DEV Community

OB
OB

Posted on

OmniDictate v2.0: The Future of Local Dictation on Windows

We are thrilled to announce the release of OmniDictate v2.0.0, a major update that completely transforms the user experience while keeping the core promise of fast, private, and accurate dictation.

OmniDictate is a free, open-source tool that brings real-time AI speech-to-text to your Windows PC. It runs entirely locally using the faster-whisper engine (based on OpenAI's Whisper), ensuring your data never leaves your machine.

✨ What's New in v2.0?

This release focuses on usability, aesthetics, and performance.

🎨 Premium Slate & Glass UI

Version 2 introduces a stunning, modern graphical interface.

  • Dark Slate Theme: Easy on the eyes and professional.
  • Frosted Glass Accents: A touch of modern elegance.
  • Intuitive Layout: Designed for clarity and focus, putting all controls right where you need them.

large-v3-turbo Support

We've upgraded the engine to support the latest large-v3-turbo model. This delivers state-of-the-art transcription accuracy with significantly improved speed, making real-time dictation smoother than ever.

🎛️ Streamlined Controls

We've simplified how you interact with the app:

  • Unified Control: Easily toggle between Voice Activity Detection (VAD) and Push-to-Talk (PTT) modes directly from the UI.
  • No More Complex Hotkeys: We've removed the confusing stop hotkeys. Now, you just speak (VAD) or hold a key (PTT).

🛠️ Enhanced Configuration

  • Hallucination Filtering: Easily add repetitive phrases to a blocklist via the GUI to keep your transcripts clean.
  • Customizable Hotkeys: Set your preferred PTT key with a simple click.
  • Auto-Save: All your settings—model size, language, sensitivity—are saved automatically.

🔑 Key Features

  • 100% Local & Private: No cloud APIs, no subscriptions, no data tracking.
  • Type Anywhere: Dictate directly into Word, Notepad, Discord, your browser, or any active window.
  • Real-Time Accuracy: Powered by optimized faster-whisper models.
  • Voice Commands: Say "new line", "delete last 3 words", or "comma" to control your text.
  • GPU Acceleration: Fully supports NVIDIA GPUs (CUDA) for lightning-fast performance.

📥 Download & Install

Get the latest version from our GitHub Releases Page.

Options:

  1. Installer (.exe): The easiest way to get started. Installs to Program Files and creates shortcuts.
  2. Portable (.7z): No installation required. Just extract and run!

Note: As a free open-source project, our installer is not digitally signed. You may see a Windows SmartScreen warning ("Windows protected your PC"). Simply click "More info" -> "Run anyway" to proceed.


🖥️ System Requirements

  • OS: Windows 10 or 11 (64-bit)
  • GPU: NVIDIA GPU with CUDA support (Highly Recommended for best performance)
  • RAM: 8GB minimum (16GB+ recommended)

🤝 Contribute & Support

OmniDictate is open source!

Enjoy the freedom of voice typing! 🎙️

Top comments (0)