███╗   ███╗██╗████████╗██╗  ██╗██████╗ ██╗██╗     
████╗ ████║██║╚══██╔══╝██║  ██║██╔══██╗██║██║     
██╔████╔██║██║   ██║   ███████║██████╔╝██║██║     
██║╚██╔╝██║██║   ██║   ██╔══██║██╔══██╗██║██║     
██║ ╚═╝ ██║██║   ██║   ██║  ██║██║  ██║██║███████╗
╚═╝     ╚═╝╚═╝   ╚═╝   ╚═╝  ╚═╝╚═╝  ╚═╝╚═╝╚══════╝

██╗    ██╗██╗  ██╗██╗███████╗██████╗ ███████╗██████╗ 
██║    ██║██║  ██║██║██╔════╝██╔══██╗██╔════╝██╔══██╗
██║ █╗ ██║███████║██║███████╗██████╔╝█████╗  ██████╔╝
██║███╗██║██╔══██║██║╚════██║██╔═══╝ ██╔══╝  ██╔══██╗
╚███╔███╔╝██║  ██║██║███████║██║     ███████╗██║  ██║
 ╚══╝╚══╝ ╚═╝  ╚═╝╚═╝╚══════╝╚═╝     ╚══════╝╚═╝  ╚═╝
MITHRIL Pixel Logo

100% Free, Privacy-First AI Voice Dictation and Assistant

🍎 Apple Silicon Macs
M1, M2, M3 chips (2020+) • 196MB
💻 Intel Macs
Intel processors (2019 and earlier) • 208MB
Download for Windows
View on GitHub
Learn More

How It Works & Privacy

Complete transparency about our data practices

ZERO DICTATION TRACKING GUARANTEE

YOUR DICTATION CONTENT IS NEVER TRACKED, SAVED, OR STORED

[INSTALLED VERSION]

Dictation content never leaves your device. Zero tracking. Zero storage.

[LOCAL VERSION]

100% offline. Dictation content never touches the internet.

Even when using AI Assistant features, your dictation is processed in real-time and NEVER saved permanently anywhere. When using the AI Assistant (that means you're using our OpenAI keys), we will keep track of only the character count and never the content being exchanged with OpenAI.

[LOCAL TRANSCRIPTION]

All voice-to-text processing uses open-source Whisper-CPP models running entirely on your device. Your voice recordings never leave your computer.

Bundled Model: Whisper Tiny (Q5_1 quantized) - Fast, lightweight transcription with excellent accuracy for most use cases.

  • [✓] 100% offline processing
  • [✓] Open-source models from OpenAI Whisper
  • [✓] Optimized with whisper.cpp for fast CPU inference
  • [✓] Audio files deleted immediately

[PRIVACY MODES]

[INSTALLED VERSION]

Requires login. Only tracks character count for AI assistant (not content).

[LOCAL SETUP]

No login required. Zero telemetry. Complete privacy.

[AI ASSISTANT TELEMETRY] - VOICE DICTATION HAS ZERO TELEMETRY

[WHAT WE TRACK] (Installed Version Only)

  • [-] Character count when using AI assistant
  • [-] Session duration and timestamps
  • [-] Model used (e.g., whisper-tiny)
  • [-] Platform (Mac/Windows)

[WHAT WE DON'T TRACK]

  • [X] Voice recordings
  • [X] Transcription content
  • [X] What you say to the AI assistant
  • [X] Any personally identifiable content

[CODE TRANSPARENCY]

Here's the exact code showing our minimal telemetry (character count only):

// From src/main/main.js - AI Assistant usage tracking
const payload = {
  transcript_chars_original: userPrompt.length,    // Character count only
  transcript_chars_cleaned: assistantReply.length, // Character count only
  metadata: {
    user_words: countWords(userPrompt),           // Word count only
    assistant_words: countWords(assistantReply), // Word count only
  }
  // NOTE: Actual content is NEVER included in telemetry
};

[IMMEDIATE FILE DELETION]

All audio files are automatically deleted immediately after transcription completes. This happens locally on your device and ensures no voice data persists anywhere in our system.