> ## Documentation Index
> Fetch the complete documentation index at: https://docs.syntheo.studio/llms.txt
> Use this file to discover all available pages before exploring further.

# Dictate in other apps

> Use a keyboard shortcut to turn your voice into text in the active field.

Keep Syntheo open, place the cursor in a text field of another app, and use your shortcut to speak instead of typing. Dictation does not create a meeting note.

Dictation uses the transcription server set up in **Settings → Transcription**, with your languages and your dictionary. Audio is not kept; your last dictation stays in memory until you turn dictation off or close Syntheo.

## Set up dictation

<ol />

1. Turn on "Enable dictation".
2. Click the "Dictation shortcut" to record a new key combination ( by default), then choose your "Microphone".
3. If needed, click "Transcription server ›", on the "Language and vocabulary" line, to check the transcription server and model.
4. Under "Try it", click in the field and test your shortcut.

Try it in the replica: turn on "Enable dictation", click in the "Try it" field, then hold  and release. A sample sentence is inserted and "Last dictation" appears. You can also click the shortcut and type another combination.

By default, hold the shortcut while you speak, then release it to finish. With "Hands-free dictation", press once to start and a second time to finish. The text is inserted when transcription is done.

With "Live transcript preview" (off by default), the words appear in a small bubble while you speak, if your transcription server accepts a real-time connection.

During dictation, a floating panel appears above the active app. Its button (1) shows the microphone level; a click finishes the dictation. The bubble (2) shows the recognized words, and the arrow on the right expands or collapses it. Try it: hold "Hold the shortcut" while the words appear, then release. Check "Hands-free dictation" to start and finish in two clicks, and press  to cancel.

Syntheo inserts the text it receives in real time. If this stream is interrupted, the OpenAI server transcribes the recording afterwards. An NVIDIA NIM model cannot: dictation then shows "The live transcription of this dictation was interrupted, and this model can't transcribe the recording afterwards. Try again.".

Dictation stops after five minutes and pauses while Syntheo records a meeting.

## Cancel a dictation

* On Windows and on Linux with X11, press .
* On Linux with Wayland, use .

You can also click "Cancel dictation" in **Settings → Dictation**.

## Linux specifics

On Wayland, approve the shortcuts in your desktop environment's confirmation window. The target app must expose an editable field through accessibility. These instructions are repeated in **Settings → Dictation**, in the collapsed section "On Linux (Wayland or X11)".

On all platforms, dictation does not work in password fields.

## If the text is not inserted

Keep the original text field active until the dictation ends. If the target changed or the app refuses the insertion, open **Settings → Dictation**, click "Copy last dictation", then paste the text yourself.

If dictation shows a setup error, click "Retry setup".

Dictation is a desktop interaction: the CLI and the MCP server do not start dictation and do not insert text into other apps.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.