Interdisciplinary audio researcher at McGill University.
I'm a researcher and engineer working at the intersection of spatial audio, signal processing, machine learning, and psychoacoustics. My work aims to make spatial audio systems smarter, more perceptually accurate, and more human.
My thesis research, SPIRAL: Spatial Impulse Response Archive for Learning, is geared towards developing a large-scale, publicly available dataset of Spatial Room Impulse Responses designed specifically for training deep learning models. By pairing high-resolution acoustic measurements with architectural, visual, and perceptual data, I'm building the foundation for a new generation of perceptually informed virtual acoustic systems.
My path here winds through a lifelong love of music and audio, professional guitar playing, mixing & mastering, and two graduate music technology programs — every step deepening my conviction that the best audio technology is grounded in how humans actually hear.
SPATIAL IMPULSE RESPONSE ARCHIVE FOR LEARNING
Thesis research centered around developing a large-scale, publicly available dataset of Spatial Room Impulse Responses (SRIRs) built with high-resolution ambisonic measurements across a diverse range of acoustic environments — concert halls, studio control rooms, lecture theatres, classrooms, etc. Paired with architectural, visual, and perceptual data for multimodal deep learning applications in spatial audio.
SPIRAL addresses the single most significant bottleneck restricting deep learning's potential in spatial audio: the absence of a robust, standardised SRIR dataset. Measurements are taken with the EM64 spherical microphone array at McGill's Immersive Media Lab, capturing a standardised 15×15 grid (225 B-format recordings per space) and used to train a variety of deep learning models.
MULTIMODAL DEEP LEARNING FOR ACOUSTIC-PERCEPTUAL CROSS-MODAL RETRIEVAL
A hybrid CNN/MLP architecture that learns high-dimensional mappings between measurable acoustic features and human perceptual qualities — immersion, clarity, warmth, envelopment, and more. The full model supports cross-modal queries: input a photograph of a room, a text descriptor like "warm and spacious," or room dimensions, and retrieve a perceptually appropriate spatial impulse response.
INVESTIGATION OF FIRST- AND HIGH-ORDER AMBISONICS FOR AURALIZATION
Investigation of the spatial resolution characteristics of first- and higher-order ambisonic microphones as capturing tools for auralization of real spaces in recording studios equipped with virtual acoustic systems.
// further publications in progress — SPIRAL dataset & SPIRAL-Net architecture papers forthcoming
I'm interested in research collaborations, industry opportunities in XR audio and spatial sound, and conversations about anything signal processing, acoustics, pyschoacoustics, or machine learning.
I'd love to hear from you!