EUGermany-based. GDPR-Compliant by design.
Desktop app

AI Voice Typing for Windows

Dictāta turns speech into clean, formatted text in any Windows app. Use Cloud mode, or run speech recognition on your PC with Local mode, with no GPU required.

Download for Windows

64-bit Windows (Intel/AMD or ARM).

What Dictāta does on your desktop

  • Dictate into any app

    Press a hotkey, speak, and the text appears at your cursor in your email client, browser, editor or chat app.

  • Prompt Styles

    Turn a spoken thought into an email, a bullet list or a ticket. Cloud mode, or Local mode with an optional on-device text model.

  • Translate while dictating

    Speak in one language and get text in another at the cursor. This needs Dictāta Cloud.

  • Smart Glossary

    Save names and terms once. In Local mode they are also passed to the speech decoder as hints.

Also in the desktop app (Cloud mode, beta): transcribe audio and video files up to 3 hours each. Read about File Transcription

Local mode: speech recognition on your computer

Local mode keeps recognition on your PC. We measured about 122–126 ms from stop to final text on a Ryzen 7 5700U laptop CPU with a replayed recording (speech model only).

  • Speech recognition runs on your device with NVIDIA Nemotron 3.5 Streaming (28 languages).
  • In our tests the final text appears about 120–200 ms after you stop speaking (Windows laptop CPU, M1 Mac, Google Pixel 9a).
  • While Local is active, audio and dictated text are not sent anywhere. The app only allows sign-in, billing, model downloads and the version check.
  • Included with the Lifetime license, every paid Cloud plan and the free trial.
  • Current limit: spoken numbers are written as words for now. Chat, translation while dictating and Memory need Dictāta Cloud.
Read the full Dictāta Local overview

Dictāta Local requirements

Platform
64-bit Windows (x64 or ARM)
Speech model download
about 475 MB
Memory (RAM)
6 GB minimum, 8 GB recommended
Free disk space
about 1.2 GB
GPU
not required

Cloud mode needs an internet connection. Optional local text models are 1.1–3.3 GB each.

Frequently asked questions

Windows voice typing (Win+H) inserts what you say into the active text field. Dictāta adds Prompt Styles, a Smart Glossary, translation while dictating (Cloud) and a Local mode with a documented network lock.

No. It runs on the CPU. The measured 122–126 ms came from a Ryzen 7 5700U laptop CPU with 4 threads.

Yes, in Local mode. Speech recognition runs on your device after a one-time model download of about 475 MB. Cloud mode, chat, translation while dictating and Memory need an internet connection.

28 languages with the NVIDIA Nemotron 3.5 Streaming model. Cloud mode supports more; see the supported languages in the docs.

Yes. Local mode is included in the free trial, the Lifetime license and every paid Cloud plan. Current prices are on the pricing page.

Start dictating on your PC

Try Dictāta free. See the pricing page for plans.

Download for Windows

Dictāta on other platforms: macOS · Windows · Android · iOS

Voice Typing for Windows — Dictāta Dictation App