Zen Whisper
MacThink it. Say it.
Written.
Dictate into Mac apps and transcribe audio with local-first speech recognition. Download speech packs, try 14 days of Pro and keep typing by voice.
Apple Silicon · macOS 14 or later
Let the words come naturally.
A reply, a rough draft, a thought you do not want to lose. Speak where you already work, and let the words land at your cursor.
A thought, already written.
See everyday dictation, word by word. This illustrative demo is silent and does not use your microphone.
Example transcription: Hi Maya, thanks for sending this over. I’ll review it this afternoon.
Hold, speak, release. Text lands in Mail, Slack, Notes, your editor, wherever you already were.
Hold your key
Right Command, Option, or Control. Ordinary typing is never interrupted.
Say the messy version
Speak naturally. Punctuation and capitals are handled for you.
Release
The text appears at your cursor. Nothing to copy, nothing to paste.
Your voice, where you work.
Types where you work
A global hold-to-talk shortcut inserts text into Mail, Notes, browsers, chat apps and editors automatically.
App aware formatting
WhatsApp and Slack stay casual, mail stays formal, terminals get light technical cleanup.
A conversation worth keeping.
Who said what
A recorded conversation comes back split by speaker, with names you can set, so an interview or a two-person meeting reads as a conversation rather than one block.
Captions you can export
The same media also comes back as timestamped caption blocks, so a video you transcribed for the notes can be the video you subtitle.
Considered down
to the details.
Local by default
Speech runs on your Mac with downloaded packs. Cloud engines are optional, off by default, and use your own API keys.
Matches the thread
Replying in a chat, it takes its cue from the last message. A casual thread stays casual.
Browser tab vocabulary
In Chrome, Arc and Safari it pulls the title, domain and issue keys from the tab you are on, so client names and ticket refs come out spelled right.
Dictionary and correction memory
Add names, brands and jargon. It also learns from corrections you keep repeating, so the same mistakes stop coming back.
Snippets and rewrite
Say a short trigger, get a long reply. Rewrite flows turn rough speech into something you can send.
Searchable transcripts
Sessions land in a timestamped history you can search later, tagged with the app you dictated into. Useful for drafting and meeting follow-ups, not just one-off dictation.
Meeting notes mode
Turns a long stretch of speech into structured lines and bullets instead of one unbroken paragraph.
Voice memos and media
Record a memo, or transcribe an audio or video file. No separate recorder, uploader and transcript toolchain.
110 languages
Including English regional variants, Hindi, Hinglish, and many regional-language workflows.
Real controls
Pick speech packs by speed, language or accuracy, and tune silence handling, punctuation and long-form behavior.
Light on your Mac
Lower memory Macs skip heavy prewarm, and models unload after quiet time instead of sitting in RAM all day.
Update notices
The app tells you when a newer build ships, so you are not checking a download page to find out what changed.
The way you actually speak.
Hinglish, in its own right.
Zen Whisper ships Hinglish as its own option, and returns Roman script rather than Devanagari. Said out loud, typed straight into the app you were already in.
110 languages. Their own scripts.
Pick a language or let it detect one. Text arrives in the script people actually write in, so there is nothing to transliterate afterwards.
Take a closer look.
Zen Whisper, on Mac. Explore every screen at your own pace.
A few steps.
Then just speak.
You do not need a paid key to start the trial. It is already built into the app.
- Download the app from Gumroad.
- Move Zen Whisper to Applications and open it.
- Grant Microphone and Accessibility permissions when prompted.
- Download at least one speech pack while onboarding.
- Bought a seat? Enter your Gumroad receipt email and license key in Preferences, under Zen Whisper Pro.
Before you
make it yours.
Core transcription runs on your Mac. Licence checks and optional features connect online. Here is what you need to get started.
- Mac
- Apple Silicon, M1 or newer
- macOS
- 14 Sonoma or later
- Microphone
- Permission to capture audio
- Accessibility
- Permission for the global shortcut and automatic text insertion
- Internet
- For the first model download, license validation, and any optional network feature you turn on
- Disk
- Several gigabytes free if you want the larger local speech packs
Start free.
Make it yours.
Try it for 14 days at no cost. Then keep 3,000 free words a week, or buy a lifetime seat. Seats are one payment, never a subscription.
One year of updates is included. The version you have keeps working after that.
Switching from Apple Dictation14-day money-back guarantee. If Zen Whisper does not work on a supported Apple Silicon Mac, email hello@zenproducts.ai within 14 days with your macOS version and a short description. Approved refunds deactivate the associated Pro license.
When does the trial start?
On your first transcription, not at install. A slow first model download does not eat into it.
What happens after 14 days?
Nothing locks you out. Transcription drops to the free tier of 3,000 words a week, and Writing polish and Rewrite switch off. Everything you already dictated stays where it is.
Does my audio leave the Mac?
Not for core dictation. Cloud engines are opt-in and only ever use API keys you add yourself.
What does the personal dictionary do?
It steers transcription toward the names, brands and jargon you actually use. Star a row to prioritize it, and scope a word to particular apps or leave it on everywhere. The whole list stays on this Mac.
What are snippets?
Say a short trigger, get a long reply. A signature, an address, a standard response. Dictation, voice memos and media import all honor the same list, and nothing leaves this Mac.
Can it transcribe files I already have?
Yes. Drop in MP3, WAV, M4A, MP4 and friends, or paste a YouTube or Instagram link. Links always transcribe on this Mac. Finished jobs wait in Transcripts until you open them.
How are voice memos different from dictation?
Dictation types straight into whatever app you are in. A memo records first and stays on the Voice memos page rather than in Transcripts. Transcribe it on device right away, or keep the audio and come back to it later.
Which speech engines can I use?
This Mac is the default, and on it nothing leaves the machine. Sarvam, OpenAI, Deepgram, Groq and Gemini are optional and run on API keys you add yourself.
How big is the speech pack download?
Standard is about 1.5 GB and is the default for everyday work. Precision is about 2.9 GB and earns its size in noisy rooms, with strong accents, and when every word has to be right.
What do I need?
An Apple Silicon Mac on macOS 14 Sonoma or later, plus Microphone and Accessibility permissions. The full list is under "What you need" above.
