Frequently Asked Questions

Common issues and their solutions.

Installation Issues

How do I install dependencies?

Follow the step-by-step instructions in the Getting Started guide. The pip install method handles everything automatically. For manual setup, make sure you have Python 3.10+, FFmpeg, and aria2 installed.

FFmpeg is not found

Install FFmpeg for your platform:

  • Windows: Download from ffmpeg.org and add to PATH
  • macOS: brew install ffmpeg
  • Linux: sudo apt install ffmpeg

Transcription Issues

Transcription shows hallucinations or repeated text

This is a known Whisper behavior. Try these solutions in order:

  1. Enable VAD filtering — this removes silence segments that cause hallucinations
  2. Use a larger model — upgrade from Medium to Large, or from Large-v3 to Large-v2 (v2 is often more stable)
  3. Enable audio separation — isolates speech from background noise

Inaccurate subtitle timing

  • Use FasterWhisper for word-level timestamps
  • Enable semantic segmentation for natural break points
  • Manually fine-tune in the subtitle editor if needed

LLM Issues

LLM request failures

  1. Verify your API key is correct and has credit
  2. Confirm the Base URL ends with /v1 (no trailing slash)
  3. Reduce thread count (start with 5, increase gradually)
  4. Check your network connection
  5. Review the log files in Settings > Logs

Getting 429 (Too Many Requests) errors

Your concurrent thread count is too high for your API plan. Reduce threads:

  • SiliconCloud: 3-5 threads
  • DeepSeek: 5-10 threads
  • OpenAI: 10-20 threads
  • Relay station: 20-50 threads

Poor subtitle quality from LLM

  • Upgrade to a better model (gpt-4o-mini → gpt-4o)
  • Lower the temperature setting (0.3 → 0.1)
  • Add manuscript hints with terminology and proper nouns
  • Enable reflective translation for better translation quality

General

Is my data processed locally?

All video and audio processing happens on your local machine. If you use cloud APIs (OpenAI, DeepSeek, etc.), only the extracted text is sent to the API provider. Video files never leave your computer.

What subtitle formats are supported?

SRT, ASS, and VTT. You can also burn subtitles directly into the video (hard subtitles) or keep them as a separate track (soft subtitles).

Still have questions?

Check the GitHub Issues page or start a Discussion.