Direct dictation into almost any suitable Windows input field
LPS Software · Local AI Productivity
LPS Windows Diktat
Note for Windows users – 14-day trial
Our Windows installers are digitally signed using Microsoft Artifact Signing (Public Trust). With new or not yet widely used installer versions, Microsoft Defender SmartScreen may still display “Windows protected your PC” when an installer is launched for the first time.
If you downloaded the installer directly from leadershipperformancesystems.com, you can continue via “More info” → “Run anyway”. Leadership Performance Systems is shown as the verified publisher.
You can also verify the digital signature via right-click the EXE → Properties → Digital Signatures.
Local AI-assisted speech recognition with direct F9 dictation into Word, Outlook, browsers and other applications.
Version 1.0.0 is available now.
Version 1.0.0Windows 11DE / EN / AUTO€59
Local AI ProductivityPlace the cursor, press F9 and dictate: LPS Windows Diktat provides both start/stop and live/streaming in a single Windows x64 application. Processing is local; a suitable GPU can accelerate recognition.
Purpose
The product in its working context.
LPS Windows Diktat transfers spoken text directly into the previously active Windows application and combines high recognition quality with a simple global F9 workflow.
Core value
Place the cursor, press F9 and dictate: LPS Windows Diktat provides both start/stop and live/streaming in a single Windows x64 application. Processing is local; a suitable GPU can accelerate recognition.
Key benefits
Concrete value rather than a feature list.
Start/stop and live/streaming in a single application
Local processing after model installation with automatic German/English handling
Key capabilities
From the model to a usable result.
- Global F9 start and stop directly in the target application
- Start/stop dictation for complete transcription after recording ends
- Live/streaming for ongoing recognition while speaking
- Local PC performance test on first launch with a recommendation for suitable recognition quality
- Available graphics hardware is tested as well; GPU acceleration is recommended only when a measured performance advantage is detected
- Enabling GPU acceleration remains a deliberate user decision
- CPU processing remains available without enabled GPU acceleration
- Automatic German/English language handling
- Review mode for corrections before insertion
- Tray status for ready, recording and processing
- Local model storage and reuse of downloaded models
- Noise gate for short impulse sounds and protection against repetition loops
Applications and target groups
For professional analysis and design tasks.
Professional Writing
Dictate longer texts, emails and documents without changing applications.
Teaching & Research
Capture manuscripts, feedback and notes efficiently.
Accessible Workflows
Reduce keyboard input and complement workflows with speech control.
Application areas
Key functions at a glance.
These views illustrate key work areas and typical steps in using the application.
F9 dictation, language and quality modeLPS Windows Diktat
Tray status, target application and local processingLPS Windows Diktat
System requirements
Local and controlled use.
- Windows 11 · x64
- Recommended minimum hardware: x64 processor with at least 4 CPU cores and 8 GB RAM
- A suitable GPU is recommended for faster processing
- Suitable AMD, Intel and NVIDIA graphics solutions can be recommended for acceleration by the integrated hardware check
- SSD and sufficient free storage for the selected speech models recommended
- Microphone
- Internet connection for installation/initial setup and the first model download
Hardware check on first launch. Speech recognition and processing of your dictation take place locally on your computer. On first launch, LPS Windows Diktat runs a performance test and recommends a recognition quality suited to the measured PC performance. The recommendation can be applied with one click; the performance test can be run again when needed.
Optional GPU acceleration. If graphics hardware is available, its performance is tested as well. If it shows a clear speed advantage on that PC with a comparable recognition result, LPS Windows Diktat recommends GPU acceleration. You decide whether to enable it. Without enabled GPU acceleration, recognition runs on the processor.
Device-dependent speed. Processing speed depends on the hardware and the selected recognition quality. On less powerful computers, the application recommends a smaller model. The free 14-day trial is the best way to verify whether speed and recognition quality meet your needs on your own PC.
Model download on first use. The required Whisper models are downloaded over the internet when first used and then stored locally. Depending on the selected quality level, several hundred MB up to about 1 GB may be transferred. Models already stored locally do not need to be downloaded again.
- Start/stop: depending on performance, available models include
small-q5_1,medium-q5_0,large-v3-turbo-q5_0andlarge-v3-q5_0. - Live/streaming:
base-q5_1andsmall-q5_1are particularly suitable for lower latency; larger models can be selected on faster systems.
Technical model source: LPS Windows Diktat uses native whisper.cpp models. Optional GPU acceleration is based on the Vulkan backend of the whisper.cpp/GGML project used by the application. Model files are obtained through the designated download provider when first required; the audio being transcribed is not sent to the model download provider.
If Microsoft Edge WebView2 is not yet available on the Windows system, the required runtime may be downloaded during installation.
Trial and licensing model
One-off payment with no automatic renewal.
14-day trial licence · one named user · up to two own devices · no automatic renewal.
Licence status and updates: When an internet connection is available, the application may check the status of the installed licence and the availability of a newer program version covered by the existing product and licence scope in the background with the LPS system. If such a version is available, the application displays this and offers a download. You decide whether and when to download and install the new version; installation is not automatic. Alternatively, the current installer can be obtained through the download area provided for the product. Without an internet connection, this supplementary check is not performed; normal desktop use with a valid local licence remains possible. Downloads, customer-portal access and individual product-related online functions may require an internet connection.
Launch price through 31 December 2026
1-Year License
Sales start: Online purchases from 1 September 2026.
Before purchase: Purchase & Licence Terms · Refunds & Withdrawal · Privacy. Payment and tax processing is handled by the Merchant of Record identified in checkout.
Before purchasing: Processing speed depends on the individual PC. A suitable GPU is recommended for faster processing, but it is not mandatory. LPS Windows Diktat tests the CPU and available graphics hardware on the individual system and provides a recommendation. Use the free 14-day trial to verify speed and recognition quality on your own PC. Details on hardware testing and model downloads.
1 year
59 €
Time-limited one-off licence with no automatic renewal.
Online purchases and the public 14-day trial are available now.
After a successful purchase, the customer portal, licence and protected installer download are provided for the purchased entitlement.
Frequently asked questions
Important questions before trial and purchase.
Does the application require a specific graphics card?
No. LPS Windows Diktat can run speech recognition on the processor. A suitable GPU is recommended for faster processing. On first launch, the application checks the available hardware and recommends GPU acceleration only when it provides a measured performance advantage on that PC. Enabling it remains your decision.
Is audio sent to a cloud service?
The speech recording and transcription itself take place locally on your Windows PC. An internet connection to the download provider (Hugging Face) is used only when a required speech model is downloaded for the first time; the speech recording to be transcribed is not sent to the model download provider as part of that download.
Which applications can receive dictated text?
The global F9 workflow is designed for suitable input fields in applications including Word, Outlook, browsers and PowerPoint.
Next step
Try Windows Diktat free for 14 days.
The 1-year licence is available online. The customer portal, licence and protected installer download are provided after a successful purchase.