NNiceVois

No local GPU required

An Applio alternative for people without a local GPU

Train and download my RVC modelSame RVC v2 output, run in the cloud.
The honest answer

Applio is good software and this is not a claim to be better than it. If you own a capable NVIDIA GPU and want full control over every training parameter, install Applio. NiceVois exists for the case where you do not have that hardware, or do not want to maintain a Python environment to get one model file.

Both approaches run RVC v2 and both produce a .pth and a matching .index. The difference is entirely about where the computation happens and how much setup you own.

Where the two differ

Applio (local)NiceVois (cloud)
Hardware neededNVIDIA GPU, typically 8–12 GB VRAM or moreAny device with a browser
SetupInstall, dependencies, model downloadsNone
Parameter controlExtensiveAudio, epochs, and model name
Output.pth and .index.pth, .index, and complete ZIP
Cost modelYour electricity and your hardwareFree during public beta
Runs offlineYesNo
Batch and repeat runsUnlimited once installedSubject to beta limits

When Applio is the better choice

When cloud training is the better choice

The files are interchangeable. A model trained here loads in Applio, and a model trained in Applio loads anywhere else. Nothing about the output is specific to the service that produced it.

What the beta accepts

Up to 15 minutes of audio and up to 100 epochs, in WAV, MP3, FLAC, M4A, or OGG. Finished files stay available for seven days unless you delete them sooner. See the dataset guide for what to upload and the epoch guide for how far to train.

Permission still applies

Train your own voice, or a voice you have clear permission to use. Our acceptable use policy explains the boundary.

No GPU? Train it here instead

Upload permitted audio and leave with the same portable RVC files Applio would produce.

Train my RVC model