Dictation and read-back for Windows and Ubuntu Linux, processed on your computer.
Talk. It types. It reads back. On your computer.
Built for people who work off the grid: a portaledge, a plane, anywhere there is no connection. Built for people whose eyes or brains read differently, who need text spoken aloud. Built for busy people with long documents and no time to read them. VozSay turns speech into text in most Windows applications. It does the work on your own machine. No account or sign-in. VozSay doesn’t upload your speech or dictated text.
Update and license checks are on by default and can be turned off. See Privacy.
- Windows 10 and 11, Ubuntu 26.04
- Dictation works offline after setup
- One-time price
Your voice stays here
Most dictation software records you, sends the audio to a company's servers, and sends text back. That is how it gets accurate, and it is why it needs your internet connection and your trust.
Your voice is never uploaded. VozSay does the work on your computer.
Unplug the network cable and VozSay keeps working. On Linux, the first-start model download must be completed first. On Windows, its voices are included with the installer. On Linux, the speech and voice models download once, the first time VozSay starts, after you agree. Optional cleanup models are downloaded only when you choose them: the app shows the size and asks first. Once downloaded, those models run locally too. The app has no account, telemetry or cloud processing. It does not send us your speech or dictated text, and its core features do not depend on an external service.
This page loads no fonts or trackers from anywhere else. The one exception is a small Cloudflare visit counter that sets no cookies (see Privacy).
What it does
Press a key, talk, and the words appear where your cursor is.
-
Types into other programs
Word, Outlook, a browser, a terminal, your editor. The text lands in whatever window you were already working in.
-
Reads back to you
Highlight text anywhere and have it read aloud. For eyes that are tired, low vision, dyslexia, or anyone who takes in the spoken word better than the page. Hear what you wrote instead of re-reading it.
-
Cleans up as you speak
Turn on VozBuff and every dictation is tidied before it lands: spelling, punctuation, and the ums and false starts removed.
-
Deep Clean, for when it matters
VozBuff tidies every dictation as you talk. When a paragraph really counts, like an email to a client or a script page, open it in VozPad and give it a deeper polish with the full-size on-device AI. Pick your pass: fix errors only, cut the filler, or tighten it up, always keeping what you meant.
-
Keeps what you said
Your words, your choice. Save your notes, keep your last five dictations handy to copy and reuse, or choose Shred to keep only the latest without saving dictation history to disk.
-
Your words, spelled your way
Teach it the names, jargon, and terms you use so it stops guessing at them.
Pricing
Your licence does not expire. One-time payment. No subscription.
One payment. The price rises to $99 once the first fifty licences are sold. If you buy now you keep the lower price, and you are not moved onto the higher one later.
How updates work. Your licence covers every update we release for your version of VozSay, at no extra cost. When a big new version arrives, with better voices or new features, it is a separate purchase, and people who already own VozSay get an upgrade discount. Old school, the way major upgrades used to work. No subscription, nothing to renew.
- Use on two of your own computers
- Updates within your version cost nothing
- Future major versions are a separate purchase, discounted for owners
- 14-day refund, no questions asked. See the refund policy
Try it first
VozSay runs for 14 days before it asks for a license. The trial is the full program, and it does not ask for a card or an email address to start. After that, VozSay keeps working on the free plan unless you buy a license.
Download VozSay
If Windows shows a blue SmartScreen screen, click More info, check the publisher says Robert Lopez, then Run anyway.
Built and tested for Ubuntu 26.04 with the standard GNOME desktop, on 64-bit Intel or AMD computers. It has not been tested on other versions of Linux, so we do not say it works there.
The Linux installer does not include the speech and voice models. The first time VozSay starts, it shows what it needs and asks before downloading anything: the speech model (about 464 MB) from huggingface.co and the voices (about 337 MB) from github.com, once. That is about 800 MB in total. On Linux, the first-start model download must be completed first.
Linux install steps (Ubuntu 26.04)
- Open Files, then Downloads, and double-click the vozsay .deb file. App Center opens. It shows "Unknown publisher", "Potentially unsafe" and "License unknown". For VozSay_Linux_amd64.deb downloaded from vozsay.com, that is expected. If you got the file from anywhere else, do not install it.
- Click Install, click Install again, and type your password. The button changes to "Installed".
- Log out and log back in.
- Click "I accept" on the License agreement.
- VozSay asks to download its speech and voice models (about 800 MB, once) from huggingface.co and github.com. Click Download. If you choose Not now, nothing is downloaded and VozSay closes; it asks again the next time it starts.
- Ubuntu's "Add Keyboard Shortcuts" window appears. Click Add.
- Only on computers with an NVIDIA graphics card: VozSay offers an optional GPU pack (1.4 GB).
Using it
The back side button on your mouse (or F7) starts dictation. The forward side button (or F8) reads the selected text aloud. If your mouse has no side buttons, F7 and F8 do the same jobs.
To use different buttons, click the VozSay icon at the top of the screen, choose Settings, click "Set button", then press the mouse button you want.
Removing it
Open Terminal and type: sudo apt remove vozsay
Known limits
Things worth knowing before you buy, rather than after.
- Windows 10 and 11, and Ubuntu 26.04 Linux. Other versions of Linux are not tested. A Mac version is coming later.
- English dictation. Other languages are available for translation, but speech recognition is tuned for English.
- It cannot type into programs running as administrator. Windows blocks that for every application, not just this one.
- A blind user's screen-reading software may not be able to read VozSay's own windows out loud.
- Recognition quality depends on your microphone and your room, and no dictation software is perfect. That is what the 14 days are for.
- VozSay uses AI models that run on your own computer. They work on your words: transcribing what you say, and rewriting the text you provide when you use cleanup. Like any AI model, they can make mistakes, so review important text before you rely on it.