 000 000 system calls
  • consumed less than 1.12 Gb of System Memory at any one time (ave. 780k)
  • used an average of 4.6 CPU% or less (2.8Ghz single processor)
  • compared 14 critical audio properties with each benchmark song and the pool as a whole
  • identified the song’s strongest and weakest audio properties, sorted and ranked them, and calculated the top and bottom three (3).
  • reanalyzed the entire song analysis collection (all previously analyzed songs) and recalculated the top 5 most recommendable songs, strongest three (3) production skills, weakest three (3) productions skills, generated a detailed PDF Summary, a Dashboard, 7 additional Summary SVG focus charts, a PDF Trace report, and a Top 15 Genres PDF report.

Astonishingly, this all took place in well under 5 minutes – and amazingly, all from a single mouse-click!