Reproduce “Designed Vocalizations Dataset: Sound-Designed Human and Animal Voices for Non-human Voice Conversion” 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
Seolhee Lee, Minsu Kang, Yangsun Lee, Woosun Min, Choonghyeon Lee, Namhyun Cho
The authors introduce the Designed Vocalizations Dataset, a benchmark for human-to-non-human voice conversion. Systems change a raw vocalization’s timbre to match a designed reference while preserving its content. The dataset provides non-parallel training audio and paired tests with seen and held-out sources and presets. A baseline scores 0.667 when both are seen and 0.610 when both are unseen on timbre similarity.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.