index.md

Audio Demo for "FlowCPCVC: A flow contrastive predictive coding voice conversion system"

---

Any-to-Any for voice conversation.

speakers in test-dataset come from vctk, all speakers are unseen during training

Man-to-female

source	target	FlowCPCVC	VQMIVC

female-to-Man

source	target	FlowCPCVC	VQMIVC

>

female-to-Man or Man-toMan

source	target	FlowCPCVC	VQMIVC

Audio of target come from libritts, which is a dataset different to vctk

---

Speakers of target in libritts. Speakers of source in vctk test-dataset. All speakers are unseen during training. We train the model only with vctk.

source	target	FlowCPCVC	VQMIVC

The converted results where mood swing of source audios are high.

The source audios come from vctk in test-dataset. Audios of target come from libritts which exclude in training.

source	target	FlowCPCVC	VQMIVC

---

Any-to-Many

The source audios come from libritts, the target timbre come from vctk in training dataset.

target timbre example audio

p236	p264	p269	p263	p259	p256

converted audios

source	to_p236	to_p264	to_p269	to_p263	to_p259	to_p256