Braina 1.80 Changelog – Offline Speech Recognition (CPU/GPU) Support

New improvements made in Braina version 1.80 are as follows:

  1. Offline speech recognition with both CPU & GPU support (Beta)
    • This update (v1.80) comes with the most awaited feature of in-built offline speech recognition.
    • Offline speech recognition will work on both CPU and dedicated GPU (for faster speed).
    • To enable offline speech recognition, Go to Settings->Speech Recognition tab and make sure that “Offline” option is selected under “Speech to text engine option” drop-down menu.
    • Various speech to text models for different needs are provided:
      • Lite – Lite model for fast speech to text results. Only supports English language. Options for both CPU & GPU available. This is the default model.
      • Basic – Slow model compared to Lite but provides better accuracy. Options for both CPU & GPU available but it may run slow on CPU if you do not have good processor. Only supports English.
      • Enhanced – Multi-language model with better accuracy than Basic model. Only works on GPU.
      • Best – Most accurate multi-language model. Needs good GPU with more than 5GB VRAM.
    • Note: Offline speech recognition only works on 64 bit OS. If you face performance issues, please use online speech recognition.
  2. Advanced AI engine is now enabled by default
  3. Fixed a bug that caused Braina to listen its own voice
  4. Status bar (can be disabled from Settings) : Shows current speech recognition options. Useful when using offline speech engine.
  5. Auto-punctuation support for dictation: Braina now supports auto-punctuation feature. Use Offline speech recognition or Online (Edge) to enable auto-punctuation.

Leave a Reply

Your email address will not be published. Required fields are marked *