Features

Dictate anywhere, fully on your computer.

Push-to-talk in 90+ languages, subtitles for any audio or video file, and translation into 29 languages. Every feature runs on your own machine.

Free forever · No account · 100% on-device

Unlimited dictation. No credits, no caps.

Dictate anywhere, — the app in action
Everything you need

One small app, every dictation feature.

All running 100% on your computer — no cloud, no account.

Push-to-talk & toggle

Hold a hotkey to dictate, or toggle hands-free — text lands at your cursor.

File & video transcription

Drop in audio or video, get a clean searchable transcript (Pro).

Subtitles from any file

Drop in audio or video and get .srt, .vtt, a video with a subtitle track, or subtitles burned into the picture. Subtitles are part of Pro and the 30-day trial.

Translation on your device

Get a file, or a dictation slot, back in any of 29 languages, with nothing sent to a server (Pro).

Speaker labels + timestamps

Diarization and timecodes for meetings and interviews (Pro).

90+ languages, offline

No internet needed after the first model download.

Vocabulary support · 29 languages

Teach it your names, jargon and acronyms for sharper accuracy (Pro).

100% on-device

Audio never leaves your computer — no account, no servers, no telemetry.

Subtitles

Subtitles for any file, in four formats.

Drop an audio or video file in and set the output to Subtitles. You get a subtitle file (.srt or .vtt), a video with a subtitle track, or the words burned into the picture. The whole job runs on your computer, so the file is never uploaded.

  • Subtitle files in .srt or .vtt
  • A video with a subtitle track, or the words burned into the picture
  • Subtitles are part of Pro and the 30-day trial. The free version transcribes files up to 3 minutes.

Transcribe a file › drop the file in › Output: Subtitles › pick the format

Coming to Windows, following the Mac release

Subtitles mode with a video file and the four output formats listed
Subtitles for any file,in four formats.
On-device translation

Text in a language you did not speak.

Set an output language other than the one that was spoken and the file comes back translated. Subtitles and transcripts cover 29 languages, and each picker marks the weaker pairs as beta or reduced accuracy. The translation happens on your own machine, so no audio and no text goes to a server.

  • 29 languages for subtitles and transcripts
  • Weaker pairs are marked beta or reduced accuracy in the picker
  • Pro and the 30-day trial; the free version does not translate

Output card › Output language

Coming to Windows, following the Mac release

Output card with the output language set to a different language from the spoken one
Text in a languageyou did not speak.
One key, two languages

Speak one language, type another.

Give a slot an output language and dictation in that slot arrives translated, in whatever app you are typing in. Speak the language you think in and the other one lands at your cursor. The overlay names the pair while you record, for example EN to TR.

  • Works in any app: mail, chat, documents, forms
  • The overlay names the pair while you record, for example EN to TR
  • Pro and the 30-day trial; the free version types what was spoken

Slots › pick a slot › Output language

Coming to Windows, following the Mac release

Recording overlay showing the pair EN to TR while text lands in a mail window
Speak one language,type another.
Slots

Five slots, five ways to work.

Each slot carries a spoken language, a model and an output language. Pro and the 30-day trial have five slots, the free version has one. Set up a slot for each language you work in and pick the one you need.

  • A spoken language, a model and an output language in every slot
  • Five slots on Pro and the trial, one slot on the free version
  • Give a slot an output language and its text arrives translated

Slots tab

Coming to Windows, following the Mac release

Slots tab with five slots, each showing a spoken language and an output language
Five slots,five ways to work.
Long recordings

A 90-minute film, around 1 GB.

A 90-minute film stays at around 1 GB of memory while it transcribes, so a feature-length recording runs on the machine you already have. Pro takes files of any length. The free version covers files up to 3 minutes.

  • A 90-minute film stays around 1 GB
  • Any length on Pro and the 30-day trial
  • The whole job runs on your computer, with nothing uploaded

Nothing to set. It applies to every file you transcribe.

A 90-minute video file transcribing with a memory reading near 1 GB
A 90-minute film,around 1 GB.
Model & performance

Built on Whisper, running fully on your device.

OpenAI's Whisper speech models run on-device with WhisperKit and CoreML — on Mac (Apple Neural Engine and GPU), and on AMD, Intel or NVIDIA GPUs on Windows. Nothing is uploaded.

Transcribes 90+ languages entirely on your device, with advanced vocabulary tuning for the 29 most-spoken. See and their accuracy tiers.

macOS 14 (Sonoma)+Apple Silicon (M1 or newer)Mic + Accessibility permissionWindows 10/11 (AMD · Intel · NVIDIA)
Built on Whisper,
98%accuracy on clear speech (largest model)
15×faster than real time, on-device
90+languages, offline
No servers100% on your device
300 MB–6 GBspeech model — base to Large-v3 (downloaded once)
Freeforever — no account or card
Unlimited dictation, free forever Talk or transcribe files as much as you want
How it works

Three steps, zero friction.

Press a hotkey, speak, and your words land as text wherever your cursor is.

1

Press the hotkey

Press your hotkey from anywhere on your computer.

2

Speak naturally

Talk the way you think — in any app, in any of 90+ languages.

3

Your text appears

Pasted at your cursor instantly — nothing ever leaves the device.

Get the app

Try every feature free.

Free forever, no account, no card. Runs 100% on your computer — your voice never leaves the device.

No credits · No limits

Download for Mac

v7.19.1

macOS 14 Sonoma+ · Apple Silicon

  • Unlimited on-device dictation
  • No account, no card, no telemetry
  • Upgrade to Pro anytime in-app

Windows

Windows 10 / 11 Windows: subtitles and translation coming soonv7.16

Native build for Windows 10 & 11 — runs on AMD, Intel and NVIDIA GPUs. NVIDIA needs the latest GPU driver + CUDA runtime; AMD & Intel just need up-to-date drivers.

Subtitles, on-device translation and five language slots are on the Mac build. The Windows build with these features follows the Mac release.

  • Unlimited on-device dictation
  • No account, no card, no telemetry
  • Upgrade to Pro anytime in-app
Download for WindowsWindows 10 / 11

Free forever. Pro is $4.99 a month, or $39 once for a lifetime licence.

macOS 14+ · Apple Silicon Windows 10 / 11 · AMD · Intel · NVIDIA NVIDIA: latest driver + CUDA Mic + Accessibility 300 MB–6 GB model, downloaded once