Dictate into your computer by holding a phone button

Hold one button on your phone, speak, and let go to have your words typed into the active computer text box.

Download Try it

Lower part of the phone app showing Realtime mode, Direct send, a text box, and a blue Hold to talk button. — With Direct send chosen, letting go of Hold to talk sends the sentence to the computer.

You hold one button on your phone instead of typing.

The microphone is active only while held down, with no one-tap start on the touch screen and no latch.

Phone app during dictation showing words on screen, a timer at 0:04, and a red Release to end button. — Words on the phone show progress, not the final text.

With a screen reader or switch control, starting and stopping are two separate steps.

Two-step screen reader control comes from code only; the computer app has no text size setting.

Notepad on Windows holding the dictated sentence with the cursor at the end. — The sentence landed in Notepad after tapping Deliver on the phone.

What does not work or was not tested

The software cannot click or scroll. Moving the cursor, clicking, switching windows, selecting text, and editing stay on your keyboard and mouse. FlowMic includes no voice commands, and it has no computer shortcut or hotkey; the talk button is on the phone.
A held press with one finger or thumb remains, which may not suit everyone. Long speaking stretches also have their own cost for some people. Short messages, corrections, and moving around still happen on the keyboard and mouse.
Words appear on the phone as you speak without flowing into your document; text lands in the computer text box only after you let go, not word by word.
Across 15 spoken passages, "Organize" removed fillers, but in 4 of 15 test passages "Organize" added or changed words never said, turning "because the train is a two" into "since the train ride is about two hours". Read "Organize" output before use.
Password fields on Windows are not refused and report "Delivered", so check cursor position before talking.
Supported macOS 11 (Apple silicon) and iOS 15 (absent from mainland China and EU App Stores) were not tested.

More limits of this kind

On the phone, "Delivered" means words reached the computer, not that they were typed into a window.
In Windows consoles, an open menu or Move mode refuses sends, and Mark mode loses words while the phone says "Delivered".
In VS Code or Cursor panels and consoles started by ssh, python, or wsl, line breaks are not joined and can run commands.
On Linux (x86-64 on X11 only), xterm and the VS Code panel get nothing while the phone says "Delivered".
We tested on Windows 10 and Ubuntu under X11 with an Android tablet using text typed into the phone and a synthetic voice from a second phone speaker. Recognition of a real human voice was never measured.
Windows 11, Cloud relay, other phones, locked screens, copied images, and cursor position after landing were not tested.
Word, Outlook, Gmail, Slack, Notion, Google Docs, Overleaf, Teams, Discord, Obsidian, VS Code, and helpdesk apps were not tested on Windows.

A few common questions

Can I review and edit my text on the phone before sending it to the computer?

A small "⇄" control by the input area selects delivery. "Direct send" is the default where release sends the text; "Manual send" shows a card with "Deliver" and "Discard" to review text before sending.

Does Local LAN need a language model set up on the computer for each mode?

On "Local LAN", "Translate", "Organize", and "AI polish" require a language model under computer Settings in "Language model", where the app states "Translate and Organize need a language model on this computer". Without one, "Realtime" arrives marked "No model set up · raw text".

Download Try it

Tested on FlowMic 0.3.120 on the computer and 0.3.118 on the phone, 6 November 2026.

Some counts come from FlowMic 0.3.119 on 5 November 2026 or noted test builds and were not repeated on 0.3.120.