Troubleshooting

SmartScreen blocks the installer#

Prefer the Microsoft Store. The direct GitHub installer is known to show an Unknown publisher warning. Download it only from the official TypeWhisper release and compare its SHA-256 hash with the published hash before running it.

Get-FileHash .\TypeWhisper-win-x64-Setup.exe -Algorithm SHA256

No microphone or wrong microphone#

1. Open Windows Settings > Privacy & security > Microphone and allow desktop apps to access it.
2. Choose the input under TypeWhisper > Settings > Dictation, or select Default.
3. For changing headsets, order the microphone priority list; TypeWhisper uses the first available device.
4. Re-run onboarding from the dashboard and check the input level in step 2.

Hotkey does not respond#

1. Open Settings > Shortcuts and record the combination again.
2. Choose a combination that Windows or another app has not registered globally.
3. Check workflow-specific hotkeys as well. TypeWhisper prevents conflicts inside the app, but not with other programs.
4. Quit the conflicting app or change one of the shortcuts, then restart TypeWhisper.

Text is not inserted#

TypeWhisper first copies the final result to the clipboard and sends Ctrl+V for automatic insertion. Some elevated, protected, or unusual text fields block simulated input.

1. Press Ctrl+V yourself immediately after dictation. The text should still be on the clipboard.
2. Check Settings > Dictation > Auto paste after transcription.
3. Test a regular field such as Notepad. If it works there, the limitation is in the target app.
4. Do not run TypeWhisper with lower privileges than the target app; Windows blocks input across that privilege boundary.

Model or extension is not ready#

1. Open Settings > Dictation and verify that the engine and model are installed and selected.
2. Open the add-on's page in the Settings sidebar and check its status, such as Ready, Setup required, or Needs attention, and its configuration. If an update is available, the page shows Update.
3. Restart TypeWhisper when an add-on shows Restart required or after changing local model acceleration.
4. If automatic acceleration fails, try CPU. Use CUDA, Vulkan, or ROCm only with matching hardware and runtime support.

API or CLI cannot reach TypeWhisper#

1. Check Settings > Advanced > API server: Enable HTTP API is on, the status reads Listening on http://127.0.0.1:<port>, and the port matches your client.
2. Test curl.exe http://localhost:8978/v1/status, or use the displayed port.
3. Do not delete discovery files while the app is running. Restart TypeWhisper if the port or token does not match the client.
4. With Require API token enabled, /v1/status stays public; every other API route needs Authorization: Bearer or X-TypeWhisper-API-Token.

A setting is locked by your organization#

Your IT department set it through a policy. Settings > Advanced > Managed settings shows the values in effect on this device and any policy problems. Only your IT department can change the value; you can't change it in TypeWhisper.

Export diagnostics#

Open Settings > Advanced > Support diagnostics and click Export diagnostics… next to Support report. The report includes app and system details, microphone names and access, models, plugins, workflow metadata, settings, and any retained diagnostic log. It excludes API keys, audio, transcripts, prompt contents, and file paths, and nothing is sent automatically. Microphone names may contain personal names; review the report before sharing and send it only through a trusted support channel.