How AI dictation works in Stenograph
You tap once, speak the way you would to a colleague, and pause. A moment later clean, punctuated text with your own names and terms sits exactly where your cursor was, in the app you were already using. This page walks through every step, from the first tap to meeting notes and file transcripts.
Listening
Tap, speak, pause
The whole interaction is one gesture. Press the global hotkey on Mac or Windows, tap the button in the Chrome extension, or tap the microphone key on your phone keyboard. Speak your sentence or your whole paragraph. When you stop talking for a natural moment, or tap a second time, the utterance ends and Stenograph starts writing.
- One tap or one hotkey starts listening
- A natural pause or a second tap ends the utterance
- No windows to manage and nothing to switch to
- The microphone stays off between utterances, with no always-on listening
Listening, processing, inserted
You always know what Stenograph is doing, because a small indicator next to your cursor shows one of three states. Listening means the microphone is on and your words are being captured. Processing means the audio is being recognized, your terms applied and the text formatted. Inserted means the finished text is in place and the microphone is already off again.
- Listening, shown with a live level meter while you speak
- Processing, usually a short moment after you stop
- Inserted, with the text placed at the cursor and ready to edit
Text lands at the cursor in the app you use
There is no separate editor to copy from. The desktop app types the result into whatever field has focus: an email reply, a contract draft, a chart note, a chat message, a commit description. In the browser the Chrome extension does the same for web apps, and on iPhone and Android the Stenograph keyboard inserts text in any app that shows a keyboard.
- Works in email clients, word processors, chat tools and browsers
- Works in code editors and terminals, with identifiers kept intact
- Keeps your undo history in the target app, so one undo removes the inserted text
Your custom terms are applied every time
Before the text is written, Stenograph checks what it heard against your personal dictionary and the professional vocabulary you selected. A client called Okonkwo, a citation, a drug such as apixaban or a function name such as getUserById comes out spelled the way you entered it. Teams share one dictionary so the whole firm writes names the same way. Learn how to build a term list on the custom vocabulary speech recognition page.
- Personal dictionary for names, products and abbreviations
- Professional vocabularies for legal, medical, software and finance work
- Shared dictionaries on Team and Enterprise
Formatting that matches the document
Spoken language is full of restarts and fillers, and written language is not. Stenograph removes the "ums", repeated words and false starts, adds punctuation and capital letters, and gives the text the shape it needs: a list when you enumerate, paragraphs for a long thought, a greeting and sign-off for an email, headings for a memo. The ai text formatting dictation page shows before and after examples.
Voice commands to fix text without the keyboard
After the text is inserted you can keep talking to change it. Commands act on what you just dictated, so you stay in the flow instead of reaching for the mouse. The complete list is on the voice commands for dictation page.
- "new paragraph" starts a new block of text
- "make that a list" turns the last passage into bullet points
- "delete the last sentence" removes it
- "replace Tuesday with Wednesday" swaps a word or phrase
- "make it shorter" tightens the last passage
- "more formal" raises the register for a client or a court
Writing styles adapt to each app
The same thought reads differently in a team chat and in a letter to opposing counsel. Stenograph knows which app is in front and applies the style you chose for it: concise in Slack, formal in Outlook, structured with headings in Notion and Word, and code aware in VS Code and terminals, where it keeps casing, punctuation and symbols as you say them. You can adjust every style on the dictation writing styles page.
Meetings and file transcription
Dictation is for writing. For listening back, Stenograph turns recordings into transcripts and notes. Record a call or an in-person meeting on your own device, or upload an audio or video file you already have. You get a transcript with speaker labels and timestamps, a summary, the decisions made and the action items, and you can export to Word, PDF, TXT or SRT. No bot joins your calls; the recording happens on your device. More on the ai meeting note taker and ai transcription pages.
- Speaker labels and timestamps in every transcript
- Summary, decisions and action items for meetings
- Exports to Word, PDF, TXT and SRT
Where Stenograph runs
One account covers every place you write. Your dictionary, styles and history follow you between devices, so a term added on the laptop is spelled right on the phone a minute later.
- Desktop app for Mac with a global hotkey
- Desktop app for Windows with a global hotkey
- Chrome extension for web apps such as webmail and online editors
- iPhone keyboard and Android keyboard for dictation on the go
- More than 50 languages, with switching inside a sentence
What happens to your voice along the way
Audio is processed to produce text and then deleted, it is not stored. Text history is optional and can be turned off or deleted. Your content is never used to train AI models. The full detail is on the dictation privacy page.
See it work on your own voice
Record a short passage in the demo, add a few of your own terms, pick a format and compare what you said with what Stenograph writes. When it fits, choose a plan on the stenograph pricing page.