From beginner to mastery of RVC tutorials

Considering that the models on

this site are manually processed through datasets and alchemy, which often does not meet the needs of most students, we present a complete tutorial on the alchemy process using the current model.

• Home
• Let's get started
  • Configuration
  • Download the integration package
  • Introduction to the RVC program
• Training (Advanced)
  • Dataset
  • Noise reduction
  • Manual dataset processing
  • Training
  • Cloud training
• Reasoning (Advanced)
  • Reasoning models
• Advanced gameplay
  • Base mold training
  • Emotional expression training
  • Sound card jumpers
  • Understand the training loss function chart
  • Common and difficult diseases

The current document structure is shown in the image above, and we will gradually distribute it to [Permanent Alchemists] via internal messages.

Before that, you need to address the following issues.

We do not provide network environment issues. Most of the links mentioned in the documentation need to be resolved through the Magic Network. This does not mean we don't migrate these files; in model training, network environment issues are almost always unavoidable, including GitHub and HF, as well as the model dependencies that need to be installed. I hope everyone uses native environments to configure these issues. This is the first problem you need to address.

Second, you need to have some basic programming knowledge. In many common configurations, due to differences in local environments, one-click packages can cause various issues. You need some programming language foundation, especially in model training. It's not just about operating but also about constantly pondering why you do it. Although the document mentions some possible issues.

If you can solve the above problems, you will be able to easily understand the documentation and thus learn about the training process for a model on this site.

I believe that after learning the above, even if Miaoyin doesn't include this part, you can easily create a model of your own.

We will send them via Feishu documents to your internal message and will be sent gradually. You can comment on any questions you don't understand, and on the last working day, we will respond uniformly to any questions you may encounter through comments.