中 Download
Features Formats Shortcuts Tech Docs GitHub Download

Troubleshooting

Fix missing previews, unresponsive voice input, OCR failures, screenshot hotkeys and update problems.

Updated 4 min read Edit this page on GitHub
All documentation
All documentation
Getting started
Getting started Spacebar preview Keyboard shortcuts
Core features
Editing and saving Offline voice input Region screenshot, OCR and color picker
Reference
Settings and system integration Technology and project layout Privacy and data Troubleshooting Frequently asked questions
On this page

First confirm that SigXor is running in the tray, that you are on the latest version, and that the app folder is writable. Most issues are answered by Settings and the tray menu. If you are still stuck, attach the logs from Documents/FNSoftware/SigXor/Logs to a GitHub issue.

Installation and startup#

Nothing happens when I run SigXor.exe#

That is expected. SigXor lives in the system tray and does not open a main window. Check the notification area, or run SigXor.exe --silent to verify.

SmartScreen says the publisher is unknown#

The release is not code-signed yet. Choose More info → Run anyway. If security software blocks it, add SigXor.exe and its folder to the allow list.

The app reports a missing .NET runtime#

Install the .NET 10 Desktop Runtime (64-bit). The ASP.NET Core Runtime or the plain .NET Runtime alone cannot run a WPF app.

The app cannot save or update#

Unzip SigXor into a writable folder such as D:\Apps\SigXor. Protected locations like Program Files break automatic updates and some save operations.

File preview#

Pressing Space does nothing#

  • Make sure a file is selected in File Explorer and focus is not inside a text field.
  • Make sure SigXor is running in the tray.
  • Some file managers or third-party shell extensions intercept Space. Use the tray menu or the command line instead.

Text files show garbled characters#

Switch the encoding in the preview toolbar to UTF-8, UTF-16 or GB18030. SigXor auto-detects when possible, but mixed or damaged encodings may need a manual choice.

Large files fail to open or ask for another view#

Text preview is limited to 100 MB; use the hex view beyond that. The hex view reads through a virtualized window, so files up to 1 GB remain smoothly editable.

Office previews look incomplete#

  • Toggle between the structured and web views at the top of the preview. The web view is closer to the original layout, but it is disabled for PPT / PPTX files larger than 8 MB.
  • Legacy .doc / .ppt (OLE2) files are unsupported and offer the hex view instead.
  • Without WebView2, the web view, Markdown rendering, audio and PDF previews are unavailable, while the structured view and text editing still work.

WebView2 is missing#

Windows 10 and 11 usually include it. If not, install it from the Microsoft WebView2 page or use the Evergreen Bootstrapper.

Voice input#

The voice hotkey does nothing#

  • Confirm Start voice service is enabled under Settings → Voice & screenshot.
  • Confirm the SenseVoice model has finished downloading.
  • Check whether an IME, game or other global-hotkey tool captures the key.
  • Check Windows microphone permissions and the default input device.

Recognition works but nothing is typed#

  • Focus the target input field before speaking.
  • If clipboard paste does not work, switch to Keyboard simulation.
  • Elevated windows reject input from normal processes (UIPI). Run both apps with the same privilege level.

The model download never finishes on first use#

Check the network and proxy settings; the speech model is about 230 MB. Use Services & models to open the model folder and check progress, delete an incomplete model and retry.

Screenshot and OCR#

Alt + ` does nothing#

  • Confirm Enable region screenshot is checked.
  • Make sure the modifier does not conflict with the voice hotkey.
  • Fn combinations on laptop keyboards usually do not reach Windows; enable No modifier `.

Screenshots are misplaced on mixed-DPI monitors#

The overlay is currently one virtual-desktop window. Multiple monitors with the same DPI work correctly; mixed-DPI layouts are still being validated. As a workaround, set all monitors to the same scale.

OCR returns nothing on first use#

The first run downloads PP-OCRv6 (about 30 MB) and falls back to PP-OCRv5 if that fails. Wait for the download or fetch the model manually from Services & models, then retry.

OCR accuracy is low#

Capture sharper, higher-contrast text and avoid scaling up tiny glyphs. OCR is tuned for screen text; handwriting, decorative fonts and low-resolution images remain difficult.

Updates#

"Cannot update automatically"#

Automatic updates are disabled for debug builds, non-writable app folders, or a missing updater. Unzip the app into a writable folder or download the latest ZIP manually from Releases.

The update failed or the app will not start#

  • The updater backs up the previous version as SigXor.exe.bak and rolls back if the new version is not ready within 30 seconds.
  • See updater.log in the app folder for details.
  • If the updater was killed, rename SigXor.exe.bak back to SigXor.exe.

Problems after upgrading from Velopack#

The portable ZIP and the old Velopack installer are not interchangeable. Uninstall the old version and unzip into a fresh writable folder; user data and models remain in Documents/FNSoftware/SigXor.

Diagnostics#

When filing a report, include:

  • The SigXor version (tray menu → About).
  • The Windows version and whether it is 64-bit or ARM.
  • Reproduction steps and the file type involved.
  • Logs from Documents/FNSoftware/SigXor/Logs.
  • For screenshot issues, the monitor count, resolutions and scale factors.

More answers are in the FAQ.