Beautiful Voice-to-Text Transcription App

A minimal, elegant voice-to-text transcription application with a stunning breathing microphone interface. Convert your speech to text with a beautiful, modern UI.

• Beautiful Minimal UI - A clean and focused interface.
• Stunning Animations - Dynamic microphone visualization with a breathing animation and an audio-responsive waveform.
• Dual AI Modes - Choose between the powerful OpenAI Whisper API or a private, offline local model using faster-whisper.
• Global Hotkeys - Start and stop recording from anywhere on your desktop (Windows only, Win+Shift+ by default).
• Theming - Customize the look and feel with multiple light and dark themes.
• Compact Mode - Switch to a super-compact, always-on-top style UI.
• Custom Title Bar - A sleek, custom title bar for a more integrated look (can be disabled).
• Smart Output - Copy to clipboard, auto-paste to your active application, or just display the text.
• Easy Settings - Modern tabbed settings for audio, transcription, UI, and output.
• Microphone Control - Auto-detects active microphone, with options to manually select and test.
• Silent Mode - Hide transcription text for privacy.

pip install git+https://github.com/quillscribe/quillscribe.git
quillscribe

1. Clone the repository:

   git clone https://github.com/quillscribe/quillscribe.git
   cd quillscribe

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application:

   python run.py

1. Get your API key from OpenAI Platform.
2. Open QuillScribe → Click Settings.
3. In the Whisper tab, select OpenAI Whisper API.
4. Enter your API key.
5. Click Save Settings.

Pros: Fast, high-quality, works immediately. Cons: Requires internet, may have usage costs.

1. Open QuillScribe → Click Settings.
2. In the Whisper tab, select Local Whisper.cpp (Private, works offline).
3. Use the dropdowns to select a model category and a specific model. Smaller models are faster but less accurate.
4. The application will handle downloading the model if it's not already present.
5. Click Save Settings.

Pros: Private, works offline, no ongoing costs. Cons: Requires model download, can be slower and less accurate than the API depending on the model and your hardware.

1. Start Recording: Click the microphone or use the global hotkey (Win+Shift+ by default on Windows).
2. Watch the Magic: The microphone animates with a live audio waveform.
3. Stop Recording: Click the microphone again or use the hotkey.
4. Get Results: Text is automatically copied/pasted based on your settings.

• Copy Only - Copies transcription to the clipboard.
• Paste Only - Pastes directly to the active application.
• Copy & Paste - Does both.
• Display Only - Shows text in the app without copying or pasting.

• Python 3.9+
• Windows 10/11 (Primary target, with global hotkeys)
• Microphone (Built-in or external)
• Internet connection (For API mode and initial model downloads)

• PySide6 - Modern Qt6 UI framework.
• sounddevice - Real-time audio capture.
• numpy - Audio processing.
• openai - Whisper API integration.
• faster-whisper - Efficient local Whisper model implementation.
• pyperclip - Clipboard operations.

• No microphone detected: Check Windows microphone permissions. Ensure the microphone is connected and enabled. Use the "Refresh" and "Test" buttons in the Audio settings. Try the "Auto-select active microphone" feature.
• Poor audio quality: Use the Test button in settings to check your audio levels. Reduce background noise and speak closer to the microphone.

• API errors: Verify your API key is correct and that you have a working internet connection.
• Local model issues: Ensure you have enough disk space for the model download. If transcription is slow, try a smaller model (e.g., tiny or base).

quillscribe/
├── src/quillscribe/
│   ├── __init__.py
│   ├── main.py              # Main application & UI
│   ├── audio_manager.py     # Audio capture & processing
│   ├── whisper_manager.py   # Whisper integration (API & local)
│   ├── settings_dialog.py   # Settings panel
│   ├── config_manager.py    # Configuration handling
│   └── ...
├── requirements.txt
├── setup.py
├── run.py                   # Application launcher
└── README.md

git clone https://github.com/quillscribe/quillscribe.git
cd quillscribe
python -m venv venv
source venv/bin/activate  # Windows: venv\Scripts\activate
pip install -r requirements.txt
python run.py

QuillScribe uses Nuitka to build optimized native executables for superior runtime performance.

1. Python 3.9+ installed
2. C Compiler (Nuitka will auto-download MinGW64 if needed)

1. Clone the repository:

   git clone https://github.com/quillscribe/quillscribe.git
   cd quillscribe

2. Install dependencies:

   pip install -r requirements.txt
   pip install -r requirements-dev.txt

3. Run the Nuitka build script:

   .\build_nuitka.ps1

   The script will:

   • Automatically download MinGW64 compiler (first run only, ~300MB)
   • Clean previous builds and caches
   • Compile QuillScribe to native code (takes 5-10 minutes on first build)
   • Create QuillScribe.exe and QuillScribe-Installer.exe

4. Run the installer or executable:

   • Use the installer: QuillScribe-Installer.exe
   • Or run directly: QuillScribe.exe

• First build: Takes 5-10 minutes as Nuitka downloads MinGW64 and compiles to C
• Subsequent builds: Faster due to caching (2-3 minutes)
• Cache management: The script automatically cleans Nuitka cache for clean rebuilds
• Output: QuillScribe.exe is a standalone executable with all dependencies bundled

Nuitka was chosen over PyInstaller for several key advantages:

• Better Performance: Native compiled code runs 2-3x faster
• Faster Startup: No extraction to temp directory needed
• Smaller Memory Footprint: More efficient resource handling
• Better Compatibility: Handles complex dependencies like faster-whisper more reliably

Required Data Files:

• src/quillscribe/icons/ - SVG and ICO icons for the UI
• src/sounds/ - WAV files for recording notifications
• src/quillscribe/app_logo.ico - Application icon

We welcome contributions! Please fork the repository, create a feature branch, and submit a pull request.

This project is licensed under the MIT License.

Made with care by the QuillScribe Team

Transform your voice into text with elegance and simplicity.