Publications
Machine learning and signal processing for music: music information retrieval, AI-generated music detection and neural audio synthesis.
2026
How Much AI Is in This Track? Quantifying the Proportion of AI-Generated Stems in Hybrid Music Mixtures
A regression CNN recovers the AI energy ratio of a mix at 0.076 MAE, R² = 0.85.
Binary detectors hit over 99% accuracy separating fully-AI from fully-human tracks, but producers mix an AI drum loop with a synthetic bassline and a human vocal. We reframe detection as regression over a continuous AI energy ratio: the fraction of a track's acoustic energy that comes from AI stems.
Binary detectors break down on such mixtures, and detectability varies sharply by instrument: drums and guitars leak strong neural-codec artifacts, vocals and bass far less.
Assessing AI-Generated Music Detection in Real-World Broadcast Monitoring
BAMM: 40 hours of AI-generated and human-made music from real television recordings.
Detectors that excel on clean audio degrade sharply in real broadcast conditions. We introduce BAMM and evaluate CNN detectors across three tiers of realism: clean foreground music, simulated TV broadcast, and genuine broadcast.
Broadcast-aware training improves robustness, but a substantial gap between laboratory results and practical deployment remains.
2022
From Words to Sound: Neural Audio Synthesis of Guitar Sounds with Timbral Descriptors
A guitar synthesizer steered by vocal commands through timbral descriptors.
A deep-learning instrument that generates guitar sounds from vocal commands, extracting timbral descriptors from spoken input to steer a latent space that is otherwise high-dimensional and semantically opaque. Built within The Sound of AI Open Source Research initiative, with more than 200 contributors.