DiktatApp for Android 8 or newer (64-bit), plus PC receiver and tablet program for Windows. No account, no ads.
- One big button: record, stop – then check the recognised text and send it.
- Speech recognition runs on the phone. Only finished text travels over the network – nobody gets your voice.
- To your PC: the text is in the clipboard, just press Ctrl+V – in the AI chat, in Word or in an email. The app finds the PC on your Wi-Fi by itself.
- No PC needed: as a voice note in the todo list on your phone, sorted by project, with an archive.
- Or by email straight through your own mailbox.
The app itself is in German; the buttons are self-explanatory.
Talk to AI – without handing over your voice
Speaking is faster than typing, especially for longer instructions to an AI assistant. But the voice input of AI services and many phone keyboards sends your recording to the provider’s servers, often in the USA. And your voice is a biometric feature: recordings can be turned into a voice profile that recognises you.
Diktat keeps these apart: you speak into your phone, Whisper turns it into text right there, and only that text lands in your PC’s clipboard. Press Ctrl+V to paste it into the chat. The AI gets your words – never your voice.
- PhoneYou speak – Whisper recognises the text on the device. The audio stays here.
- text only, on your own Wi-Fi
- Your PCThe text is in the clipboard and in the receiver’s history.
- Ctrl+V
- AI chatgets text as if typed – no recording, no voice profile.
How the text gets where it belongs
Audio and recognition stay on the phone. Only the text you approve moves on.
- 1
Record
Pick a target, press the button, speak, stop. Recording continues with the screen locked.
- 2
Recognise on the phone
Whisper turns the recording into text right on the phone – with punctuation, without internet.
- 3
Check and send
You see the text, correct it if needed and send it: to a PC’s clipboard, as a todo or by email. Or record again.
Voice notes as todos – no PC needed
The app is useful on its own too: instead of going to a PC, your dictation goes into a todo list on the phone.
- Pick a todo project as the target, speak, check the text, save – and the note is done.
- Projects such as Garden, Office or Shopping keep tasks apart. Dictate, type, edit, move.
- Ticking off moves a task into the project’s archive – where you can restore it.
- Email the open todos; delete a whole project with all its entries once it’s done.
- Everything stays on the phone – no account, no cloud.
What Diktat does
- One button, then a check
- Record and stop on one big button, plus pause and resume. After recognition the text is shown for checking: send, delete or repeat.
- Recognises while you speak
- Long dictations are split at pauses and recognised in the background. After you stop, usually only the last piece is left – the text is ready for checking shortly after.
- Whisper on the phone
- Recognition uses whisper.cpp with the large-v3-turbo model right on the phone. The app downloads the model (about 550 MB) once; after that everything works offline. Smaller models are faster.
- Straight to the PC clipboard
- The Diktat receiver runs in the PC’s notification area. The app finds it on your Wi-Fi; pair once with a 4-digit code – done. Several PCs are possible; you pick the one you want.
- History on the PC
- The receiver keeps every text. A double-click puts an older text back into the clipboard.
- Todo with projects
- Create projects and dictate tasks straight into them. Done items move to the project’s archive. A project can be deleted with all its data.
- Email straight from the app
- The app sends through your own mailbox. Recipients build up a list; the last one used is preselected.
- Nothing gets lost
- If the PC is off or there is no network, the text waits on the phone and goes out automatically as soon as possible.
- Windows tablets too
- A separate program for Windows: one button, recognition on the tablet, text for checking, sent by email.
Your voice stays on the device
Most dictation features and AI voice modes send the recording to the provider’s servers. Diktat recognises everything itself.
- Audio never leaves the device: recording and recognition happen on the phone or tablet.
- Text only to your targets: your paired PC on your own network, or through your own mailbox.
- No voice profile: no AI provider, Google or Apple ever receives a recording of your voice.
- No account, no cloud: there is no server of mine in between.
- One-time model download from Hugging Face – only the model is downloaded, nothing is sent.
Set up in three steps
Set it up once, then just speak. The detailed guide (in German) is in the project’s README.
App and speech model
Install the app and download the speech model on first start – preferably on Wi-Fi, it is about 550 MB. For the large model 4 GB of RAM are recommended; on older phones pick a smaller one.
Pair the PC receiver
Start the Diktat receiver on the PC (Windows 10/11, no installation). In the app choose Ziel (target) → “PC suchen und koppeln” and enter the code the PC shows.
Email (optional)
Under Einstellungen → E-Mail-Versand enter your mailbox’s SMTP details and send a test mail. Todos need neither PC nor email.
Note: between phone and PC the connection on your own network is not encrypted, but only paired devices may send. Email is the way to go when you are away.
Download
Everything is free. For todos and email the app is enough; for the clipboard add the PC receiver.
App
Recording, speech recognition, todo projects, sending to PC and by email.
Android 8 or newer, 64-bit
Downloaddiktat.apk · Version 5.5 · 6,4 MB · Updated 26.10.2026
New: Mit Wave
PC receiver
Puts texts from the phone into the clipboard and keeps a history. A single program, no installation.
Windows 10/11
Downloaddiktat-empfaenger.zip · Version 1.0 · 35 KB · Updated 26.10.2026
Tablet program
Dictation on a Windows tablet: one button, recognition on the device, check the text, send by email.
Windows 10/11, 64-bit
Downloaddiktat-tablet.zip · Version 2.0 · 1,6 MB · Updated 26.10.2026
Install the app
- Open this page on your phone and download the app.
- Open the file diktat.apk, via the notification or the “Downloads” folder.
- The first time, Android asks whether the browser may install apps from unknown sources. Allow it, go back, tap Install.
- If Google Play Protect reports an unknown app: Install anyway.
- Open the app, download the speech model, allow the microphone on your first dictation.
Updates simply install over the existing version. Todos, settings and paired PCs are kept.
Windows may warn on first start (“Windows protected your PC”) because the programs are not signed: More info → Run anyway. The firewall asks once for the PC receiver – allow “Private networks”.
Questions and feedback are welcome at admin@duko.de.