Windows 10 / 11 · 14-day trial
Speak. The text writes itself.
One key. Your voice. Text, here, on this computer.
Voice Dictation is the Windows app that turns what you say into text. Push-to-talk (hold F9), toggle (start and stop) or hands-free (confirm at silence and continue). French, English, Spanish.
Enable LIVE in Settings → General: in hands-free or push-to-talk mode, confirmed words are added progressively to the chosen field while you speak, without waiting for a pause. Keep the cursor in that field and avoid typing during dictation. The preview can change; pasted text stays in place. A pause or F9 completes the remaining words. Recognition latency depends on the model and your PC, especially for translation. Toggle keeps text in the notepad, which also retains the complete phrase.
Speak French, write in English. In Settings → General, choose French as the spoken language and English as the written language. Translation runs on your PC, including the LIVE preview. Choose “No translation” to keep the spoken language.
Parakeet v3 FP32 for French dictation. An optional local CPU engine with about 2.55 GB of weights downloaded on first use: smaller than Whisper large-v3 (3.09 GB), larger than Whisper small (486 MB). English translation and other spoken languages automatically use Whisper.
The speech engines run on your PC: audio is not sent to a transcription service. Licensing, updates and initial downloads use the network, as described below.
Open-source components and LGPL rights
The corresponding sources for the distributed components are available free of charge from the same server as the installer: Qt Base 6.11.2, Qt SVG 6.11.2, PySide 6.11.2 and FFmpeg 8.1.
The app's terms do not restrict LGPL rights to modify or replace these libraries, relink the app with modified versions, or reverse engineer to debug those changes. Licence texts and notices are included in the installation and available through “Licences des composants…”.
Install
A Windows installer. No cloud account. The 14-day trial starts on its own.
Press
F9 by default — or the key you choose. A beep: it is recording.
Speak
Audio is transcribed on your PC. Automatic mode uses an available NVIDIA GPU or falls back to CPU. LIVE is optional.
Read
The text is written where you were — or in the dictation window, to copy, edit, undo.
Trial
Free, no card. All features. Included in the download.
Lifetime
One payment. After confirmed payment, copy the code from the confirmation page into Licence… → Activate.
BuyPrivacy. Audio recognition stays on your PC, with no cloud ASR service. Licensing and updates contact licence.le-sieur.ca with a pseudonymous machine fingerprint and licence/version information. Technical crash reports can be disabled in Settings → Advanced. Initial model downloads use Hugging Face; NVIDIA libraries use PyPI and its file servers. Payments use Stripe. No advertising tracking or usage analytics. The clipboard is used for local pasting.