01
Press the shortcut
In any app’s text field, press the shortcut you set up.
AI voice dictation · Mac, Windows, iPhone and Android
Say it.
Say it.
It’s written.
Hold the shortcut, speak, let go — one to two seconds later your words appear right where the cursor is. WeChat, LINE, WhatsApp, Notes: any text field works.
Download link TBD All four platforms. Same app.
Demo: saying “Um… so, let’s meet at three tomorrow — I mean, three thirty” becomes “Let’s meet at three thirty tomorrow.”
02
Just talk
Say it however it comes out. Stumbles, repeats and filler words are fine — you are not typing while you think.
03
Let go, and it’s there
One to two seconds later, the cleaned-up text appears where your cursor was.
Features
More than speech-to-text
Built-in dictation hands you a raw transcript and leaves the tidying to you. SayType uses the one or two seconds after you let go to turn it into something you can send as it is.
Let go — it’s already in the cursor
Finish talking, let go, and the text lands where you were writing. No window switching, no copy and paste, no separate dictation app to open first.
Filler words cleaned up
No need to delete “um”, “uh” or “I mean” yourself. AI smooths the phrasing, adds punctuation, and splits long stretches into paragraphs.
Chinese and English, mixed
Switch languages in the middle of a sentence and it still gets it right — no input-method switching halfway through.
Numbers and idioms come out right
Say “three thousand five hundred” and get 3500. Say an idiom and it stays an idiom — it won’t be forced into digits.
Learns the words you use
Add the names, companies and technical terms you use often to your personal dictionary, and recognition gets sharper. Fix a word once and it learns your spelling.
Audio is only turned into text
Recordings are only used to produce text. We don’t train AI on your voice, and you can delete dictionary entries whenever you want.
Use cases
Less typing, everywhere you already type
Living in Taiwan, Hong Kong, Malaysia, Singapore or North America means switching between languages and apps all day. SayType behaves the same way in every text field.
Replying on WeChat and LINE
“On my way, two minutes” is a single sentence. Say it, send it — far quicker than tapping it out on a phone keyboard.
Also works with WhatsApp and Telegram
Talking to clients
Quotes, delivery dates, follow-ups — speak and it’s written. Put client and product names in your dictionary so they never come out wrong.
WhatsApp Business, email
Answering email
Say the whole thought, let AI turn it into clean sentences, adjust two words and send. Ten minutes before a meeting is enough for three replies.
Gmail, Outlook, work email
Longer notes
Say whatever comes to mind; long stretches get split into paragraphs. Staring at a blank page takes longer than writing — talk first, trim later.
Notes, Notion, documents
Meeting notes
Narrate the key points as you listen and they land in the document — no reconstructing from memory afterwards.
Minutes, class notes
Any text field
Address bar, admin panels, code editors, AI chat boxes — anywhere you can type, you can talk the same way.
No plugins, no new workflow
Privacy
What you say is only turned into text
Dictation is personal: messages to family, quotes for clients, drafts you haven’t figured out yet. So we only do what’s needed — and we’re clear about what we can and can’t promise.
Recordings are only turned into text
Never used to train AI. What you say does not enter any training data of ours.
You can delete your dictionary anytime
The names, companies and technical terms in it are added by you — remove them whenever you like.
We don’t say “completely secure” — nobody can promise that. The points above are what we actually do today. Full details on data handling, retention and third-party services will be in the published privacy policy (TBD).
FAQ
Questions you might have
Which apps does SayType work with?
Anywhere you can type: WeChat, LINE, WhatsApp, Telegram, email, Notes, documents, browsers, admin systems, even AI chat boxes.
It isn’t tied to any app, because the text goes straight into whatever field your cursor is in — not into a separate window you copy from.
Does it need an internet connection?
Yes. Audio has to be turned into text and sent back to your cursor, so it won’t work offline.
Connection quality affects speed; on a slow link it can take longer than one or two seconds.
Which languages are supported?
Chinese and English, and you can mix them in one sentence without switching input methods.
Support for other languages: TBD.
How is this different from built-in dictation?
System dictation gives you what you actually said — including “um”, “uh” and the false starts. You still have to clean it up.
SayType gives you the tidied version: fillers gone, punctuation added, paragraphs split — and it remembers your names and terms.
How does the personal dictionary work?
Add names, companies, products and industry terms. Recognition then leans towards those spellings instead of similar-sounding words.
Words you correct yourself are remembered: next time the same word comes up, it uses your spelling.
How much does it cost?
Pricing is still being finalised: TBD.
What happens to my audio?
Recordings are only turned into text and never used to train AI. You can delete dictionary entries at any time.
More detail in the privacy section above.
Which devices are supported?
All four: Mac, Windows, iPhone and Android. Installers are in final testing, so download links are TBD.