Software & Research
github – for more information, demo videos, code, etc., get in touch.
Software:
- Music Information Retrieval Tools for SuperCollider (2019) (github)
- Neural Network in SuperCollider (client-side) (2019) (github) (performance)
- Live Laptop Improvisation & Performance (2012-present)
- Poisson Disc sampling in n-dimensional space (2019) (github)
- Linear assignment algorithms ported to SuperCollider (2019) (Munkres & Auction) (github)
- TSNE port for SuperCollider (2019) (github)
- Non-real time FFT analysis (and inverse) in SuperCollider (client-side) (github)
- Live video and audio sampler (2017) (performance)
- Voice Modulation for Eric F. Avery's production of The Life and Death of Eric F. Avery (2016) (github)
- LFO / Arpeggiator / Gate / Trigger for Endorphin.es Shuttle Control (2016) (github)
- Microtonal Keyboard (2016) (github)
Presentations / Papers:
Non-negative Matrix Factorization for Spatial Audio (2020)
Due to COVID-19 the 2020 Spatial Music Workshop in the Cube at Virginia Tech was cancelled, but the organizers invited alumni to give talks about some aspect of their work with spatial audio. I presented this algorithm which I find very useful for spatializing audio in interesting ways.
slides
Due to COVID-19 the 2020 Spatial Music Workshop in the Cube at Virginia Tech was cancelled, but the organizers invited alumni to give talks about some aspect of their work with spatial audio. I presented this algorithm which I find very useful for spatializing audio in interesting ways.
slides
Interference Patterns: analysis of interacting feedbacks in hollow (2020)
This presentation analyzes the feedback system of my piece, hollow, which uses three large PVC tubes to create feedback at the resonant frequencies of the tubes. Through filtering, delay line modulation and serial feedback routing, various emergent sonic properties arise. Analysis of the resulting sounds provides some insight into the behaviors of the system.
slides
This presentation analyzes the feedback system of my piece, hollow, which uses three large PVC tubes to create feedback at the resonant frequencies of the tubes. Through filtering, delay line modulation and serial feedback routing, various emergent sonic properties arise. Analysis of the resulting sounds provides some insight into the behaviors of the system.
slides
Polynomial Functions in Žuraj's Changeover (2019)
A mathematical analysis of Vito Žuraj's orchestral work Changeover. Knowing that Žuraj composes using custom made computer-aided composition tools, this analysis reverse engineers some of the equations and algorithms that he may have used. A generative example using Žuraj's methods is included.
paper
A mathematical analysis of Vito Žuraj's orchestral work Changeover. Knowing that Žuraj composes using custom made computer-aided composition tools, this analysis reverse engineers some of the equations and algorithms that he may have used. A generative example using Žuraj's methods is included.
paper
|
Preserving User-Defined Expression through Dimensionality Reduction (2019)
This is a talk a I gave at the FluCoMa Plenary Session at CeReNeM at the University of Huddersfield in the UK. It demonstrates various machine learning algorithms implemented in my improvisation software and how I use those algorithms to explore new modes of expressivity. video of talk slides |
|
Machine Learning Applications for Live Computer Music Performance (2019)
Presentation at the University of Chicago Digital Media Workshop. This presentation demonstrates three uses of machine learning in live computer music performance: (1) using a neural network to classify no-input mixer timbres for light control, (2) a frequency modulation synthesizer that predictions synthesis parameters based on novel incoming spectra, and (3) a TSNE based dimensionality reduction system for low-dimensional control of synthesizers with high-dimensional parameters spaces.
slides
Approaches to Live Performance and Composition with Machine Learning and Music Information Retrieval Analysis (2019)
This presentation offers three creative uses of machine learning: (1) using audio descriptor analysis and machine learning to organize grains of audio into a performable two dimensional space, (2) using a neural network to classify no-input mixer timbres for light control, and (3) using a traveling salesperson pathfinding algorithm to re-organize audio grains into a new sequence.
slides