#7790 Each line must contain 3 parts separated by '#': audio_name.wav#audio_text#lang_code

This format audio_name.wav#audio_text#lang_code is not a valid reference-audio entry format, and combined with your clone requirement it will fail.

The correct reference-audio format requires only 2 parts separated by #: filename.wav#text spoken in the audio. There is no third lang_code field. Adding a third field makes the software treat the whole string as the spoken text, so the reference audio cannot be located correctly and dubbing fails.

Correct steps:

1. Prepare a clean, single-speaker WAV audio, 3-10 seconds, mono, no background noise, no leading/trailing silence.
2. Copy it into the software directory folder f5-tts.
3. Open Menu -> TTS Settings -> Set reference audio and fill in exactly:
   myaudio1.wav#the exact text spoken in this audio
4. Save, then select myaudio1.wav in the Voice role dropdown on the main interface.

Additional requirements for reliable cloning (already implied by your clone tag):

1. In Menu -> Tools/Options -> Advanced Options -> ASR Settings, set VAD:max speech duration(s) to 6-10 and VAD:min speech duration(ms) to 3000-4000. Reference clips shorter than 3 seconds are very likely to fail cloning.
2. Enable Separate Vocal/BGM in More... to improve reference audio quality.
3. Use an AI translation channel (DeepSeek/OpenAI) with Send full SRT format checked.

Note: clone only works inside the Video Translation function, not in the standalone Text-to-Speech panel.

请查阅相关文档：

• 语音克隆
• 全部配音渠道说明