Dicto - The private AI-powered voice dictation
Dicto is an intelligent voice dictation app that transcribes your voice and automatically formats it based on where you're typing. Whether you're sending a casual message to a friend or drafting a professional email, Dicto adapts your words to fit the context.
- Context-aware formatting — Dicto detects your active app and applies the right style automatically
- No typing needed — Hold a key to record, release to paste. It's that simple
- Smart AI processing — Your transcription is polished by AI before it reaches your clipboard
- Works everywhere — Slack, iMessage, Gmail, Notion, and more
- Custom dictionary — Add technical terms and proper nouns for accurate transcription
- Text shortcuts — Create abbreviations that expand into full phrases
Dicto_.Speak.It.Writes.2.mp4
Choose between two transcription modes:
| Mode | Engine | Privacy | Speed | Accuracy |
|---|---|---|---|---|
| Local | Whisper Small | ✅ Fully offline | Fast | High |
| Cloud | Deepgram Nova-3 | Encrypted | Real-time | Very High |
Local mode uses OpenAI's Whisper Small model running entirely on your device. No internet connection required, no data sent anywhere. Your voice stays on your machine.
Cloud mode offers real-time streaming transcription via Deepgram for users who prefer speed over complete privacy.
- macOS (coming soon)
- Windows (planned)
- Linux (planned)
A minimal, always-on-top widget that stays out of your way until you need it. Click or use a keyboard shortcut to start recording.
Dicto automatically detects where you're typing and applies the appropriate writing style:
| App Category | Example Apps | Default Style |
|---|---|---|
| Personal | iMessage, WhatsApp, Telegram | Casual — all lowercase, minimal punctuation |
| Work | Slack, Teams, Discord | Professional — sentence case, full punctuation |
| Mail, Gmail, Outlook | Formal — proper greeting, structured paragraphs | |
| Notes | Notes, Notion, Obsidian | Structured — markdown headers, bullet points |
Powered by Deepgram's Nova-3 model, Dicto captures your voice with high accuracy in real-time. Watch your words appear as you speak.
Once you stop recording, Dicto uses Groq to intelligently format your transcription. The AI only adjusts:
- Capitalization
- Punctuation
- Line breaks
Your words remain unchanged — just properly formatted for the context.
Add technical terms, proper nouns, and acronyms to ensure they're transcribed correctly:
- Company names (e.g., "PostgreSQL", "OAuth")
- Industry jargon (e.g., "KPI", "ROI")
- Names of people and places
Create abbreviations that expand into longer phrases:
| Trigger | Replacement |
|---|---|
brb |
be right back |
omw |
on my way |
addr |
123 Main Street, City, State 12345 |
Choose your preferred style for each category:
Personal
- Casual — all lowercase, minimal punctuation
- Brief — sentence case, periods only
- Expressive — lowercase with exclamation marks
Work
- Professional — sentence case, full punctuation
- Casual — lowercase with dashes
- Direct — sentence case, periods only
- Formal — proper greeting with colons
- Professional — standard business format
- Friendly — warm tone with exclamations
Notes
- Structured — markdown headers and bold labels
- Bullets — organized bullet points
- Prose — flowing paragraphs
Dicto supports transcription in multiple languages:
- English (US, GB)
- Spanish
- French
- German
- Italian
- Portuguese
- Japanese
- Korean
- Chinese
Dicto is built with modern, performant technologies:
- Frontend: React + TypeScript + Vite
- Desktop: Tauri (Rust)
- Transcription: Deepgram Nova-3
- AI Processing: Groq
- Database: SQLite (local)
- Styling: TailwindCSS + shadcn/ui
# Install dependencies
bun install
# Run development server
bun run dev# Navigate to desktop app
# Run Tauri development
bun run dev