RXS

是不是虚拟环境没激活

重新开个虚拟环境吧

> 在看了各位大佬的分享经验，终于在mbp上跑起来了，感谢大家的分享。 但是有个问题，就是推理出来的声音没有大家演示的那么好。第一次推理默认步数，发现声音呼呼的没有字音，然后又增加了训练步数到15和30。然后推理还是不行。最后试了试训练后模型中步数最小的两个推理，有字音了，但是不那么理想。不知道这个和训练素材的长短是否有关？我现在是切割后11条，时长是56秒的素材。另外，这个素材的长度，以及与训练步数的设置，是否有什么关联呢？辛苦大佬指点一二，谢谢啦。  建议云端训练试试，我测下来一样情况

Have you guys tried CPU inference? I've currently tested So-vits-svc and bert-vits2 for CPU inference.

> Have you guys tried CPU inference? I've currently tested So-vits-svc and bert-vits2 for CPU inference. CPU inference Work!

> Have you guys tried CPU inference? I've currently tested So-vits-svc and bert-vits2 for CPU inference.

> > > Have you guys tried CPU inference? I've currently tested So-vits-svc and bert-vits2 for CPU inference. > > > > > > CPU inference Work! > > >...

> Same issue, and I fix it by redownload the 7B model. > > [facebookresearch/llama#73](https://github.com/facebookresearch/llama/pull/73) > > First time, I download by ipfs, and second time, I download by magnet,...

> 真的有点蒙啊，希望可以有个Mac的文档 获取 Outlook for iOS > […](#) > ________________________________ 发件人: GH3a ***@***.***> 发送时间: Friday, January 19, 2024 10:06:11 PM 收件人: RVC-Boss/GPT-SoVITS ***@***.***> 抄送: Stanley-baby ***@***.***>; Comment ***@***.***> 主题: Re:...

> ### Describe the feature or enhancement > Add the ability to export/import CSV data from Websites, on own server and database. Like umami cloud does. > > #2381 #2451...