Skip to main content
The defaults are right for most people. Change these only if a specific app misbehaves. The paste method, clipboard handling, trailing space, and auto submit settings live in Settings → Dictation → Output. Spoken emoji is under Settings → Dictation → Extras.

Paste method

The default is Clipboard (Ctrl+V) on Windows, Clipboard (Cmd+V) on macOS, and Direct on Linux, because clipboard paste is less dependable across Linux compositors.

Clipboard handling

Other output settings

Spoken emoji

Under Settings → Dictation → Extras. Say “happy emoji” or “thumbs up emoji” and SpeakoFlow inserts 😊 or 👍. Common aliases and minor speech-to-text errors are handled on your device. Unknown names stay as text. Off by default.

Filler-word removal

On a local speech engine, SpeakoFlow removes filler words (“um”, “uh”, false starts, repeated words) automatically on every dictation. The local engines have always done this, and there is no switch on the local side. On a cloud provider, the Remove filler words switch controls both layers: it asks the provider to strip fillers server-side and runs the app’s own filter as a backstop. Turn it off to get verbatim output, word for word. See Cloud transcription.

Paste-target restore (Windows)

On Windows, SpeakoFlow remembers which window was in the foreground when you started recording. If one of its own windows takes the foreground during the recording (because the overlay or the assistant panel appeared), it hands focus back to that window before pasting. This only happens when a SpeakoFlow window is what took focus. If you alt-tab to a different app during the recording, SpeakoFlow respects that and pastes into the new app.

The order things happen

Once a recording is transcribed, the text passes through these stages:
1

Transcription

Custom-word correction and filler removal. See Custom words.
2

AI cleanup (optional)

Only on the dictation-with-cleanup shortcut, or when cleanup is set to run on every dictation. See AI cleanup.
3

Spoken emoji, then text replacements

Both opt-in. Replacement rules run last, so they always win. See Text replacements.
4

Insert

Trailing space, paste method, then auto submit.
Typing tool appears in Settings → Dictation → Output only on Linux, and only while the paste method is Direct. It picks which tool types the text. The dropdown lists only the tools present on your system (wtype, kwtype, dotool, ydotool, xdotool). Leave it on Auto unless a specific tool misbehaves.Clipboard writes prefer wl-copy under Wayland, which handles non-ASCII text better.
Two settings in Settings → Debug affect insertion (Debug appears after pressing Ctrl+Shift+D, or Cmd+Shift+D on macOS):Paste delay (default 60 ms). How long to wait after writing the clipboard before sending the paste combo. Increase it if the wrong text pastes.Extra recording buffer (default 0 ms). How much longer to keep recording after you release the key. Increase it if your last word or two gets cut off.
Any method that synthesizes keystrokes releases every modifier afterwards, even if the combo was interrupted. This guards against Ctrl or Shift appearing stuck down after a paste.
Last modified on August 30, 2026