SayLocal›Offline dictation on Windows
Offline dictation on Windows 10 and 11
SayLocal turns your voice into text on your own PC and types it into any Windows app. Hold the right Ctrl key, speak, let go. Your recordings stay on your computer.
For Mac and Windows · Trial with an email address, no credit card
How it works on Windows
- Hold the right Ctrl key – or any other key you set in SayLocal.
- Speak normally, even quietly.
- Let go – the text appears at the cursor in your email program, Slack, Word, the browser, a ticket system or any other app where you can type.
For long texts, tap the space bar while holding the key and keep dictating hands-free. Filler words like “um” are removed, and silence produces no text.
Runs on your graphics card – or on the processor
With a graphics card that supports Vulkan, SayLocal runs the speech recognition on the GPU and your text appears after about 0.2 seconds. Without a suitable graphics card it runs on the processor and takes a few seconds.
On PCs without a suitable graphics card, SayLocal can offer an optional cloud recognition: the recording is sent encrypted to Cloudflare, turned into text and not stored. It is off by default, only runs after your explicit tick and can be switched off at any time. On PCs with a suitable graphics card this option does not exist.
Offline after activation
On first launch, SayLocal downloads about 550 MB once. After that you can dictate without an internet connection. Once or twice a day it briefly checks your license and looks for updates – only your license key goes over the line. Without a connection it keeps working for up to 7 days. No account is needed.
Stays on your computer
- Your voice recordings – turned into text, then deleted
- Your dictated texts
- Your learned words and word list
- Names read from the window you are replying to – in working memory only, never stored
Goes over the network
- A one-time download of about 550 MB after installation
- A brief license check once or twice a day – only your license key, server in the EU. Without a connection SayLocal keeps working for up to 7 days.
- The update check – new versions install only if they verifiably come from us (can be switched off)
- Only if you agree: anonymous usage figures (no audio, no text), an end-to-end encrypted copy of your word list for your second device, and cloud recognition on less powerful computers
System requirements
| System | Windows 10 (22H2) or Windows 11, 64-bit |
| Processor | x64 processor with AVX2 (Intel from 2013, AMD from 2015) |
| Graphics | recommended: graphics card with Vulkan support – much faster |
| Memory | 8 GB |
| Storage | about 1 GB free |
One license for PC and Mac
Your license covers two of your own devices – for example an office PC and a MacBook. SayLocal learns names and technical terms after a single correction, and if you want, the two devices sync their learned words end-to-end encrypted. Using a Mac too? See offline dictation on the Mac.
Frequently asked questions
Does SayLocal need an internet connection on Windows?
Only for the one-time download after installation and a brief license check once or twice a day. Dictation itself works offline, and SayLocal keeps running for up to 7 days without a connection.
Do I need a graphics card?
No, but it is recommended. With a Vulkan-capable graphics card your text appears after about 0.2 seconds; on the processor it takes a few seconds.
Which Windows versions are supported?
Windows 10 (22H2) and Windows 11, 64-bit, on an x64 processor with AVX2.
Can I change the dictation key?
Yes. The right Ctrl key is the default; you can set another key in SayLocal.
Does SayLocal send my audio anywhere?
No. The recording is turned into text on your PC and then deleted. Only if you explicitly switch on the optional cloud recognition on a PC without a suitable graphics card is it sent encrypted to Cloudflare for transcription and not stored.
More about SayLocal
Updated 19 October 2026 · Conto IT, Itzehoe
