whisperMeOff

WhisperMeOff is a powerful, privacy-first voice-to-text app that runs entirely on your machine using cutting-edge AI. Built on whisper.cpp (the blazing-fast C++ port of OpenAI's Whisper) and powered by llama.cpp for intelligent text formatting, it delivers professional-quality transcription without ever sending your voice to the cloud.

Hold a hotkey, speak your mind, and watch your words appear — automatically pasted into whatever app you were using. No subscriptions, no account required, no data leaves your computer.

Features

• Voice Recording: Hold Ctrl+Shift+R to start recording, release to transcribe (or toggle mode)
• Recording Overlay: Visual spectrum analyzer VU meter shows real-time audio levels during recording with colorful animated bars
• Session Stats: Today's and all-time transcription count, word count, audio duration, processing time, and words-per-minute (WPM) tracking (shown on separate lines)
• Auto-Paste: Automatically pastes transcribed text to the previously active window
• Whisper.cpp Powered: State-of-the-art local speech recognition using OpenAI's Whisper models (tiny to large)
• Whisper Translation: Translate non-English speech to English directly with Whisper (for supported languages)
• Llama.cpp Text Formatting: Optional AI-powered formatting cleans up your transcription with proper punctuation and paragraphs
• Llama Translation: Translate transcribed text to 27+ languages using Llama AI
• Cancellation Support: Long-running transcription can be cancelled (internal API support)
• Auto-Recovery: Automatic recovery from GPU hangs or native library errors with timeout protection
• 100% Offline: Works completely offline - no internet required after model download
• HuggingFace Integration: Download GGUF quantized models directly from HuggingFace
• 7 Visual Themes: Light, Dark, Nord, Dracula, Gruvbox, Monokai, and Synthwave themes
• System Tray: Run in background - record without the app being active
• Minimize to Tray: Option to minimize to system tray instead of taskbar
• Launch at Login: Automatically start the app when Windows boots
• Custom Vocabulary: Define word lists to improve recognition accuracy for domain-specific terminology
• Word Replacements: Automatically replace spoken phrases with different text after transcription
• History Management: Clear all transcriptions or filter by age (older than 1 hour/6 hours/24 hours/7 days/30 days/90 days)
• Model Validation: Visual badges showing valid/invalid model status with size and sample rate info
• Recording Modes: Support for push-to-talk, push-on talk, and push-off talk modes
• Download Progress: Real-time download progress with auto-select after completion

Top Benefits

• Boosts Productivity: Speak 3x faster than you type. Capture ideas instantly without breaking your workflow.
• Improves Accessibility: A voice-first interface makes content creation accessible to everyone, regardless of typing ability.
• Enables Hands-Free Use: Keep your hands on the keyboard or mouse while capturing your thoughts via voice.
• Reduces Typing Errors: Voice transcription eliminates typos and spelling mistakes from your workflow.
• Cuts Injury Risk: Reduce strain on your hands and wrists from excessive typing - ideal for those with RSI or carpal tunnel.
• Enhances Focus and Flow: Stay in the zone by capturing thoughts naturally without interrupting your creative process.
• Saves Time Overall: Less time typing means more time for what actually matters - creating and thinking.
• Adapts to Users: Choose from multiple Whisper model sizes and Llama formatting options to match your needs and hardware.

Screenshots

App Screens

Requirements

• Windows 10/11
• .NET 10.0 Runtime

Dependencies & NuGet Packages

This project uses the following NuGet packages:

Audio & Transcription

AI & ML

UI & Desktop

Data & Utilities

Installation

1. Download the latest release
2. Run whisperMeOff.exe
3. The app will guide you through initial setup

Usage

Quick Start

1. Select a Whisper Model: Go to the Whisper tab and download a model
2. Set Hotkey: The default is Ctrl+Shift+R (you can change this in General settings)
3. Start Recording: Hold Ctrl+Shift+R to start recording
4. Release to Transcribe: Release the keys to stop recording and transcribe
5. Auto-Paste: The transcribed text is automatically pasted to your previous window

Configuration

Whisper Settings

• Language: Select the language or use "Auto Detect"
• Translate: Enable to translate output to English
• Model: Select a Whisper model size (tiny, base, small, medium, large)

Llama Settings (Optional)

• Enable Llama text formatting for cleaner output
• Translation: Enable Llama translation to translate transcribed text to 27+ languages (English, Spanish, French, German, Italian, Portuguese, Russian, Chinese, Japanese, Korean, Arabic, Hindi, and more)
• Download models from HuggingFace (search for GGUF quantized models)
• Enter your HuggingFace token for private models

General Settings

• Hotkey: Change the trigger key (default: R)
• Launch at Login: Enable to automatically start the app when Windows boots
• Download Paths: Customize where models are saved
• Recording Mode: Choose Push-to-talk (hold to record) or Toggle (press to start/stop)
• Clipboard: Configure clipboard restore behavior
• Minimize to Tray: Minimize to system tray instead of taskbar
• Theme: Choose from 6 visual themes (Light, Dark, Nord, Dracula, Gruvbox, Monokai)

Model Sizes

Keyboard Shortcuts

Building from Source

dotnet build

License

MIT License