Designed Vocalizations Dataset: Sound-Designed Human and Animal Voices for Non-human Voice Conversion

Researchers released the Designed Vocalizations Dataset for non-human-style voice conversion.

The dataset pairs human speech and animal vocalizations with professionally effect-processed versions.

It includes standardized seen/unseen test splits and baseline results for reproducible evaluation.

The benchmark fills a gap for robotic, monster, and other designed vocal sounds beyond natural speech.