FastVits is a powerful PC voice-cloning tool that runs fully offline on your machine. Built to help users clone a specific voice and turn text into near-indistinguishable speech output. No installation—unzip and go, simple and fast.

FastVits needs no network—it runs entirely on your local computer. Turn any text into speech anytime, ideal for audiobooks and voice applications. The advanced local AI engine makes the package 10 GB, ensuring high-quality output and fast processing.
With powerful features and friendly operation, FastVits delivers an efficient, convenient voice-synthesis solution—for personal fun or professional production, an outstanding tool.
The goal is to create a voice template customized to your needs.
1. Pick the audio folder to train: the app auto-processes WAV folders—slicing, removing background music and noise, and other preprocessing.

2. Use a pretrained model: usually no manual action—the app loads the author's model by default. To choose manually, pick from the model folder.
3. Name your model in English.
4. Choose a speech-recognition model—the default base model is fine.
5. Set epochs: 200 recommended; though dozens often suffice, 300 gives better results.
6. Set batch size: adjust to your VRAM—keep it small.
7. Learning rate: use the default. When set, click Start Training to build the template.

With the template built, start turning text into speech.
1. Type or import text: enter it directly or import a TXT file. (If both exist, the TXT takes priority.)
2. Pick the trained audio model—the G-prefixed model from Step 1.


3. Keep defaults: SDP mix ratio, noise level, speed and more stay as default.
4. Generate: click Generate and the app produces speech from your text and chosen model.
These two steps let you easily build personalized voice templates and turn any text into a specific voice.