ICMC 2022 Workshop:

Teaching Critical Machine Listening and Machine Learning with FluCoMa


This workshop will present the foundations of a curriculum developed by the Fluid Corpus Manipulation team, designed to teach critical machine listening and machine learning within a Max, SuperCollider, or Pure Data class.

The Premise

The FluCoMa project enables techno-fluent artists to tackle programmatic data mining within their mastered creative coding environment. A great deal of music making with computers involves processing, combining and playing with recorded sound. As storage gets cheaper and collections of sounds get larger, they become harder to manage and explore. Advances in signal processing and machine learning show promise for working fluently with these collections, but their musical potential has so far been hard to explore due to a range of factors such as availability, legibility, and quality of documentation. The FluCoMa project aims to bring these tools to techno-fluent artists with accompanying documentation and pedagogical resources, making these technologies approachable, learnable, teachable, and inspiring.

We believe that questions of musical potential are not, and cannot be, solely technical. Moreover, as demonstrated by repeated controversial uses of machine learning within industry, culture, and government, individuals’ understanding of and fluency with these technologies and their limits are increasingly urgent to society. Artistic uses of machine learning and data science offer entry points and practical applications for building understanding and intuition about what these technologies can and cannot achieve and how they can succeed as well as go wrong.

We have found, via our series of commissions, user support, and workshops, that to enable artists to engage with these questions, a critical perspective on the technology's role in creative practice needs to be taught alongside an understanding of what these technologies can do, all at the right level of complexity and granularity.

Finally, we think that these questions and affordances should be introduced early into the process of making artists techno-fluent. We have therefore devised an inviting approach that is designed to integrate into creative coding courses.

The proposed workshop

The day-long workshop will provide participants with first-hand experience of the FluCoMa tools, their supporting pedagogical materials, and a sample curriculum, all of which will enable a discussion around workshop participants' experiences and practice with similar technologies in pedagogical context.

Thus, the workshop will:

  1. Present the FluCoMa project’s agenda and general findings (1h);
  2. Give participants a hands-on introduction to the Fluid Corpus Manipulation toolkit, its framework and affordances, and their possible usefulness in the classroom. (3h) We will cover various FluCoMa workflows and tutorials, including:
    • real-time and non-real-time audio descriptor analysis and dataset creation
    • data science and machine learning for sonic comparisons and pattern finding
    • sonic hybridizations and more creative affordances of programmatic data mining
  3. Present the FluCoMa curriculum (1.5h), including
    • the toolset's help files/documentation/learn website/example code framework
    • the video tutorials and their place as entry points with key concepts
    • the “Made-With FluCoMa" series highlighting artists and works
    • the other toolkits in the wider ecosystem with similar pedagogical questions
  4. Attendees will be invited to submit a 10-15 minutes-long presentation on how they tackle similar teaching aims, both from a technological and pedagogical perspective. (1.5h)
  5. Open discussion: in the light of the workshop contents and attendees’ presentations, we will hold an in-depth group discussion on the larger questions of pedagogical and artistic affordances of these technologies, (1h) and hopefully to start a community of sharing practices

Who is it for

Teachers of Max / SuperCollider / PureData, from within or outside academia. They may already teach, or be interested in including, machine listening related concepts. Participants will need a computer with one of these environments and the FluCoMa package installed

Number of places

  • 16 participants in person.
  • listening-only should be possible online (no coding support)

Outputs & Outcomes

  • basic understanding of the FluCoMa toolset and how it can be integrated into one’s teaching
  • a sample curriculum, mostly ready to adapt for the next academic year as part of their creative coding environment classes
  • a knowledge exchange on teaching machine learning, data science, and their creative affordances to creative musicians
  • seeding a community of practitioners and artists engaging with similar questions


  • this workshop is free
  • online participation is possible but might be difficult because real-time support will not be available.


Pierre Alexandre Tremblay, Owen Green, James Bradbury, Ted Moore. For bio, see the people's page.