2013 → 2022 · SMC 2022 · ISMIR 2022
Audio Metaphor
A pipeline of computational tools for generating soundscapes from a sentence — search, segmentation, classification, and mixing modules that run independently or in concert. The 2022 rebuild grew the audio database from around 4,000 files to over 395,000, replaced hand-tuned classifiers with trained regression models, and moved the whole tool online: sketch a valence/arousal curve over time and it composes a soundscape, ready to export as an OMF file for a DAW.
Project site
Read the paper
2019–2020 · ISEA
Aeon
A real-time performance instrument for solo audiovisual improvisation. Live animation "voicings" built in Max/MSP-Jitter meet a multi-agent drone system in SuperCollider that listens to the performer and negotiates its own musical elaborations — no fixed script, no two performances alike. Performed live by Miles Thorogood (percussion/electronics) and Aleksandra Dulic (live animation).
Read the paper
2026 · Preprint
Glitch Audio Plug-In
Packet-loss-concealment algorithms are built to invisibly repair dropped network audio; deliberately pushed past the point of transparency, that same repair mechanism becomes a glitch-synthesis engine. Ships as Dropout, and as the Packet Loss module bolted onto Codec Crusher +.
2025 · IEEE Access
Audio Packet-Loss Concealment
A closed-form continuous-time neural network — a "liquid" architecture loosely modeled on a microscopic worm's nervous system — predicts what a dropped network-audio packet should have sounded like. Listeners rated its repairs more convincing than a classic linear model, even though the simpler model scored better on raw error metrics. Streams live inside JamLink.
2026 · HCII · Sustainability (MDPI)
BeaverWorld
A place-based 3D simulation for climate education, built around Indigenous ecological knowledge and beaver-driven habitat restoration — borrowing its challenge structure from Soulslike games on the premise that hard-won struggle teaches an ecological lesson better than information delivery alone.
Read the paper
2026 · Virtual Reality (Springer)
User Preference for Immersive Sound in XR
A listener-first study of spatial audio in XR headsets: preferences are drawn from participants' own reported experience rather than a predetermined technical benchmark, inverting the usual order of spatial-audio evaluation research.
Read the paper
2026 · Submitted to SMC
Spectral Manifold Splatting
Repurposes 3D Gaussian Splatting — the go-to method for photorealistic scene reconstruction — entirely for sound: 64 "audio splats" scattered in a Fibonacci spiral drive phase-vocoder synthesis, producing spatial audio participants rated as more enveloping than a mono reference.
2026 · Submitted to TEI
Tactile Game Level Creator
Snap a photo of objects arranged on a desk and a computer-vision pipeline (FastSAM) turns the layout into a playable Unity platformer in seconds — no code, no menus. First-time designers were playtesting within five minutes of sitting down.
2024 · SIGUL
Language Tech & Indigenous Methodologies
Grew out of a speech-to-text aligner for nsyilxcn, a Salishan language with fewer than 81 fluent speakers. The paper turns the build process itself into a decolonizing framework for language-technology research with Indigenous communities, built on Respect, Relevance, Reciprocity, and Responsibility.
Read the paper
2018–2023 · ISEA · Sustainability (MDPI)
Waterways
A four-year, community-engaged museum exhibition treats interactive visualization itself as the research method — combining Indigenous water knowledge with ecological data to explore the Okanagan waterscape with the public, not just report on it afterward.
Read the paper
2023 · ISEA
Light Up Kelowna
The infrastructure behind a recurring urban-screen program: rear-projected frosted-vinyl windows standing in for expensive LED walls, and networked conductor/player software that keeps a rotating lineup of artists running unattended, night after night.
2022 · ISEA
Networked A/V Authoring System
A general-purpose authoring layer for multi-channel installation art: one authored media library can be reassembled across a different number of screens or speakers without re-editing, powering works like Journey of a Pod and Light Up Kelowna.
Read the paper
2021 · Book chapter, Routledge
A Sound Design Creative-AI Methodology
Distills Audio Metaphor, the soundscape affect classifiers, and Impress into a general framework for how creative-AI sound design research should be framed, evaluated, and situated within existing practice — a field guide rather than another single system.
Read the chapter
2019 · ISEA
Approaching Sustainability
Argues that generative art's own feedback loops and emergent behaviour are a working model of complex-systems thinking, not just a metaphor for it — the theoretical blueprint that later underpinned the Waterways exhibition.
Read the paper
2019 · Book chapter, Routledge
Soundscape Generation Systems
A survey chapter mapping the field of procedural soundscape generation — from rule-based synthesis to database retrieval to ML-based composition — consolidating scattered approaches into one reference for newcomers and practitioners.
Read the chapter
2019 · Book chapter, Routledge
Soundscape Online Databases
Turns critical attention to the online audio databases — Freesound and similar — that generative soundscape systems quietly depend on, examining the metadata quality, tagging, and licensing issues that shape what those systems can reliably produce.
Read the chapter
2013–2019 · NIME · Audio Mostly · JAES · SMC · JNMR
Soundscape Affect Classification
Seven studies spanning a decade — from Impress' real-time affect prediction to a full valence-arousal model — synthesized into the classification framework that underlies affect labels in Audio Metaphor and later soundscape systems.
Read the paper
