Real-time voice typing for Mac

Your words, typed as you speak

Press one key and talk. Voicecape recognises what you say as you say it and lands it at your cursor in any app — fillers cleaned, your wording intact.

Download free See how it works

macOS · 14 days free, no card · then $9.9 a month.

Scripted preview

Everything in Voicecape

Real-time cloud dictation is the product. The rest is what makes it survive a working day.

Cloud dictation Real-time speech to text Clean modes Filler removed, wording kept Personal dictionary Names and jargon, spelled right Tone per app Code, terminal, chat or prose Quick note Capture without a window Backup Dictionary and settings, portable Offline fallback Keeps typing without network Accounts & devices One subscription, three Macs

All features →

491msp95, end of speech to final text — measured through our relay
5languages, detected automatically
2modes — Exact Clean · Verbatim
14days free — no card needed

Two modes, one rule

Your words, minus the noise of speaking

The same sentence, spoken once. Exact Clean removes what you didn’t mean to say. Verbatim keeps every sound. Nothing gets rewritten in either.

What you said

“Um, I think we should send it on Friday, after the the review, you know.”

Exact Cleanwhat lands at your cursor

I think we should send it on Friday, after the review.

um — filler , you know — trailing filler phrase the the — accidental repeat

Verbatimwhen every sound matters

Um, I think we should send it on Friday, after the the review, you know.

quotes · transcripts · legal notes

Features

Accurate voice typing, not AI rewriting

Straight to your cursor

No app switching, no copy-paste. One key, then talk.

Privacy is not a tier

Zero-retention terms apply on the trial and on the paid plan alike. Nobody has to move up a plan to get them.

Personal dictionary

Names, jargon, spellings — teach them once. The list is stored on your Mac. A cloud dictation carries it with the request as recognition hints, and the server keeps none of it.

Five languages, measured

English, Korean, Spanish, Japanese and Chinese. You never pick one before you talk, and a sentence that mixes two comes back mixed the same way.

Real-time, not batch

Text grows while you are still talking, instead of appearing after you stop. From the end of a sentence to its final text is 491ms at p95, measured through our relay.

Undo is one keystroke

Inserted the wrong take? It goes in as a single edit, so your app’s ⌘Z takes it straight back out.

The contract

Cleaned, never rewritten

Most AI dictation tools “improve” your writing. Voicecape treats your words as final — it only removes what you didn’t mean to say.

Voicecape removes

  • Fillers — “um”, “uh”, hesitation sounds
  • Accidental word repeats — “I I think”
  • Trailing filler phrases — “…, you know”
  • Extra spaces and messy line breaks

Voicecape never

  • Rewrites your sentences or “improves” your tone
  • Summarizes, translates, or reorders what you said
  • Adds greetings, sign-offs, or content you didn’t say
  • Changes names, numbers, or negations

Where it earns its keep

Built for people who type for a living

Writers & makers

First drafts at speaking speed

Dictate the messy first pass, keep your own voice. You edit your sentences — not an AI’s.

Typing hurts some days

Fewer keystrokes, same words

When RSI flares up, speak instead. What lands is your wording — no proofreading someone else’s rewrite.

Developers

The words around the code

Commit messages, code comments, PR replies — dictated while your hands stay where the work is.

Privacy

Same terms on every plan

  • Zero-retention terms apply on every plan — there is no privacy tier to upgrade into
  • Audio is processed in transit and not stored, under a data processing agreement we have signed
  • Every company that can see your speech or your text is named on this page, and in the policy
  • Cloud dictation is a setting — with it off, recognition runs on this Mac

Don’t take our word for it: turn cloud dictation off and dictation still works in airplane mode.

Where it runs

One app, on your Mac

A Mac app that listens on a keystroke and types where your cursor already is. No window to switch to, no recording to export, no paste at the end.

  • macOS 13 Ventura or laterApple silicon and Intel
  • Lives in the menu barOne global shortcut — no window in your way
  • No network? It falls backRecognition drops to this Mac and keeps typing
  • Five languages, detected automaticallyEnglish · 한국어 · Español · 日本語 · 中文

Languages

Five languages, nothing to switch

Voicecape works out which one you are speaking from what you say, and the cleanup afterwards is told never to translate — so a product name in another language comes back exactly as you said it, inside the sentence you said it in.

