1. Training the Voice Model: Collect at least 10 minutes of quality audio samples in a quiet environment. Name your training model and provide the directory path for your audio files. Keep sample settings at default, then click “Process Data” and “Feature Extraction.”
2. Using Model Inference Online: After training, select a cloned voice from the generated list. Download the required index or .pth file into the asset directory. Refresh the voice list, provide vocals, select an extraction method, and click “Convert” after setting the output directory.
3. Adding Audio Files: Use MimicPC’s default directory system to upload audio files. Navigate to the desired directory, click the three dots button, choose upload, and select files from your device.
4. Removing Vocals and Reverb: Use the second tab in RVC to separate vocals and reverb. Choose the appropriate model, drag and drop files, set the output directory, select an export format, and click “Convert.”
5. Converting to ONNX: Go to the Export ONNX tab, select the .pth model file, and specify the output .onnx file path to convert your model.
