Everything Lazytype does, in one place

This page describes every feature in the current release (v1.8.2). Prices and plans are on the pricing page; frequently asked questions live in the FAQ.

Hold-to-talk dictation (modifier-safe)

The core of Lazytype is one gesture: press and hold a key — left Ctrl by default — speak, and let go. On release, your speech is transcribed by OpenAI's Whisper large-v3-turbo running on Groq at roughly 216× real time, and clean, punctuated text is typed wherever your cursor is. A short sentence is back in well under a second, in any app with a text field: email, chat, documents, browsers, code editors, even the terminal.

The hotkey is modifier-safe: because dictation only starts on a sustained hold, quick taps and chords like Ctrl+C, Ctrl+V and Ctrl+Z keep working exactly as they always did. You don't give up your most-used shortcuts to gain dictation. The hotkey itself is configurable — left Ctrl, right Alt, Ctrl+Shift, F8–F10 and more than ten other combinations.

Unlike dictation tools that transcribe into their own window and make you copy the result over, Lazytype types directly into the app you're already working in.

Realtime preview

While you're holding the key and speaking, a floating preview bar shows your words arriving chunk by chunk — before you've even finished the sentence. Audio is streamed to the transcription engine during recording, so you can see where the text is going, catch a misheard word mid-sentence, and simply keep speaking to correct course. When you release the key, the final cleaned-up text is typed into your app.

The preview is a small overlay that follows your dictation and can be toggled on or off in the tray settings. It was added in v1.3 and refined since; it works with every engine mode and every language.

No other mainstream dictation tool — Wispr Flow and Superwhisper included — shows you a live transcript while you're still speaking; they only reveal the text after you stop.

100+ languages, auto-detected

Lazytype transcribes more than 100 languages, and you don't have to tell it which one you're speaking — the language is detected automatically per dictation. You can dictate an English email, answer a Dutch chat message, and drop a French quote into a document without touching a setting. Accuracy is strongest in English and Dutch, and the full list runs from Afrikaans to Yoruba.

Auto-detection matters for people who live in two or more languages a day, which is most of the non-English-speaking world. Built-in tools like Windows Speech Recognition and Apple Dictation typically require you to pick a language up front and switch manually; Lazytype just listens.

Translation mode

Translation mode turns dictation into simultaneous interpretation: speak in one language and Lazytype types the text in another. Speak Dutch, and polished English lands directly in your email — no translation tab, no copy-paste round trip. Any of the 100+ supported languages can be the source, and you choose the target.

The translation runs in the same pass as the AI clean-up, so filler words and stutters are removed at the same time. You can enable translation globally in the tray menu, or — more usefully — assign it to its own hotkey as a voice macro, so one key dictates in your language and another translates on the fly.

Among the major dictation apps this is rare: Superwhisper offers no translation, and Wispr Flow's is limited, while Lazytype treats it as a first-class mode.

Voice macros

Voice macros let you assign different AI behaviours to different hotkeys. One key dictates clean text. A second key translates as you speak. A third runs a command on the text you just wrote — hold it and say "make this shorter" or "make it formal", and Lazytype rewrites the last dictation in place. Each macro is a hotkey paired with its own AI instruction, configured in the settings dashboard.

In practice this means the same app covers three workflows that would otherwise need three tools: transcription, live translation, and voice-driven editing. All hotkeys are freely configurable, so the layout adapts to your keyboard and habits rather than the other way round.

Most competing dictation tools offer exactly one behaviour per app — dictate — which is why macros are one of Lazytype's clearest differentiators.

Context-aware tone

Lazytype detects which application is active and adapts its clean-up style to match. The same spoken sentence comes out casual in Slack, properly punctuated and formal in Word, and with technical terms preserved in a code editor. There is no manual mode switch — the app you're focused on is the mode.

The detection happens locally: Lazytype reads the active window title and process name on your machine to infer context, and that information is never sent anywhere. Only the appropriate clean-up instruction changes.

Dictation tools without this feature give you one register everywhere, which usually means editing the output by hand in half your apps — the exact typing you were trying to avoid.

Voice commands

A few spoken commands keep the whole flow hands-free. Say "scratch that" and the last dictation is deleted. Say "new line" or "new paragraph" and the formatting happens as you go. Combined with hold-to-talk, this means you can draft, structure and correct a message without your hands ever leaving their resting position.

Commands are recognised inside your dictation, so there's no separate command mode to enter and exit. This is the kind of detail that built-in dictation on Windows and macOS handles awkwardly or not at all, and it's what makes voice-first writing actually sustainable for longer sessions.

AI clean-up

Raw speech is messy: "ehm", false starts, repeated words, missing punctuation. Lazytype's optional AI clean-up pass removes filler words and stutters, fixes punctuation and capitalisation, and delivers text that reads like you typed it carefully — not like a court transcript. Say "send the the deck to maria before noon and cc finance" and what lands is "Send the deck to Maria before noon, and cc finance."

Clean-up is a separate, optional layer on top of transcription, so you can turn it off when you want the literal transcript. It also powers the translation and command macros, which reuse the same server-side pass.

Compared with plain speech-to-text tools that hand you an unpunctuated stream, this is the difference between dictation you publish and dictation you rewrite.

Offline engine (whisper.cpp)

Lazytype includes a fully offline, on-device transcription engine built on whisper.cpp. Switch to it in the tray menu (Engine → Local) and no audio ever leaves your machine — no network, no cloud, no account. It's included in every plan, including the €25 one-time Personal license.

The offline engine exists for two situations: places without a connection, and work where audio simply must not leave the device — doctors, lawyers, anyone under confidentiality obligations. In cloud mode, audio is sent only to transcribe a clip and is never stored; in local mode, it never leaves at all.

Few dictation apps ship any offline option — Wispr Flow has none — so having one bundled alongside the fast cloud engine, switchable per situation, is unusual.

Statistics dashboard

The built-in statistics dashboard (added in v1.8) shows what dictation is actually doing for you: words dictated, characters typed, and the time saved compared with typing them, per session and cumulatively. At an average typing speed of 40 words per minute and a comfortable speaking pace of about 130, the saved-time number grows quickly enough to be motivating.

The dashboard lives one tray-click away, alongside the rest of the settings: engine, language, post-processing, hotkeys and history. Nothing is buried in nested menus.

It's a small feature, but most dictation tools give you no feedback loop at all — you never learn what the habit is worth.

Config persistence across updates

Since v1.8.2, all of your settings — hotkeys, macros, engine choice, language preferences, history — survive app updates. Configuration is stored in your user profile (Windows AppData), separate from the application itself, so installing a new version never resets your setup.

This sounds like table stakes, and it should be: plenty of small utilities lose your configuration on every update, and a tool you've tuned with custom macros and hotkeys is exactly the kind of setup you don't want to rebuild. Update, keep working, nothing to redo.