Code coming soon
(Repository created for the ICASSP 2026 submission.
Full training and inference scripts will be released after the paper is published in the proceedings.)
👉 Interactive demo on GitHub Pages
We provide listening comparisons for Reference, Post-render, and DDSP-only.
Listening aligns with Tables 2–3 in the paper:
_Post-render ≈ Reference > DDSP-only.