Event Synthesis: Algorithmic Music Composition
Explore the geometry of sound and machine learning through creative coding, from generative MIDI to procedural composition for performance, games, and installations.
This new online course is for musicians, programmers and mathematically-inclined artists interested in new ways of thinking about music. The central goal is to present composition through unconventional conceptual lenses drawn from physics, geometry and data science to encourage the playful use of mathematics in art. We will build custom tools for composing electronic music by using PyTorch for machine learning and construct playful user interfaces for audio systems using JavaScript.
Participants interested in developing generative MIDI systems or music software will gain insight into musical representation, control design, and sequencing strategies. Composers will encounter procedural composition techniques and their applications in performance, video game music, and interactive installations. Mathematicians and scientists are invited to experience familiar concepts through the lens of sound and musical language.
Rather than relying on large end-to-end AI models, the workshop emphasizes ML for portable systems: tools that learn from artist input within a single composition session or through carefully shaped reward functions. Topics include intuitive probability modeling, learning-based sequencing, and vector-field approaches that reframe musical events as flow through a continuous, differentiable landscape. Each session will blend conceptual grounding, historical reference, artistic context and live demonstrations.
Course Logistics
Enrollment Deadline: March 31, 2026
Dates:
April 7, 2026
April 14, 2026
April 21, 2026
April 28, 2026
(Every Tuesday in April)
Times:
6 PM – 9 PM PT / 9 PM – 12 AM ET / 1 AM – 4 AM GMT
Location: Online
Cost: $480 for Live Online Access.
Scholarship: We also offer Diversity Scholarships.
Apply by March 31, 2026. Scholarship notifications will be sent within 1 week after the deadline.
Experience Level: Beginner
Prerequisites:
- Participants should have a strong interest in music and basic programming literacy.
- No formal background in mathematics or machine learning is needed. Curiosity to explore abstract ideas is what matters most.
- For hands-on experimentation, access to a DAW capable of receiving MIDI input (such as Ableton Live, Logic, Reaper, or similar) is highly recommended, as participants will work with generative MIDI streams that can be monitored, shaped, and played in real time.
Additional Information:
• No Refunds or Exchanges.
• View our FAQ here.
• Contact [email protected] with any questions.
Workshop Outline
- Session 1: Stochastic Modeling / Embracing Indeterminacy
◦ Markov chains and probabilistic sequencing
◦ From discrete distributions to continuous probability fields
◦ References: Terry Riley / Scrying Patterns and Divination - Session 2: Spectral Decomposition / Geometry of Musical Time
◦ Spectral thinking: Fourier analysis and singular value decomposition revealing form / extracting features
◦ Geometric interpolation of musical trajectories using Gaussian processes
◦ References: Jazz Language / Iannis Xenakis - Session 3: Reinforcement Learning / Games, Strategy and Music
◦ Music generation framed as game design: states, actions, and rewards
◦ Reinforcement learning for event sequencing using small, interactive training loops
◦ Agents learning musical behavior through play
◦ References: John Zorn / Pauline Oliveros - Session 4: Differential Modeling / Musical Weather
◦ Vector fields / Differential equations and phase space as models of parameter motion
◦ Attractors, flow, and emergent musical behavior
◦ Live performance of flow-based, continuous generative systems
◦ References: Laurie Spiegel / J.S. Bach
Educational Goals:
- Present mathematics and machine learning as artistic materials
- Develop geometric, probabilistic and learning-based thinking about musical structure
- Enable participants to experiment hands-on with generative musical systems
- Emphasize small, playable learning systems over large-scale data dependence
- Inspire further exploration on ML embedded in artist workflows
About Technologies:
We’ll use a combination of Python / PyTorch for signal generation, JavaScript for visualization and Ableton Live for audio monitoring / processing
Become a Gray Area Member for early access to our workshops and courses, and up to 20% off.
Gray Area Members enjoy extraordinary access to our interdisciplinary arts & education programs, while supporting our mission to catalyze creative action for social transformation. Get free admission, exclusive content, and more!
