LiveCap v1.3.1 Update

Thank you for using LiveCap! 👍

This update introduces various improvements and important fixes related to the Whisper transcription feature.

✨New Features

• "Auto" Language Setting Added
  
  • You can now select "Auto" as the input language when using the real-time transcription service "FasterWhisper."
    
  • Selecting "Auto" will automatically detect and select the speaker's language.
    
  
  
  
• Automatic Initial Configuration
  
  • Upon first launch, LiveCap now automatically detects your system's language settings and configures the transcription language accordingly.
    
  • If no GPU or an unsupported GPU is detected for Whisper, LiveCap will automatically switch to CPU mode.
    
  • For CPU mode, lightweight models (tiny, base) are recommended.
    
  
  

🔧Changes

• FFMPEG Integration
  
  • Fixed an issue where LiveCap referenced the user's installed FFMPEG instead of its own bundled version.
    
  • Included a bundled FFMPEG within the application, and LiveCap now correctly uses this bundled version.
    
  • Removed dependency on FFMPEG-python.
    
  
  
  
• Updated Model Data Storage Paths
  
  • Changed WhisperS2T model download path to "Models/WhisperS2T."
    
  • Changed FasterWhisper model download path to "Models/FasterWhisper."
    
  • Models will no longer be saved to the User/app_data folder.
    
  • Resolved the issue where symbolic links were incorrectly created for model data.
    
  
  

🐞Bug Fixes

• Fixed the issue where a progress bar incorrectly appeared during real-time transcription.
  
• Improved error messaging to automatically recommend switching to FasterWhisper upon WhisperS2T execution errors.
  
• Resolved GUI error messages incorrectly displaying raw JSON data.
  
• Localized and corrected several error messages.
  
• Enhanced clarity of error messages under specific conditions.
  

We continue to welcome your feedback and suggestions for LiveCap!