The error "Upload WAV Ref Audio" indicates that you are trying to use a GPT-SoVITS voice cloning model, but the required reference audio file is missing or invalid.
To resolve this, follow these steps:
- Select a Pre-trained Voice: In the main interface, under the "Voiceover (TTS)" section, click the dropdown menu next to "Role." Choose a voice from the list that does not have "(Clone)" in its name (e.g., "zh-CN-XiaoxiaoNeural" for Edge-TTS). These are pre-trained voices that do not require a reference audio file.
If You Must Use GPT-SoVITS (Voice Clone):
- Ensure you have locally deployed the GPT-SoVITS service. The software cannot upload files to an external service.
- In the main interface, go to Menu -> Settings -> GPT-SoVITS.
- Enter the correct local API address (e.g.,
http://127.0.0.1:9880). - Click "Refresh" to load the available models from your local server.
- Then, select a model and a reference audio file that is already on your local server (not uploaded from your PC).
Important: The "Upload WAV Ref Audio" button in the software is for sending an audio file to your locally running GPT-SoVITS server to create a new voice model. It is not for selecting a file for direct use in translation. For standard translation, use a pre-built TTS voice like Edge-TTS.
If you continue to have issues, please provide the complete error log from the C:/Mac/Home/Desktop/pyvideotrans/win-pyvideotrans-v3.86/logs/ directory.