MAE-VC: Voice Conversion Using Learnable Similarity-Guided Masked Autoencoder
Compared with other methods
VCTK
F2F, Our dispute is with the company, not the travelling public.
F2M, I feel we might now have the foundations to go forward.
M2F, She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.
M2M, The actual primary rainbow observed is said to be the effect of super-imposition of a number of bows.
Source
Target
ADAINVC
AGAINVC
MediumVC
FragmentVC
MAE-VC
VCC2020
F2F, Their solution requires development of the human capacity for social interest.
F2M, Income rose zero point four percent last month.
M2F, There is a connection between all of this.
M2M, The problems I mentioned are not insurmountable.
Source
Target
ADAINVC
AGAINVC
MediumVC
FragmentVC
MAE-VC
LibriSpeech
F2F, Turning the first corner he came to, and the next and the next, he stepped into a mews, the court of which seemed empty, and slipped behind the gate.
F2M, In India it is usually made with chicken, but any kind of meat does nicely.
M2F, All this against which he now inveighed so bitterly was but a necessary stage in the shaping of his odd destiny.
M2M, And the subordinate care of justice and the revenue was delegated to seven consulars, three correctors, and five presidents, who governed the fifteen regions of Italy according to the principles, and even the forms, of Roman jurisprudence.
Source
Target
ADAINVC
AGAINVC
MediumVC
FragmentVC
MAE-VC
Aishell-3
F2F, 锁链叮叮铛铛的掉在石头地板上。
F2M, 来一首那些花儿。
M2F, 他正在开发监测用户在开车时是否在打盹儿的应用。
M2M, 此次个人住房公积金存贷款利率相应调整。
Source
Target
ADAINVC
AGAINVC
MediumVC
FragmentVC
MAE-VC
Evaluation on mask ratio and FSC
VCTK
F2F, Our dispute is with the company, not the travelling public.
F2M, I feel we might now have the foundations to go forward.
M2F, She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.
M2M, The actual primary rainbow observed is said to be the effect of super-imposition of a number of bows.
Source
Target
Base_4c_16b_0r_w/o FSC
Base_4c_16b_50r_w/o FSC
Base_4c_16b_90r_w/o FSC
Base_4c_16b_0r_FSC
Base_4c_16b_50r_FSC
Base_4c_16b_90r_FSC
Evaluation on bottleneck
VCTK
F2F, Our dispute is with the company, not the travelling public.
F2M, I feel we might now have the foundations to go forward.
M2F, She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.
M2M, The actual primary rainbow observed is said to be the effect of super-imposition of a number of bows.
Source
Target
Base_4c_2b_50r
Base_4c_4b_50r
Base_4c_8b_50r
Base_4c_16b_50r
Base_4c_36b_50r
Evaluation on cascaded number
VCTK
F2F, Our dispute is with the company, not the travelling public.
F2M, I feel we might now have the foundations to go forward.
M2F, She can scoop these things into three red bags, and we will go meet her Wednesday at the train station.
M2M, The actual primary rainbow observed is said to be the effect of super-imposition of a number of bows.