Voice Dictation Tool

A modern, GUI-driven AI dictation app powered by RealtimeSTT and OpenAI's Whisper model (large-v3).

Features

• High-Quality Bluetooth Headset Support: Dynamically toggles Windows PyAudio streams on and off to prevent Bluetooth headsets from dropping into low-quality Hands-Free (HFP) modes while dictation is idle.
• Global Hotkey: Press Alt+K from anywhere in Windows to start/stop dictation. The text will instantly be typed into your active window.
• Modern UI: A sleek, dark-themed PyQt6 interface that tracks your dictation history, similar to the Windows Clipboard window.
• Dynamic Microphone Selector: The top right of the GUI application features a dropdown menu containing all valid audio hardware devices on your PC (filtered heavily so you aren't overwhelmed by virtual lines). Swap them on the fly!
• Background Mode: Minimizes to the system tray so it is always ready to use without cluttering your taskbar.
• Graceful VRAM Clearance: When the app is closed, it forcefully triggers resource cleanup to ensure no background threads hang onto your massive NVIDIA VRAM allocation.
• Smart Sounds: Uses soft Windows system sounds to notify you exactly when the microphone starts and stops recording.
• Timestamped History: Each dictation entry shows the time it was recorded and its word count.
• Clear History: A "Clear All" button in the header lets you wipe the history list when it gets long.
• Live Audio Level: A real-time volume meter bar appears while recording so you can confirm the mic is picking up your voice.
• Auto-Copy to Clipboard: Every transcription is automatically copied to your clipboard, useful for apps where keyboard.write() doesn't work — just Ctrl+V to paste.
• Error Resilience: If typing into the focused window fails, the text is still available in your clipboard and history.

Prerequisites

• A compatible NVIDIA GPU (tested on RTX 5090) for utilizing the large-v3 model locally without latency.
• Python 3.9+

Installation

1. Create a virtual environment:

   python -m venv venv
   call venv\Scripts\activate.bat

2. Install dependencies:

   pip install -r requirements.txt

3. Run the application: Double-click launch.bat, or run the Python script without a console window:

   pythonw dictate.py

Usage

• Run the tool and use the Dropdown in the top right to select your preferred microphone.
• You do not need the window open! Press the Minimize button (_) and it will hide in the System Tray near your clock.
• Press Alt + K to toggle recording. When you hear the ping, speak. Press Alt + K again to stop, and it will magically type out your text into whatever window you have focused.
• All dictations are saved into the history interface where you can quickly copy them again.
• To exit the program completely and free up GPU memory, click the Close button (X) on the main window, or right-click the system tray icon and click "Exit App".