Abstract
Stem retrieval, the task of matching missing stems to a given audio submix, is a key challenge currently limited by models that discard temporal information. We introduce PHALAR, a contrastive framework achieving a relative accuracy increase of up to approximately 70% over the state of the art while requiring less than 50% of the parameters and a 7x training speedup. By utilizing a Learned Spectral Pooling layer and a complex-valued head, PHALAR enforces pitch-equivariant and phase-equivariant biases. PHALAR establishes new retrieval state-of-the-art across MoisesDB, Slakh, and ChocoChorales, correlating significantly higher with human coherence judgment than semantic baselines. Finally, zero-shot beat tracking and linear chord probing confirm that PHALAR captures robust musical structures beyond the retrieval task.
Michele Mancusi
Senior Applied Scientist, PhD
Senior Applied Scientist at Music.AI (Moises), developing and optimizing autoencoders and generative models for speech, music, and general audio enhancement