#4464 RuntimeError: {'message': 'tts failed', 'Exception': 'giongnuct.wav not exists'}

This error occurs because a TTS engine requiring a reference audio file (typically used for voice cloning) could not locate it. The file giongnuct.wav was expected but does not exist. This usually happens when you have selected a clone voice role (e.g., F5‑TTS, CosyVoice, OmniVoice‑TTS, GPT‑SoVITS) and the automatic extraction of the original audio segment for that subtitle line failed or was not created.

Solution

1. Adjust voice recognition parameters for clone compatibility

   • Go to Menu → Tools → Advanced Options → Voice Recognition Parameters
   • Set Shortest speech duration (ms) to 3000
   • Set Longest speech duration (seconds) to 10
   • Enable Whisper pre‑split audio? and Merge too‑short subtitles to adjacent
     These settings ensure each subtitle segment falls within the 3–10 second range required by most cloning engines.

2. Improve the reference audio quality

   • In the main window’s More settings..., enable Separate human voice and background sound. This isolates the speaker’s voice before cloning, which often fixes extraction failures.

3. Force a fresh re‑process

   • Check the Clear generated option on the main interface to remove cached files, then run the translation again.

4. Check for audio‑related issues

   • Verify that the original video actually contains an audio track (see FAQ #15).
   • Ensure the video file path is short and contains only English characters / numbers to avoid ffmpeg path‑length problems.

5. If the problem persists

   • Switch to a non‑clone voice role (e.g., a built‑in Edge‑TTS voice) to test if the issue is cloning‑specific.
   • Examine the log file in logs/ for earlier ffmpeg errors that might indicate why the reference audio could not be extracted.

Please refer to relevant documentation:

• How to clone original video voice for dubbing
• Improve video translation quality – best practice for clone roles
• Common error: video contains no audio track