Notes on talking to your computer.

Compare dictation apps

Wispr Flow isn't open source. Four open-source dictation apps with their licence, platforms and GitHub stars: Handy 33,992, OpenWhispr 9,495, VoiceInk 6,833 and SpeakoFlow 318.

Updated October 17, 2026 10 min read

Wispr Flow isn't open source: 4 free alternatives that are

Wispr Flow is polished, but its code is closed and your speech is transcribed on its servers. These free, open-source apps type into any app from your own computer.

Free dictation on Windows: SpeakoFlow, Handy and OpenWhispr transcribe on your PC, while the built-in Win + H voice typing needs the internet.

October 17, 2026 9 min read

Wispr Flow alternatives for Windows, free and open source

Seven ways to dictate into any app on a Windows PC, what each one costs, and where your speech is transcribed.

Meeting notes written for you: a summary, key takeaways and next steps with owners Marco, You and Priya.

October 17, 2026 10 min read

Open-source Granola alternatives: free meeting notes with no bot

Six free apps that write meeting notes from your computer's audio, set against Granola and Wispr Flow's new Notetaker.

SpeakoFlow and Wispr Flow side by side: $0 against $15 a month, speech turned into text on your computer against in the cloud, no word limit against 2,000 words a week, and the platforms each runs on.

SpeakoFlow vs Wispr Flow: free and local vs $15 a month

Both put your spoken words into any app. They differ on price, on where your speech is transcribed, and on which devices they run.

Updated October 17, 2026 9 min read

Wispr Flow's plans: Free at $0 for 2,000 words a week, Pro at $15 a month or $12 billed yearly, and Growth at $23 per user a month.

Pricing

Wispr Flow pricing 2026: is it free, and what does Pro cost?

Every Wispr Flow plan as its own pricing and help pages list it on October 17, 2026, with the free plan's word limits and the one-year and three-year totals worked out.

The SpeakoFlow 2.0 window under the headline Runs offline, showing that your voice is transcribed on your computer, with 99 languages, 65 speech models, no telemetry, MIT open source code, and free builds for Windows, macOS and Linux.

Answer

Does Wispr Flow work offline? No. These dictation apps do

Wispr Flow turns your voice into text on its own servers, so with no internet there is nothing to transcribe it. Here is what its docs say, and which apps keep working offline.

Updated October 17, 2026 10 min read

Four free dictation tools for Linux with their packages and licences: SpeakoFlow, Handy, OpenWhispr and nerd-dictation.

Roundup

Best dictation software for Linux in 2026: 8 tools that work

Press a key, talk, and the text lands in whatever app you are using. These are the Linux tools that do that today, what each one ships, and how they cope with Wayland.

Guides and research

"Just talk. It types for you." beside a Gmail reply, where the SpeakoFlow pill cleans up a dictated sentence by striking out "um" and "Thursday, no wait".

Guide

Voice dictation for ChatGPT, Cursor and Claude Code prompts

Every place you prompt has a text box, and a dictation app that types at your cursor works in all of them. Here is how to set one up, with the shortcuts and settings that matter for prompts.

A SpeakoFlow voice conversation: the user asks for help planning launch week, the assistant answers out loud, and the user moves the emails to Friday while it listens.

Guide

How to get a Jarvis AI on your PC, free and offline

What a voice assistant on Windows, macOS or Linux can really do in October 2026, what it still cannot do, and how to set one up in a few minutes without writing code.

Updated October 17, 2026 11 min read

Bar chart of one 0.8B model before and after fine-tuning: overall score 47.3% to 70.7%, restraint 89.7% to 92.6%, edit accuracy 4.9% to 48.8%, against a dashed line at 50% for doing nothing.

Guide

How to clean up dictated text: rules first, then a small model

Deleting the word "um" is a table lookup. Deciding whether "no, Friday" cancels "Thursday" is not. Build those as two stages, in this order, and measure them separately.

Updated October 17, 2026 10 min read

Seven cleanup models ranked by overall score with a dashed empty bar for the do-nothing baseline sitting fourth, three models above it and four below.

Release

SpeakoFlow Mini: local dictation cleanup that knows when to stop

Half of dictation cleanup is knowing when not to edit, and that half is where every model I tried fell over. So I trained one on the rule instead of prompting for it.