We have two books. One book 1200 pages Rashi font, and a second book 1200 pages Niqqud-Hebrew. We have attached only one page of each as a 5MB sample. Each page needs header, margin, and footnotes electronically trimmed off. The margin (verse and chapter numbers) and these trimming also need to be saved on the corresponding page/file number. Some pages may need fine tesseract alignment, (deskewed). Select all 27 individual letters-shapes, plus 15 Niqqud points, and the cantillation marks and get these ligatures from raster to vector for each book. We hope to get 99.99% of the lines of text in good order and be put into a spread sheet line by line. We need to keep the Niqqud points from interfering with the good sequenced lines of letters. We expect the accuracy of the Niqqud points to be lower. Select for us a collection of the 3 best of each ligature that will provide 3 sets (a good/better/best) for us to adjust later to a keyboard or for reprinting. The best of three (or a combination of) is what must work in the tuned tesseract product. We expect the software pipeline complete with read me instructions in addition to the two spreadsheet outputs of the two tesseract operations sent to us. The books proper page number must correspond with the page number in the spreadsheet and we should be able to duplicate the OCR output with the open source software you will provide.