Command Mode shortcut does nothing
What you see: you press the Command Mode shortcut (Option + Shift / Alt + Shift) and nothing happens. No sound, no indicator, no error. Why: Command Mode has its own switch, and it is off by default. While it is off, the shortcut is silent. Fix:1
Open Command Mode
Click Command Mode in the sidebar.
2
Connect an AI provider if asked
Without a connected provider, the page shows a locked card with a Connect an AI provider button. Connect one under Language models first.
3
Turn on Command Mode
Turn on the Command Mode toggle at the top of the page.
Text is not pasted
What you see: the dictation shows up in Home › History but never lands in your app.
- Check that Settings › General › Auto-paste transcription is on. It is on by default. When it is off, Text Insertion Method, Drag-to-Insert Banner and Auto Enter are greyed out.
- Click into a text field in the target app before you dictate.
- On macOS, check that Accessibility is granted. See permission problems.
- Switch Settings › General › Text Insertion Method between Paste method and Direct insertion.
- Quit Vowen from the menu bar or tray icon and open it again.
A “Drag to insert” card appears instead of the text
What you see: on macOS, after a dictation, a card shows your text with a copy button instead of pasting it.
- Accessibility is not granted. The card then adds “Grant Accessibility permission so Vowen can insert text for you.”
- The focused element is not a text field, such as a Finder list or an image.
Apps like Google Slides now paste
Some apps report their editing area as a generic container instead of a text field. Google Slides does this for its canvas. Vowen now pastes into these instead of showing the card.Wrong spacing or capitals around your dictation
What you see: a dictation lands with a missing or extra space, or with the wrong capital letter for the sentence around it. Why: Settings › General › Smart Spacing and Capitalization reads the text around your cursor to add the space a dictation needs and match the capitalization of the sentence. When it cannot read that text reliably, it leaves the dictation as spoken. It is off by default. Fix: if the result is wrong in a particular app, turn the setting off. If you are on Windows and want it, turn it on.Dictations pile up in my clipboard manager
What you see: every dictation appears in Maccy, Raycast, Alfred or Ditto.
- Check that Settings › Developer › Allow Dictations in Clipboard History is off. It is off by default.
- To keep dictations off the clipboard entirely, switch Settings › General › Text Insertion Method to Direct insertion.
- Settings › General › Restore clipboard after paste (on by default) puts back what was on your clipboard before the dictation.
Stuck on Listening (macOS)
What you see: the indicator stays on Listening after you release the shortcut. Why: under heavy load, macOS can stop delivering key events to Vowen, so the key release never arrives. This fixes itself. After 15 seconds without a release, Vowen ends the recording and transcribes what it captured. If it happens often, something is saturating your CPU. Check Activity Monitor and see performance.Every shortcut stopped at once (macOS)
What you see: dictation, Command Mode, hands-free and the meeting panel all stop responding together. Why: the same loss of key events, while nothing was recording. This fixes itself within a few seconds. Vowen also clears any modifier keys it thought were held. If shortcuts are still dead after that, see shortcut problems.Windows: blank window or dead shortcuts after sleep
What you see: after waking the PC, the Vowen window is blank, the indicator buttons do not respond, or shortcuts do nothing. Why: Windows can leave app windows and keyboard delivery broken after sleep. This is repaired on wake. If a recording was in progress, Vowen waits for it to finish. If it is still broken after a few seconds, quit and reopen Vowen.Recording indicator does not show
What you see: you hear the start sound but see no indicator. Fix:- Check Settings › Recording › Recording Indicator Position. It offers Top (the default), Bottom and Don’t show.
- On macOS, check Indicator style in the same section. Notch draws the indicator in the notch area instead of as a floating pill.
- Check your other displays.
- Restart Vowen.
”Thank you” appears in the output
What you see: the text reads “Thank you.” although you said nothing like it. Why: Whisper-family models invent this phrase over silence. Vowen already discards it on local models when it is the whole transcript. Fix:- Start speaking as soon as you press the shortcut.
- Turn on Settings › Experimental › Enhanced silence detection. It raises the silence threshold for cloud models in noisy rooms. A Sensitivity slider from 1 to 5 appears once it is on. The default is 2.
First or last word cut off
What you see: the start or end of what you said is missing. Fix:- Hold the shortcut for a moment after you finish speaking.
- Open the entry in Home › History and choose Retry transcript. Retry uses the original recording, so it shows whether the audio or the model lost the word.
A model I was using disappeared
What you see: a banner reads [model] has been removed, and dictation does not run. Why: the speech model was retired in an update, and your settings still point at it. Fix: click Choose a model on the banner and pick a replacement. Also check the model set on any tone you use, since a tone overrides your main model.”Local Whisper server is not running”
What you see: this banner, and Restart Server does not help. Fix:- Delete the model on the Speech models page and download it again.
- Check that antivirus has not quarantined the model or the Whisper program.
- Try a different model to see whether the problem is the model or the server.
- On Windows, if you use the portable build, install the Microsoft Visual C++ Redistributable. The installer adds it for you.
Model download fails with a certificate error
What you see: “self signed certificate in certificate chain” or a similar error, usually on a work machine. Why: corporate networks with SSL inspection (Netskope, Zscaler and similar) replace certificates, and Vowen’s downloader rejects them. Fix:- Download the model in your browser and place it in Vowen’s models folder. See manual model installation.
- Ask your network administrator to exempt
huggingface.coandassets.vowen.aifrom SSL inspection (on Vowen 0.6.2 and earlier, also the update hostpub-9d034521e4604ac7861eae261b5b8673.r2.dev). - To test, try the download over a personal hotspot.
Cloud services time out or fail
What you see: cloud transcription or AI requests fail, often with a network error, while other apps work. Why: some networks advertise IPv6 but cannot route it. Fix:- Check that Settings › Developer › Prefer IPv4 Connections is on. It is on by default and applies immediately. Only turn it off on an IPv6-only network.
- Re-enter your API key on the Speech models or Language models page.
- Check the provider’s status page and your usage limits.
- Quit and reopen Vowen.
Vowen is slow or uses a lot of memory
- Switch to a smaller local model, or to a cloud model.
- Turn on Settings › General › Resource efficient mode, which loads the model only when you dictate. It appears only for local Whisper models, and for Parakeet on Windows.