Remembers you, on your device.
Memory is optional, editable, and off until you turn it on. It helps the assistant write in your style without sending a profile elsewhere.
Just speak.
Privately. Anywhere.
Free, open-source dictation and a private AI assistant for your desktop. Nothing you say ever leaves your machine.
- Free foreverNo trial, no upsell
- Open sourceMIT licensed
- 100% localRuns on your machine
- No accountNothing to sign up for
Dictation
Talk. It appears.
Press a key and speak. Your words type straight into whatever app has focus: email, your editor, a chat box, anywhere. Your selected speech engine runs on your machine, so your voice never touches the cloud. Hold to dictate, or go hands-free with a single tap.
Left Ctrl + Left Super to transcribe into any app
- 150+
- WPM speaking
- 45
- WPM typing
- 3×
- faster, hands-free
The assistant
A real assistant,
A real assistant,
one hotkey away.
Ask by voice or type and get a streaming answer in a floating panel that stays out of your way. It remembers the conversation, then collapses when you are done.
Left Ctrl + Left Alt to summon it from any app
And it talks back.
Hear answers through local Kokoro or the voice engine you choose, with the pace and reply length under your control.
Screen vision
It sees what you see.
Ask about an error, a chart, or anything else on your screen. SpeakoFlow captures the display you are using only when you ask, then answers with that context.
It learns how you like to work.
Private memory and switchable profiles make the same assistant useful for different parts of your day.
Different jobs, different voices.
Switch between a warm companion, a terse quick-answer mode, or a profile of your own. Each can have its own voice and reply length.
More than dictation.
Still free.
The best-known voice tools are polished dictation products. SpeakoFlow adds an assistant, screen context, spoken replies, and source code you can inspect.
SpeakoFlow
$0
Free and open source
- A full assistant. Chat, screen vision, memory, and spoken replies.
- Your stack. Local models, your own server, or a provider you choose.
- Your software. MIT licensed, inspectable, and usable without an account.
Official individual pricing checked July 2026. Product plans and limits can change.
Privacy
Your voice never leaves your machine.
Local speech-to-text is the fixed boundary. Everything beyond it follows the model, memory, and network choices you make.
Transcription stays local.
Your selected on-device speech engine turns audio into text. The privacy promise is not tied to one hardcoded model.
The assistant follows your route.
Use the built-in offline model, your own local server, or a provider you explicitly configure.
Memory stays inspectable.
Personal memory is off by default, stored locally, and available for you to review, edit, export, or erase.
FAQ
Questions, answered.
Everything people ask before downloading. Still curious? The GitHub discussions are open.
Yes. SpeakoFlow is free and open source under the MIT license, with no trial, subscription, or upsell.
Your voice is transcribed on your machine and never uploaded. The assistant only contacts the model provider you choose, which can be a fully local one.
No. You can run entirely on the built-in local model. If you would rather use OpenAI, Anthropic, Groq, or others, plug in your own key. Your choice.
Windows, macOS, and Linux.
SpeakoFlow builds on Handy's excellent local dictation core and adds the assistant panel, screen vision, spoken answers, personal memory, and profiles. Full credit to Handy below.
Ready when you are.
Download SpeakoFlow and start talking with a private assistant that is yours to keep.
Also on macOS and Linux. See all downloads.