The error indicates that the voice-cloning reference audio file nu_viet.wav is not present in the expected folder: C:/Users/Admin/Downloads/win-pyvideotrans-v4.15/f5-tts. Custom reference audio for cloning must physically exist in that folder and be registered before it can be selected in the 配音角色 (voice role) dropdown.
How to fix it (1. Non-clone path issues; 2. Correct reference-audio workflow):
Verify the file location
- Copy your audio file into exactly this folder:
C:/Users/Admin/Downloads/win-pyvideotrans-v4.15/f5-tts - Make sure the filename matches exactly, including extension and case:
nu_viet.wav (not .WAV, not nu_viet (1).wav).
Check the WAV requirements
- Format must be WAV, mono, sample rate 16 kHz–44.1 kHz.
- Length: 3–10 seconds.
- Content: single clear speaker, no background noise, no leading/trailing silence.
- If your original file is mp3/m4a/etc., convert it to WAV first.
Register the reference audio in the software
- Go to menu:
Menu -> TTS Settings -> Set Reference Audio (菜单 -> TTS 设置 -> 设置参考音频). - Enter the entry in the exact format
filename.wav#transcript, e.g.:
nu_viet.wav#the text actually spoken in this audio - Save it.
Then select it correctly
- In the main window's
配音角色 (voice role) dropdown, pick nu_viet.wav (as registered). - If you instead intend to clone the original video's voice, select the role
clone — do NOT point to a custom file.
- Restart the software after copying the file and saving the reference-audio entry, so the dropdown list refreshes.
Note: If your goal is cloning the original speaker's voice rather than a custom audio, use the clone role and ignore the f5-tts folder entirely.
Please refer to the related documentation：