Whispering is a deep integration of OpenAI's Whisper model directly into your VS Code workflow. Designed specifically for developers who want to transcribe thoughts into code or documentation without compromising speed, privacy, or system resources.

[!NOTE] Optimized specifically for Visual Studio Code on Windows.

✨ Key Features

• 🚀 Zero-Config Auto-Setup: Say goodbye to manual Python environment wrestling. Whispering automatically creates its isolated virtual environment and installs all necessary dependencies (including PyTorch and backend libraries) via a robust internal Node.js process the first time you run it.
• 🧠 True Multi-Backend Support: Run inference where it makes the most sense for your machine:
  • CPU: Highly optimized fallback for any system using whisper.cpp (GGUF).
  • CUDA (NVIDIA GPUs): Blazing fast transcription leveraging faster-whisper.
  • OpenVINO (Intel NPUs/iGPUs): Harness the power of your laptop's AI accelerator without draining battery or CPU cycles using optimum-intel.
• 🎙️ Seamless Push-to-Talk: Hold your configured hotkey (default: F8) to record, and release to instantly transcribe. The text is automatically pasted at your cursor location.
• 👀 Visual Feedback: Real-time progress bars natively integrated into the VS Code UI for model loading, backend switching, and initial installations. You always know what the engine is doing.
• 🔒 Privacy First: 100% offline. Audio data never leaves your machine. Your code and your thoughts remain strictly local.

🛠️ Requirements

• Python 3.10+: Must be installed and accessible.
• FFmpeg: Required for audio processing. Must be installed and added to your system's PATH.

Note on Hardware: While CPU execution is supported, using CUDA on NVIDIA GPUs or OpenVINO on Intel AI hardware is highly recommended for near-instantaneous transcription of large models.

⚙️ Detailed Setup & Configuration

Whispering is highly customizable directly from your VS Code Settings. Navigate to Preferences: Open Settings (UI) and search for Whispering to fine-tune your experience.

Modifying any of these settings dynamically restarts the background listener with visual progress indicators.

Hardware Backend (whispering.device)

Choose the engine that best fits your local hardware:

• cpu (Default): Uses whisper.cpp under the hood. Perfect for laptops without dedicated GPUs. It's highly optimized and quantized, making it very RAM-friendly.
• cuda (Recommended for NVIDIA): Uses faster-whisper. If you have an NVIDIA GPU, select this for blazing fast, real-time transcription.
• openvino (Recommended for Intel AI PCs): Uses optimum-intel. Specifically designed for Intel NPUs and modern integrated graphics, providing hardware acceleration without draining your battery.

Model Selection (whispering.model)

Select the Whisper model size. Larger models are more accurate but require more resources and take slightly longer to process.

• mini: Uses tiny model (~0.5GB). Extremely fast, but may struggle with strong accents.
• fast: Uses base model (~0.5GB). Very fast, a bit more robust than mini.
• small (Default): Uses small model (~1.0GB). The sweet spot. Excellent accuracy for most dictation tasks while remaining very fast.
• medium: Uses medium model (~2.5GB). High accuracy. Recommended for complex technical jargon or architecture.
• ultra: Uses large-v3 model (~4.5GB). Studio-grade transcription. Best used with the cuda backend.

General Settings

⚖️ License & Credits

Whispering is built upon the incredible work of the open-source community:

• OpenAI Whisper: The core speech recognition models are developed by OpenAI and released under the MIT License.
• Faster-Whisper: High-performance implementation using CTranslate2.
• Whisper.cpp: High-performance C++ implementation for CPU inference.

You can use these models freely for personal and commercial projects under the terms of the MIT License.

🚀 Getting Started

1. Install the Extension: Download and install from the VS Code Marketplace.
2. Prerequisites: Ensure you have Python 3.10+ installed and FFmpeg added to your system's PATH.
3. Auto-Setup: Press your configured hotkey (default: F8). If it's your first time, Whispering will automatically create a virtual environment and securely download all required libraries (PyTorch, Whisper, etc.).
4. Talk: Once the notification says "Bereit!" (Ready), einfach F8 gedrückt halten, sprechen und loslassen. Dein Text wird magisch an der Cursorposition eingefügt!

📜 Changelog

0.1.4

• Process Management: Improved lifecycle management

0.1.3

• Localization: Added full support for English and German.
• Settings Refactoring: Unified configuration keys (whispering.*) with legacy support for whisperingSTT.

0.1.2

• Process Management: Improved lifecycle management for Python background processes to ensure clean exits.

0.1.1

• CUDA Runtime Fixes: Resolved issues with missing NVIDIA DLLs on Windows systems by automatically including local library paths.