How to Make Voice Models

Last update: July 17, 2025

- Simple model training tutorial, using RVC -


1. Prepare dataset

  • In the context of RVC, the dataset is an audio file containing the voice the model will replicate. It can be either speaking or singing.

  • For the best results, having a clean dataset is crucial, so take the time to remove any undesired noises.

  • Learn how.


2. Set up RVC

  • With your dataset ready, it's time to set up RVC to train the model.

  • There are plenty of versions of RVC, but these are the best ones for beginners. Pick according to your needs:

    Locally

    On the cloud


3. Train the model

  • Before you start training, we inform you that the training guides are oriented around using TensorBoard. Read about it & install it after setting up RVC.

  • Good luck & remember to be patient! As this won't be an instant process.


You have reached the end.

Report Issues


⚠️
SECURITY ALERT: COMMUNITY SAFETY NOTICE
  1. Discord Vanity Hijack: AI HUB official vanity discord.gg/aihub was hijacked on April 30, 2026. Do not join it, it's a malicious clone server. Please use our official permanent invite: discord.gg/mmRR2TUJF5.
  2. Weights.gg Vanity (At Risk): We are aware of actors targeting AI HUB's Former Partner Weights.gg's vanity discord.gg/weights. To ensure you remain on the legitimate server, please use our permanent invite: discord.gg/CHGgVRCHvm. Be aware that Weights.gg's Discord Server is semi closed since the shutdown on April 1st 2026.
  3. "W8ights" Phishing Scam: Malicious actors are impersonating AI HUB's former partner, Weights.gg. These "W8ights" sites are scams. We have no affiliation with them and are currently investigating into them.
  4. TL;DR: YOU ARE SAFE. This is the original, legitimate AI Hub Docs. We are just informing users about impersonations.