README.md

March 9, 2026 · View on GitHub

Dictto — Voice to Text for Windows

Free, open source voice-to-text for Windows.
Hold a hotkey, speak, release — clean text appears at your cursor. Everywhere.

Download for Windows · How It Works · Features · Build · Contributing

https://github.com/user-attachments/assets/be8bda6e-2b05-43d3-8982-63cf2f811cf2

_{Hold a hotkey, speak, release — text appears at your cursor in ~2 seconds.}

How It Works

Dictto uses a push-to-talk model. No always-on microphone, no wake words, no background listening.

Step	What happens
1. Press	Hold `Ctrl + Win` — Dictto starts recording from your microphone
2. Speak	Talk naturally. A floating waveform bar shows you it's listening
3. Release	Let go — audio is sent to OpenAI Whisper for transcription
4. Done	AI cleans up filler words, fixes grammar, and pastes the text at your cursor

The entire flow takes about 1–2 seconds after you stop speaking. Audio is never saved to disk.

Why Dictto?

Works everywhere — Any text field in any app. VS Code, Slack, Notepad, Chrome, Word, Telegram — if you can type there, Dictto can paste there.
Your keys, your data — Bring your own OpenAI API key (BYOK). Audio goes directly from your device to OpenAI. No middleman, no Dictto servers.
Actually private — Zero telemetry, zero analytics, zero accounts. Your API key lives in Windows Credential Locker. Read our privacy policy →
Open source — Fully auditable. AGPL-3.0 licensed. You can read every line of code that touches your microphone and your clipboard.

If you've used Wispr Flow, SuperWhisper, or other AI dictation tools and wanted something open source, private, and free — Dictto is built for you.

Features

Push-to-talk — Ctrl + Win hotkey. Speak only when you want. No always-on microphone.
AI text cleanup — GPT removes "um", "uh", "like", fixes grammar, and formats your text naturally.
102 languages — Supports all languages available in OpenAI Whisper, including mixed-language dictation (Spanglish, Franglais, etc.).
Lightweight — ~5 MB installer. Built with Tauri + Rust, not Electron. Minimal CPU and memory usage.
Smart clipboard — Saves your clipboard before pasting, restores it after. Your clipboard history stays intact.
Auto-start — Optionally starts with Windows and lives quietly in your system tray.
Transcription history — All your transcriptions are saved locally in a SQLite database for your reference.
Fully offline-capable architecture — The app itself runs locally. Only the API calls go to OpenAI (using your key).

Install

Download

Download the latest installer from GitHub Releases.

Setup

Run Dictto_x64-setup.exe
Open Dictto from the system tray (bottom-right of your taskbar)
Go to Settings → API and enter your OpenAI API key
Hold Ctrl + Win and start talking

Windows SmartScreen

Note

You may see a "Windows protected your PC" screen. This is expected and safe — here's why.

Windows SmartScreen flags apps that are not code-signed with a paid certificate. Most open source projects — including Dictto — don't have one yet because certificates cost hundreds of dollars per year.

This is not a virus warning. It's a "we don't recognize this publisher" warning. Dictto is fully open source — you can audit every line of code that runs on your machine.

To install:

Click "More info" (the small text link, not the button)
Click "Run anyway"

We are applying for a free code signing certificate through SignPath Foundation (used by major open source projects). Once approved, this warning will disappear.

Verify your download: Every release includes SHA-256 checksums. Compare with:

certutil -hashfile Dictto_0.1.0_x64-setup.exe SHA256

Privacy

Dictto is designed to be private by default:


Telemetry	None. Zero analytics, zero crash reports, zero tracking.
Audio	Sent directly to OpenAI via your API key. Never stored on disk. Never sent to Dictto.
API key	Stored in Windows Credential Locker (OS-level encryption). Never logged or transmitted.
Accounts	None required. No sign-up, no email, no database.
Data storage	Settings and transcription history stored locally in `%LOCALAPPDATA%`.

Read the full Privacy Policy →

Tech Stack

Layer	Technology
Desktop framework	Tauri v2 (Rust backend + WebView frontend)
Frontend	React 19 + TypeScript + TailwindCSS v4
Audio capture	WASAPI (native Windows audio via Microsoft `windows` crate)
Transcription	OpenAI Whisper API (BYOK)
Text cleanup	OpenAI GPT (BYOK)
Local storage	SQLite (settings + history) + Windows Credential Locker (API key)

Build from Source

Prerequisites: Node.js (LTS), pnpm, Rust (stable), Tauri CLI prerequisites

git clone https://github.com/dictto-app/dictto.git
cd dictto
pnpm install
pnpm dev          # Development mode with hot-reload
pnpm build        # Production build (.exe installer)

See CONTRIBUTING.md for detailed setup instructions.

Alternatives

Looking for voice-to-text on other platforms or with different trade-offs? Here are some alternatives:

App	Platform	Open Source	Price	Key Difference
Dictto	Windows	Yes (AGPL-3.0)	Free (BYOK)	Push-to-talk, local-first, your own API key
Wispr Flow	macOS, Windows	No	$12/mo	Closed source, subscription model
SuperWhisper	macOS, Windows	No	$8.49/mo	Closed source, local Whisper model option
Whisper Writer	Windows, Linux	Yes	Free	Python-based, local Whisper, no text cleanup
Buzz	Cross-platform	Yes	Free	Transcription GUI, not push-to-talk dictation

Contributing

Contributions are welcome! See CONTRIBUTING.md for setup instructions and guidelines.

License

AGPL-3.0 — You can use, modify, and distribute Dictto freely. If you modify it and offer it as a service, you must share your changes under the same license.

Built with Tauri and Rust.