Abstract
This paper introduces GlOttal‑flow LPC Filter (GOLF), a novel method for singing voice synthesis (SVS) that exploits the physical characteristics of the human voice using differentiable digital signal processing. GOLF employs a glottal model as the harmonic source and LPC filters to simulate the vocal tract, resulting in an interpretable and efficient synthesis approach. We show it is competitive with state‑of‑the‑art singing voice vocoders, requiring fewer synthesis parameters and less memory to train, and runs an order of magnitude faster for inference. Additionally, we demonstrate that GOLF implicitly learns to model the phase components and formants of the human voice, having the potential to control and analyse singing voices in a differentiable manner. Our results highlight the effectiveness of incorporating the physical properties of the voice production mechanism into SVS and underscore the advantages of signal‑processing‑based approaches, which offer greater interpretability and efficiency in synthesis.