Skip to main content
Joe's Tech Solutions
Open Source · Privacy First

Whisper WalkieYour voice, your machine.

Nothing leaves.

Hold a hotkey, speak naturally, release. The transcribed text types directly into whatever window has focus — no clipboard, no cloud, no account. Whisper AI runs entirely on your machine.

Simple by design

How It Works

Three seconds from voice to text in any application on your computer.

01

Hold

Press and hold the push-to-talk hotkey (default: Right Alt). The app starts recording immediately.

02

Speak

Talk naturally into your microphone for as long as you need. No time limit, no wake word.

03

Done

Release the key. Whisper transcribes locally and types the text directly into your focused window.

First time setup

Quickstart Guide

Up and running in under five minutes.

1

Download

Click Download, choose your platform. The installer handles everything.

2

Install & Open

Windows

Run the installer. If SmartScreen appears, click “More info” → “Run anyway” — the app is open source and safe.

macOS
  1. Open the downloaded .zip — it extracts automatically
  2. Open the WhisperWalkie folder
  3. Right-click WhisperWalkieOpen
  4. Click “Open” when macOS asks to confirm

Grant Accessibility permissions when prompted: System Settings → Privacy & Security → Accessibility.

Linux
  1. Extract the downloaded .tar.gz file
  2. Open the WhisperWalkie folder
  3. Double-click WhisperWalkie to launch

Optional: run ./install-linux.sh to add a desktop shortcut to your app menu.

3

Choose Your Mic

Open Settings in the app and select your microphone from the dropdown.

4

Try It!

Hold Right Alt (or your chosen hotkey), speak naturally, release. Text appears wherever your cursor is!

Why switch

Why Whisper Walkie

Most voice tools trade your privacy for convenience. You should not have to choose.

Other Voice Tools

The trade-offs you accept

  • Cloud required — audio uploaded to their servers
  • Monthly fees that add up over time
  • Locked to a single application
  • Account and login required
  • Always listening in the background
Whisper Walkie

What you actually get

  • 100% local — audio never leaves your machine
  • Free forever — no subscriptions, no fees
  • Works in any application on your OS
  • No account, no login, no tracking
  • Push-to-talk — only records when you hold the key
What's inside

Features

Everything you need. Nothing you don't.

100% Local

Audio is transcribed entirely on your device using faster-whisper. Nothing is sent to any server, ever.

Works Everywhere

Browser, Slack, VS Code, Word, games — if your OS can focus it, Whisper Walkie can type into it.

GPU Accelerated

CUDA support on Windows and Linux for near-instant transcription. Falls back to CPU automatically.

Cross-Platform

Native support for Windows, macOS, and Linux. One codebase, three platforms, zero compromises.

Push-to-Talk

Hold Right Alt (customizable), speak, release. It only records when you tell it to — total control.

Open Source

MIT licensed. Read every line of code. Fork it, modify it, ship it. Free now and always.

The story

Why I Built This

I was using tools like Willow and OpenWhispr for voice transcription. They worked — but every word I spoke was going to the cloud, where other companies could use it to train their models. That didn't sit right with me.

I was already running Ollama and Open WebUI for local LLMs. I wanted the same thing for voice: fast, private, and completely offline. So I built it.

Whisper Walkie was built entirely using the agentic AI workflow with Claude Code from Anthropic — proving that one developer with the right tools can ship production-quality, cross-platform software that stands up against anything built by a traditional engineering team.

I'm giving it away for free because I built it for people like me who care about privacy. And because I want to contribute something real to the open source community — working software for Windows, macOS, and Linux that anyone can use, inspect, and build on.

— Joe, Joe's Tech Solutions

Get started

Download

Whisper AI model is bundled — no internet needed after download.

All downloads on GitHub Releases. Whisper AI model bundled — no internet needed after download.

Windows: If SmartScreen says “Windows protected your PC,” click “More info” “Run anyway.”

macOS: Right-click the app and choose “Open” on first launch. Grant Accessibility access when prompted.

Linux: Extract the folder and double-click WhisperWalkie to launch.

These warnings are normal for independent software. Whisper Walkie is fully open source — inspect every line on GitHub.

Human + AI collaboration

Built with AI

This entire app was built through human-AI collaboration with Claude by Anthropic. The architecture, the cross-platform backend, the GUI — all designed and iterated through a tight loop between a developer who knew what the app needed to do and an AI that could help ship it faster without cutting corners.

That is not a disclaimer — it is the point. AI-assisted development lets a single developer build and maintain production-quality, cross-platform software that would otherwise require a team. Whisper Walkie is a working proof of that model.

Read the source on GitHub
Common questions

FAQ

Everything you might want to know before you download.

Yes! SmartScreen warns about apps from new publishers — this is normal for any independent software. Whisper Walkie is fully open source, so you can read every line of code on GitHub. Just click ‘More info’ → ‘Run anyway’ to proceed.

This is Apple’s Gatekeeper blocking unsigned apps — normal for independent software. Go to System Settings → Privacy & Security, scroll down, and click ‘Open Anyway’ next to the Whisper Walkie message. You only need to do this once.

Yes, completely free and open source under the MIT license. No trials, no premium tier, no subscriptions. Forever.

Never. Whisper Walkie runs the AI model entirely on your machine. Your audio is processed locally and never touches a network. No accounts, no telemetry, no cloud.

No. The base Whisper model runs comfortably on any modern CPU. If you have an NVIDIA GPU (Windows/Linux), transcription will be even faster with CUDA acceleration.

Yes. The speech recognition model is bundled with the download. No internet connection needed after you download.

Whisper supports 99 languages. It auto-detects the spoken language, or you can set a specific one. Accuracy varies — English is the strongest, with about 10 other languages performing well.

Yes. Open Settings in the app and choose from: Right Alt, Scroll Lock, Pause, F13, F14, Insert, or Right Ctrl.

Go to System Settings → Privacy & Security → Accessibility. Click the + button and add the Whisper Walkie app (or your Terminal if running from source).

Try Whisper Walkie Today

Free. Private. Works in any app.

MIT License · No account required · Audio never leaves your machine