  • English

    Send it on Friday, after the review.

  • 한국어

    금요일 검토 끝나고 보내 주세요.

  • Español

    Envíalo el viernes, después de la revisión.

  • 日本語

    金曜日、レビューのあとに送ってください。

  • 中文

    周五评审之后再发。

Automatic by default

You do not pick a language before you talk. If you would rather be explicit, set one in Settings and it is obeyed exactly.

A mixed sentence stays mixed

If you drop an English term into a Korean sentence, the cleanup keeps both exactly where you put them. It is instructed never to translate, in either direction.

Measured, not asserted

Character error rate on our benchmark set: 4.4% for Korean, 2.3% for Japanese. Same synthesised audio for every engine, scored against the text a dictation product should produce — not against a transcript of the sounds.

Shipping, not planned

The three most recent builds. Every one of them is in the update feed the app itself reads.

All releases →

Pricing

14 days free, then $9.9 a month

Fourteen days, no card. After that, one price with everything in it — there is no tier above this one to be upsold into.

$9.9

per month · or $79 a year · macOS · 14-day free trial

  • Real-time recognition in five languages
  • Exact Clean and Verbatim modes
  • One subscription, up to 3 Macs
  • Personal dictionary for names, jargon and spellings
  • Zero-retention terms — the same ones the trial gets
  • Cloud fair use: 240 dictation sessions and 400 clean-ups per Mac, per day — local dictation has no limits
Download free Subscribe now

The trial is the full app — nothing is held back and nothing expires except the trial itself. Checkout runs on Polar, our merchant of record, and a refund within 14 days needs no explanation. Questions first? [email protected]

FAQ

Questions, answered plainly

Where does my voice go?

With cloud dictation on, the audio of that dictation goes to Soniox Inc., which turns it into text. It is processed in transit and not stored, under a data processing agreement we have signed. The text can then go to Cloudflare, Inc. to be cleaned up, with Google LLC, OpenAI, L.L.C. and Groq, Inc. behind it when the first is unavailable. Every one of those companies is named in the privacy policy, and the terms are the same whether you are on the trial or paying. With cloud dictation off, recognition runs on your Mac and none of that happens.

How is this different from AI writing tools?

AI dictation tools usually rewrite what you said into “better” text. Voicecape does the opposite: it removes fillers and accidental repeats, and keeps your words, order and meaning exactly. If a sentence needs rewriting, that stays your decision.

What happens without a network?

Recognition falls back to your Mac and dictation keeps working — on a flight, in a dead zone, on a locked-down network. Accuracy is not identical between the two paths, and the figures on this page describe the cloud one.

Which apps does it work in?

Any app with a text field. Voicecape inserts text at your current cursor position system-wide — email, documents, chat, browsers, code editors. In browsers and Electron apps insertion goes through your clipboard, and Voicecape puts back whatever you had copied straight afterwards.

Which languages are supported?

English, Korean, Spanish, Japanese and Chinese. You do not pick one before you talk: leave it on automatic and Voicecape follows what you say, and a sentence that mixes two comes back mixed the same way, because the cleanup is instructed never to translate. On our benchmark set the character error rate is 4.4% for Korean and 2.3% for Japanese — synthesised audio, scored against the text a dictation product should produce.

Why a subscription?

Because recognition and refinement run on servers that bill us by the minute. Every word you dictate costs something to produce, so a single payment could not honestly cover an unlimited number of them. It is $9.9 a month, or $79 a year — and the fourteen days before that cost nothing and ask for no card.

What happens when the 14 days are up?

The app stops inserting text. Your personal dictionary, your settings and anything you have already written stay exactly where they are, and starting a subscription picks up where you left off.

What if I get a new Mac?

One subscription activates up to 3 Macs at a time. If you have used all three and are retiring one, open Settings → Licence on that Mac and press “Release this Mac” — the slot frees immediately. If you no longer have it, email [email protected] and we release the old slot the same day.

What if it does not work for me?

Ask for a refund within 14 days of paying and you get one — no questions, and using the app during those 14 days does not disqualify you. The full terms are on the refund policy page.

How do I get support?

Email [email protected]. Support is handled directly by the team that builds Voicecape — there is no ticket queue to route through.

Talk instead of typing

Fourteen days free, no card. Dictation that keeps your words intact.

Download free

Questions? [email protected]