De-Mix Audio using FluCoMa in Max

Membership plan: Going Deeper | Topics: Sound Design

Course overview

The first time I heard an example of what this algorithm can do, I was super impressed. It broke apart a mono drum loop into three tracks. One had only the snare sounds, one had only the kick drum sounds, and one had only the cymbal sounds. This sparked a lot of thinking about how I could use this for creativity. Then I found out it also can break apart and identify specific sounds in real-time audio. In this workshop I'll show you how these tools work in the FluCoMa toolkit and some of the ways I've been using it in my own work.


It's suggested that you have completed the FREE course: Using Machine Learning Creatively via FluCoMa in Max

What you'll learn

  • De-mix recorded audio into sound objects
  • Deconstruct real-time audio streams
  • Identify important sounds in real-time audio
  • Seed this process to help the algorithm find the sound objects we're interested in

Who is this course for?

  • Sound Designers searching for new ways of combining sounds
  • Audio Engineers intrigued by de-mixing possibilities
  • Creative artists interested in machine learning algorithms for music making

Course content

  • Max patches, audio files and PDF
  • 1. Introduction
  • 2. Workshop Outline
  • 3. Audio Decomposition with NMF
  • 4. Background on Non-negative Matrix Factorization
  • 5. fluid.bufnmf~ Attributes
  • 6. Playing Activations as Envelopes
  • 7. Decomposing Real-time Audio Streams with fluid.nmffilter~
  • 8. Combining Real-time Activations and Bases
  • 9. More with fluid.nmffilter~
  • 10. Real-time Spectral Matching with fluid.nmfmatch~
  • 11. Creative Examples
  • 12. Individual Experimentation
  • 13. Seeding Bases
  • 14. Seeding Bases for Analyzing Large Files
  • 15. Seeding Activations
  • 16. Over-de-composing a Source Sound
  • 17. NMF for Multi-channel Spatializatio
  • 18. Timbre-transfer with fluid.bufnmfcross~
  • 19. Outro


  • A working PC, Laptop, iMac or MacBook
  • Max software
  • Download the FluCoMa Package from the Package Manager
  • It is suggested to first watch the free course Using Machine Learning Creatively via FluCoMa in Max

Course schedule

Meet your instructor

Ted Moore (he / him) is a composer, improviser, and intermedia artist. He holds a PhD in Music Composition from the University of Chicago and recently served as a Research Fellow in Creative Coding at the University of Huddersfield, investigating the creative affordances of machine learning and data science algorithms as part of the FluCoMa project.‚Äč His work focuses on fusing the sonic, visual, physical, and acoustic aspects of performance and sound, often through the integration of technology.

Notify of

Inline Feedbacks
View all comments
Would love your thoughts, please comment.x