What you said
uh I think we should probably move this meeting to tomorrow
What Just Say inserts
I think we should probably move this meeting to tomorrow.
Speak naturally. Just Say turns your voice into writing, on your Mac or PC. Local-first. No account. No API key.
macOS 14 or later · Apple Silicon · Windows 10 or 11, 64-bit
Listening
On your computer
uh I think we should probably move this meeting to tomorrow
Just Say
I think we should probably move this meeting to tomorrow.
Inserted at the cursor.
01
Hold Left Option on a Mac or Left Alt on Windows, or the shortcut you set.
02
Just Say listens while you hold it.
03
Transcription runs on your computer.
04
The sentence lands at the cursor.
Mac · Local-first
Hold a shortcut, talk naturally, and Just Say turns speech into writing where you already work.
What you said
uh I think we should probably move this meeting to tomorrow
Just Say
I think we should probably move this meeting to tomorrow.
Writing cleanup
English cleanup can drop fillers, finish punctuation, and keep the meaning. “Probably” stays.
Messages
Cursor is here
I think we should probably move this meeting to tomorrow.
Where you work
Speak in the app you already have open. Just Say inserts there. No chat window.
Languages
English, German, Hindi, or Russian. Transcription, not translation.
Selected region
Not the whole screen
Kept with this dictation.
Screen context
Optional. Capture a selected area while you dictate. Kept on your computer, not uploaded.
Writing
This is the English path. German, Hindi, and Russian keep your wording. Just Say only normalizes spacing and applies dictionary entries saved for that language.
What you said
uh I think we should probably move this meeting to tomorrow
What Just Say inserts
I think we should probably move this meeting to tomorrow.
“Uh” and “um” can come out. “Probably” and “actually” stay.
A line that reads as a sentence can gain a capital and a closing mark.
“John — I mean James.” can become “James.”
Conservative, Natural, Professional, Casual, Concise, or Developer. Cleanup does not change the facts.
Where you work
Just Say is a voice layer for the Mac. It types into the app you already have open. If it cannot find a text field, it says so.
Where you type
Speak, then release.
Where you type
Speak, then release.
Cursor is here
I think we should probably move this meeting to tomorrow.
Where you type
Speak, then release.
Where you type
Speak, then release.
Where you type
Speak, then release.
These are places you already type. Just Say does not ship a separate plugin for each app.
Screen context
Screen context stays off until you turn it on. While you dictate, you select a region. Just Say keeps that image on your computer, with the dictation. Text in the image is read on the device. It is not treated as words you spoke, and it is not uploaded.
Cancel discards a capture that is still in progress. Dictation still inserts if capture fails.
Selected region
Not the whole screen
The image stays with this dictation.
Spoken words
The sentence you said.
Read from the image
Kept on your computer, separate from speech.
Four languages
You pick one language. Just Say transcribes it. It does not translate.
Privacy
01
Your voice
02
Your Mac
03
Just Say
After the model is on disk, transcription stays here. Just Say does not send audio or text to a cloud speech service.
Internet is for a model download or repair you choose, installer updates from this site, and a daily check for a newer version (it sends nothing about you; turn it off in Settings → About). The download is the model file, not your transcript.
History, dictionary, styles, notes, and settings live in ~/Library/Application Support/ShabdaTaraFlutter/ on a Mac, or %LOCALAPPDATA%\ShabdaTaraFlutter on Windows. There is no required account and no API key.
Workflows
Writing
Speak the thought. English cleanup and a style you choose finish the sentence without changing the facts.
Coding
Developer style can format spoken case names and symbols. It does not write the program for you.
Messaging
The words land at the cursor in the app you already have open. Just Say does not open a chat.
Notes
History and notes stay on your computer. Search, copy, or delete them. Empty history stays empty.
The app
Just Say lives in the menu bar. Home, History, Dictionary, Styles, Language, and Settings are there when you need them. The listening bar stays on screen while you speak.
Listening
Hold Left Option or Left Alt. Speak. Release.
Voice bar
Cancel inserts nothing. Confirm finishes, the same as letting go of the shortcut. Escape cancels too.
Recent
I think we should probably move this meeting to tomorrow.
Ship the loading state when the request fails.
History
Transcripts stay on your computer. Search, copy, or delete them. Empty history stays empty.
Language
English, German, Hindi, or Russian. Just Say transcribes. It does not translate.
Selected region
Optional
Kept with this dictation.
Screen
Optional. Capture a selected area while you dictate. The rest of the screen is not recorded.
Drawn from the app’s structure. Not a screen capture.
Speech is transcribed on your Mac once the model you need is on disk.
Hold a shortcut and talk in the app you already have open.
English, German, Hindi, or Russian. You choose. Just Say does not translate.
When you want it, a region you select stays with the dictation. On your computer.
No required account and no API key. For Mac and Windows.
Mac uses a disk image. Windows uses a small installer. Both are direct downloads from this site.
Mac
macOS 14 or later. Apple Silicon. Download the DMG, drag Just Say to Applications, then open it from the menu bar.
Windows
Windows 10 or 11, 64-bit. Download JustSay-Setup.exe, run it, then open Just Say from the Start menu.
Quick start
Direct download · no account required.
A menu-bar app for Mac. Hold a shortcut, speak, and Just Say transcribes on your Mac, cleans English locally, and inserts the text where you were already typing. It does not open a chat.
Transcription runs on the Mac. English uses a local Parakeet model, then Apple’s on-device English speech if needed. German, Hindi, and Russian use a local Whisper model after you install it. Just Say does not send dictation audio or transcripts to a cloud speech service.
After the speech model you need is already on disk, dictation does not require a network. Internet is used when you choose to download or repair a model. That transfer is the model file, not your transcript. Installer updates are downloaded from this site. Once a day the app also asks GitHub whether a newer version is out. Nothing about you, your voice, or your text is sent, and you can turn it off in Settings → About.
English, German, Hindi, and Russian. You pick one in Settings. Just Say transcribes that language. It does not translate. German, Hindi, and Russian need the multilingual model from Settings → Models. Automatic language detection is not available.
Yes. Download JustSay-Setup.exe from this site and run it. Windows 10 or 11, 64-bit. Mac uses JustSay-macOS.dmg.
Microphone, to hear you. Accessibility, to type into other apps. Screen Recording is optional and is used only when you select a region. Speech Recognition is used only if English dictation uses Apple’s on-device engine.
On your computer: ~/Library/Application Support/ShabdaTaraFlutter/ on a Mac, %LOCALAPPDATA%\ShabdaTaraFlutter on Windows. Temporary dictation audio is deleted after transcription unless you turn on Keep audio files. History, notes, dictionary, styles, and screenshots stay in that folder. Just Say does not upload them.
Speak naturally. Keep the words. Download Just Say for Mac from this site.