VoiceFlow Desktop v0.2.0 is Live•100% Offline AI

Type at the speed of thought. 100% offline.

Press Ctrl + Space anywhere on Windows—Chrome, Word, Notepad, Slack, or VS Code. Powered by local CPU-accelerated OpenAI Whisper. Zero audio ever leaves your PC.

Windows 11 / 10 (64-bit) • Single installer (~25 MB) • Download as .zip

100% Private (Zero Cloud)
AVX2 CPU Vector Accelerated
Works in Any Windows App
Airplane Mode Ready
AVX2 Realtime CPU
audio_engine.rs — fn process_voice_stream(audio: &[f32]) {
▶
VoiceFlow
Listening...
Switch Application:
Engineered for Windows

Why VoiceFlow is the Ultimate Dictation Tool

Designed from the ground up to replace expensive cloud subscriptions with pure, local, private computing power.

100% Private & Air-Gapped

Zero Cloud Audio Processing

All speech-to-text inference runs locally in memory on your PC using OpenAI Whisper. Zero audio recordings, zero transcripts, and zero telemetry ever leave your device.

Works in Airplane Mode (No Wi-Fi)
Zero audio cached or saved to disk
Safe for healthcare & NDA work
No accounts, passwords, or tracking
Everywhere

Global Ctrl + Space Hotkey

A seamless low-level Windows keyboard hook lets you dictate into any focused application: Notepad, Word, Slack, Chrome, VS Code, or WhatsApp.

Ctrl+SpaceInvokes anywhere
Zero GPU Needed

AVX2 CPU Acceleration

Compiled with AVX, AVX2, and FMA vector math instructions. Transcribes 3x faster than real-time speech on modern Intel & AMD CPUs without heavy GPU power drain.

3x faster than real-time speech
~25 MB Single File

Self-Contained Runtime

Requires zero developer tooling. No Node.js, Python, or Git needed on your laptop. Includes everything inside a single lightweight installer.

Zero configuration required
Native Windows

Silent Tray & Auto-Start

Closes silently to the Windows System Tray and registers with Windows Registry to start automatically on system boot with minimal RAM footprint (~80 MB idle).

Lightweight background process
Minisign Verified

Cryptographic Auto-Updater

Receive future models and improvements automatically. Signed cryptographically with Minisign and checked at most once every 12 hours from GitHub Releases.

Secure 1-click update prompt
Zero Learning Curve

Up and Running in 3 Steps

No command-line setup, no configuration files, and no cloud accounts required.

01

Download & Launch

Run the single setup wizard. VoiceFlow opens and automatically downloads the optimized 140 MB Whisper base model on first launch.

No Python or Node.js required
02

Press Ctrl + Space

Click into any application where you can type. Press Ctrl + Space and the dictation pill appears floating above your cursor.

Intercepts globally across Windows
03

Speak & Instant Insert

Speak your thoughts naturally. Press Enter or wait for silence; VoiceFlow simulates instant keystrokes right where your cursor is blinking.

3x faster than real-time speech
The Better Choice

VoiceFlow vs. Cloud Dictation Services

Why pay monthly subscription fees to companies that listen to your private audio when your CPU can do it for free?

Feature
VoiceFlowRecommended
Cloud Dictation SaaS
Audio Processing
100% Local CPU (Zero Cloud)
Streamed to Remote Cloud Servers
Pricing & License
100% Free & Open Source (MIT)
$12 - $30 / month recurring fee
Works Without Internet
Yes (Airplane Mode Ready)
No (Fails immediately with no Wi-Fi)
Privacy & Data Retention
Zero audio kept or logged to disk
Voice data retained & used for training
System-Wide Windows Hotkey
Ctrl + Space (Any active app)
Restricted to browser or custom app
AI Model Transparency
OpenAI Whisper (Auditable)
Proprietary black-box cloud APIs

Seeing "Windows protected your PC"?

Standard Windows SmartScreen

When installing VoiceFlow on Windows 10 or 11, Microsoft Defender SmartScreen may display a blue warning screen. This happens with all newly published open-source tools because Microsoft requires indie developers to purchase a costly $300-$500/year EV Code Signing certificate. VoiceFlow is 100% free and community-driven, so we have not purchased an enterprise certificate yet.

How to install in 2 clicks:Windows Defender
1

Click "More info"

Click the small underlined text beneath the paragraph.

2

Click "Run anyway"

A new button will appear at the bottom right. Click it to launch.

100% open-source under MIT License. Clean, audited, and virus-free.
Inspect Code on GitHub
Common Questions

Frequently Asked Questions

Everything you need to know about VoiceFlow privacy, offline AI, and performance.

No! Once VoiceFlow downloads the lightweight base AI model (~140 MB) on your initial launch, it works 100% offline. You can dictate seamlessly on airplanes, trains, or in high-security facilities with zero Wi-Fi.