Skip to content

Releases: PlayVoice/whisper-vits-svc

HiFTNet Release

05 Oct 10:27
5cc5fbb
Compare
Choose a tag to compare

HiFTNet is much faster than BigVGAN

BigVGAN Release

11 Jul 06:35
4024615
Compare
Choose a tag to compare

Mix Encoder : hubert soft & whisper large v2 audio encoder's output of 24L

dependency

28 Jun 02:34
3baf64a
Compare
Choose a tag to compare
about mix

final model architecture of hifigan

18 May 02:58
Compare
Choose a tag to compare

code: so-vits-svc-5.0-hifigan-code.zip
pretrain: sovits5.0_main_1500.pth

6G memory GPU can be used to trained

final model architecture of bigvgan

29 May 12:48
Compare
Choose a tag to compare

pth is pretrained model file, can be used for fine-tune
模型包含生成器和判别器,可用于微调

sovits5.0 48k debug

26 Apr 12:44
503b32e
Compare
Choose a tag to compare

Debug model for sovits5.0 48k. Train 2 days, just for test.

sovits5.0 16k debug

19 Apr 14:33
503b32e
Compare
Choose a tag to compare

For code debug, not last state.

sovits5.0 preview release

10 May 14:14
503b32e
Compare
Choose a tag to compare

预览模型,未进行第二阶段训练,存在音色泄漏;附上训练日志;

v3.0 32k

19 Mar 17:29
9cbad41
Compare
Choose a tag to compare
disable fp16