whisperstream free local ai dictation
Top AI Tools

Whisperstream: Free Local AI Dictation for Windows and Mac

Visit website

Whisperstream is free, open source AI dictation that runs fully on your Windows or Mac, with local cleanup, per-app styles, 39 languages and no…

whisperstream free local ai dictation

Whisperstream is dictation that runs entirely on your computer. You think and speak at roughly three times the speed you type, and Whisperstream closes that gap without sending a syllable to the cloud. It is free and open source, with no subscription and no account. The product appeared on Hacker News Show HN on 11 November 2026. You can see the official site at https://whisperstream.io/.

What is Whisperstream?

Typing caps most people at 40 to 60 words a minute. Whisperstream lets you speak instead, in any app you can type in, and turns your speech into text right on your PC or Mac. Its own cleanup model, Stream-1 Mini, takes what you said messy and returns it clean: filler cut, corrections applied, and the result typed straight into the app you are using.

The tool also adapts to where you are. Per-app styles mean Gmail gets an email and Slack gets a message, and you can write your own prompt for any app. A custom dictionary spells your names right: add names, products, and jargon once, and cleanup matches them by sound. Agent Speak extends the same idea to coding agents, letting you hear a Claude Code or Codex response read aloud and answer by speaking.

Key features

  • On-device speech recognition for Windows and Mac, with no cloud transcription and no internet needed after setup.
  • Stream-1 Mini cleanup that cuts filler and applies spoken corrections without sending a word off your computer.
  • A custom dictionary that matches names, products, and jargon by sound.

Who is it for?

Whisperstream fits anyone who writes a lot and would rather speak: writers, developers dictating prompts to a coding agent, and people answering email or chat all day. It works in the apps you already use, so there is no new editor to learn and no copying text between windows.

It also fits people who will not trade privacy for convenience. Your voice never leaves your computer, and the company says it never receives your recordings or your text. The app only goes online to download a model at setup, check for updates about once a day, and for cloud cleanup, which stays off until you turn it on and sends text, never audio.

Pricing

Whisperstream is free: every feature included, no subscription needed. There is no trial, no account to set up, no license key, and no device limit, so you can install it on each computer you use. It is open source under the GPL-3.0 license, so you can read every line of code on GitHub. The homepage compares that $0 price with leading cloud dictation apps at $140 or more a year, and $0 against $420 or more over three years.

Pricing may change; see the official website for current pricing. Details above reflect the official site as checked on 13 November 2026.

Getting started

Download Whisperstream for Mac or Windows from the official site and install it with no account. You need internet once to download the models, then dictation works offline. The requirements are Windows 10 or 11 in 64-bit, which works on CPU with no GPU required, or a Mac with Apple silicon running macOS 14 or later. Local AI cleanup recommends at least 4 GB of memory for the default model and 8 GB for the customizable model. Start by dictating a real email or a prompt to your coding agent, and add the names you use every day to your dictionary.

KEEP EXPLORING

More tools to discover.

Find your next starting point