Say it once. It's already typed.
LecSync listens on a shortcut and writes into the window you are already in — cleaned up, punctuated, and exactly as long as what you said.
Included in the LecSync desktop app.
Message
What you said
Book the meeting room for Tuesday afternoon, um, no wait, make it Thursday. Invite Xiao Li and Xiao Wang, it's the quarterly review.
The only interface is that pill. The text lands in whatever app is in front of you — and one click brings back the raw transcript, word for word.
Works everywhere you write
Dictation types into the window you are already in. If you can type there, you can speak there — no plugins, no integrations.
Gmail
Notion
Claude
Cursor
Google Docs
WhatsApp
Telegram
Figma
Linear
Obsidian
Gemini
GitHub
WeChat
Discord
Jira
Perplexity
Trello
Zoom
Xcode
LINE
Todoist
Substack
Reddit
XDictate
It types what you meant, not what you said
Speech is messy: you pause, you backtrack, you change your mind halfway through. LecSync cleans that up before a single character lands in your document — and never adds anything you did not say.
Filler words disappear
Um, uh, and the long pauses in between never reach the page.
Only your final version survives
Say “no, change that to Thursday” and only Thursday is typed — even when the thing you corrected was two sentences back.
Spoken lists become numbered lists
Say “first… second… third…” and you get 1, 2, 3, with the counting words stripped out.
Numbers and spellings come out right
“Extension eight zero six two” becomes 8062. Spell a name out loud and it gets fixed in place.
Your own vocabulary
Add up to 32 names, products and technical terms. LecSync spells them your way instead of guessing.
Speak any language
The language is detected from your audio every time, so there is no mode to switch before you start talking.
Your details
Say “my address” and the address types itself
Phone number, email, shipping address — the things you retype every day. Fill them in once. After that you just talk, and the real value lands at the moment you actually hand it over. Most dictation tools stop at getting your words right; this goes one step further.
Filled in once
You dictate
Put my phone number on the form and ship it to my address.
LecSync types
Put +1 (415) 555-0132 on the form and ship it to 1600 Market St, Apt 5B, San Francisco.
It fills in only when you are actually handing the detail over. Say “my address changed last month” or “do you have my phone number?” and nothing is touched — when in doubt it leaves your words alone.
Built for AI tools
Your prompt stays a prompt
Dictating into ChatGPT, Claude or Cursor is what people use voice for most — and what most dictation tools quietly break. LecSync treats what you say as text to type, never as an instruction to follow.
You dictate into ChatGPT
Write me a Python script that reads a CSV and plots a line chart.
LecSync types
Write me a Python script that reads a CSV and plots a line chart.
Nothing leaves your hands
Speech is transcribed on our own GPU nodes. The audio produces your text and then it is gone.
Audio is never stored
Your speech is transcribed as it streams and then discarded. The dictation gateway never writes audio to disk.
History stays on your computer
Recent dictations are kept locally by default. Turn on sync only if you want them in your LecSync account.
The raw transcript is always yours
Every dictation keeps the untouched transcription next to the cleaned-up text, so you can always go back to your exact words.
How it works
Three keys and your voice
Press the shortcut
Any app, any text field. A small indicator tells you LecSync is listening.
Say it out loud
Talk the way you would talk to a colleague. Pauses, corrections and second thoughts are fine.
Press it again
About a second later the finished text is typed into the field your cursor was already in.
Questions
Which apps does dictation work in?
Any app on macOS or Windows that accepts typed text — chat, email, documents, code editors, browser text boxes, AI chat interfaces. LecSync inserts the text into the frontmost window, so there is nothing to integrate.
Do I have to pick a language first?
No. The spoken language is detected from your audio each time, and mixed-language speech is handled too. There is no mode to switch before you start talking.
Why does macOS ask for Accessibility permission?
Inserting text into another app requires macOS Accessibility access. If you would rather not grant it, dictation still works — you copy the last transcript from the indicator instead of having it typed for you.
What if the cleanup removes something I meant to keep?
Every dictation stores the raw transcription alongside the cleaned-up version, and one click restores it word for word. Recent dictations are listed on the Dictation page of the desktop app.
How fast is it?
Speech is transcribed while you are still talking, so letting go of the shortcut leaves only the last fragment to process. The text usually lands about a second after you finish.
Free yourself from the keyboard
Install the desktop app, pick a shortcut, and start talking.