VQVC-DEMO

Abstract

Voice conversion (VC) is a task that transforms the source speaker’s timbre, accent, and tones in audio into another one’s while preserving the linguistic content. It is still a challenging work, especially in one-shot setting. Auto-encoder-based VCmethods disentangle the speaker and the content in input speechwithout given the speaker’s identity, so these methods can further generalize to unseen speakers. The disentangle capabilityis achieved by vector quantization (VQ), adversarial training, or instance normalization (IN). However, the imperfect disentan-glement may harm the quality of output speech. In this work, to further improve audio quality, we fuse skip-connection mod-ules into an auto-encoder-based VC system. We find that to leverage skip-connection, strong information bottleneck is necessary. The VQ-based method, which quantizes the latent vectors, can serve the purpose. The objective and the subjective evaluations show that the proposed method performs well in both audio naturalness and speaker similarity

DEMO ( Unseen Source and Target, both of them are random sampled, and converted by only one utterance)

DEMO ( Seen Source and Target, both of them are random sampled, and converted by only one utterance)