Capture2Text is a surprisingly simple (and portable) OCR and speech recognition tool.
The program can capture text from anywhere on your screen, for instance. Just move your mouse cursor to the top left of the text; press the hotkey (Win+Q by default); move the mouse cursor to the bottom right of the text; click the left mouse button, and the decoded text ends up in the clipboard.
Capture2Text's speech recognition follows a similar pattern. Press a hotkey (Win+A by default); speak a few words or sentences; when you're done, a list of the most likely transcriptions will be provided, and whichever you select will be sent to the clipboard.
If any of this doesn't work quite as you expect then there are lots of configuration options which can help. You're able to customise the hotkeys, for instance; set a new language for OCR or speech recognition; or maybe tweak the output settings so the text goes directly to a program, rather than the clipboard.
We had a few problems when testing the program. The hotkey didn't work consistently, for instance. And the OCR accuracy was generally poor.
There are plenty of issues which may influence this, though, so if you're interested in the core idea then there's no harm in giving the program a try: you may have more luck.
Recent changes include:
[Version 3.5 (7-17-2014)]
- Capture box should be less jumpy.
- Preview will now only update when the user has stopped moving the capture box for at least 400 milliseconds.
- When preview is setting to "Dynamic", the positioning should be less jumpy.
[Version 3.4 (7-10-2014)]
- Added option to strip furigana from Japanese text.
- Added the "Auto" choice to the "Text direction" preference.
- Removed the option to toggle "OCR pre-processing" from the Preferences. It may still be edited in settings.ini.
- Changed the default "OCR pre-processing" hotkey to Shift-Ctrl-Windows-B.