Please recommend gratis OCR software that can prioritize speed over accuracy.
Tesseract is quite good, but too slow for the task at hand (unless there is a way to speed it up using its options file - please see this Stack Overflow question if you can help in that regard).
The task requires performing OCR on 20K English-language web browser screenshots. This task will need to be performed on a semi-regular basis (1-2 times per month).
75% accuracy will be sufficient for this task, and only a few key words need to be identified in each screenshot. The purpose of the OCR is to use the OCR output to sort the screenshot files by keyword. I'll write the code to allow appropriate tolerances for the sort to work within the specified accuracy.
Cross-platform software is ideal, but Windows compatibility is currently a minimum requirement.