Title

U2VC: One-Shot Voice Conversion using Two-Level Nested U-Structure

Description

This page is for audio demo samples of our works. The demo samples show that our proposed approach improves the naturalness of the converted whlile preserving the similarity between the converted speech and the target speech. It has to be noted that the model is only trained with VCTK corpus. All the speakers speak English in this corpus.

Demo

Seen-to-seen conversion

Uneen-to-unseen conversion

Cross-lingual conversion

VCC2020 corpus is used for cross-lingual conversion, Mandarin is mainly used for the evaluation because the raters are (Mandarin-English) bilingual speakers. Other language samples are also shown here.