Reproduce “RIVET: Robust Idempotent Voice Attribute Editing” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Dareen Alharthi, Bhuvan Koduru, Rita Singh, Bhiksha Raj
RIVET edits age and gender in speech while aiming to preserve speaker identity. It adds an idempotency objective that penalizes differences between speech representations before and after reconstruction, supporting stability under noisy labels. In EARS experiments it generally retains higher identity similarity as synthetic label noise rises, though attribute-accuracy trends have exceptions; evaluation is limited to age and gender.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.