• Languages included: Danish, Dutch, English US, English UK, Finnish, French, German, Hebrew, Italian, Norwegian, Portuguese, Spanish, Swedish, Yiddish
• Automatically Identifies Columns and Tables
• Automatically Distinguishes between text and graphics
• Automatically Ignores Noise Areas
• Font Trainer
• Spell-Checker (Hebrew-English one pass)
• Text Analysis
• Zone Selection
• Retains many text and document attributes
• Connects directly to the scanner and scanner software without leaving the program
• Supports the leading image file formats and desktop scanners, including Twain and HP AccuPage
• Can recognize mixed English and Hebrew texts in one pass.
• Saves Typing Time
• Up to 99.9% Accurate
• Seamless integration into Dagesh Pro Wordprocessor

Y. Alderstein (1/1/2002) wrote:

"Due Recogniton"

Think most-important-innovation-since-sliced-white-bread, and apply it to Hebrew-English wordprocessing. What you have is the new add-on to Dagesh Pro – its newly released Optical Character Recognition (OCR) Package.

Dagesh OCR is a tool you will quickly learn to love. All of us who use email trade documents that have been inputted as text. In other words, at some point, someone had to type in the text, letter by letter.

What about text which hasn’t been inputted – magazine articles, paragraphs from a book you wish to share, portions of a faxed message? Until OCR, there was no option, other than taking a “picture” of the original using a scanner, and sending it along as a (very large!) graphics file. And there was no efficient way to grab only a few lines, or to actually modify the content or appearance of the text.

OCR changes that. It starts with the “picture” you scan in, but then examines each letter, and enters it for you as text. Better commercial programs, such as OmniPage Pro, preserve layout, formatting, etc., and claim accuracy rates of better than 99%, up considerably from the 95% of a few years ago. (Your typist probably does better. But if you had to enter pages of text, it is cheaper to get a less-qualified proofreader to make corrections in an OCR-generated file, than having a typist copy it from scratch by retyping.)

Until recently, no Hebrew OCR was available to the consumer. The best product, Ligature, aimed at the commercial market, and cost a whopping $1400. TES cut a deal with Ligature to offer an OCR add-in to its popular Dagesh Pro, and offers it at $399 as a stand-alone, or $499 bundled with Dagesh Pro (Having addressed almost all of the complaints about previous editions, Dagesh Pro is everything you could ask for in a multilingual wordprocessor. It handles about every non-Asian language and alphabet you can think of, and comes close to matching the fine-tuning of the much-more expensive Microsoft Word. It usually goes for between $149-$189.)

OCR is an extremely complex process. The program must deal with differences in the shapes and sizes of letters owing to different fonts. It must take into account embedded pictures, symbols, and text flowing between different columns and in tables.

Practically, this means that the user needs to spend a bit of time learning the program, and tweaking it to suit his/her individual scanner. Applying OCR to Hebrew is even more complex! Yet, on my very first try with a page of clean, modern text, the program read the page nearly perfectly, and allowed me to paste lines of Hebrew text into a larger document.

Emboldened, I tried some Rashi text, and learned that it does not yet recognize Rashi, although its Font Trainer allows you to quickly teach it to recognize a particular Rashi font which you may have continuing recourse to.

Why not build Rashi recognition into the Dagesh OCR? After all, probably half of the classic Torah works still come in some sort of Rashi orthography. TES responded that the variety of Rashi fonts, coupled with the poor quality of paper that the older sefarim were printed on (poor paper will produce letters which are blurred, rather than the sharp image that the OCR prefers), made it impracticable. They have a point!

Writing this column in the waning days of the shemita year, I slipped a copy of the RCA pruzbul into the scanner. Dagesh OCR read the document perfectly, except for two characters I had to change, and some underlining I had to supply. I was then able to “personalize” the document by changing to a font and size of my choosing, and adding the city and names of the rabbinic judges who would serve on the court - all at a fraction of the time it would have taken me to copy the text by typing.

There is no question that Dagesh OCR represents a quantum jump for the home and office user of Hebrew text. Dagesh OCR in a must-have addition for rabbis, teachers, and offices that occasionally use documents containing some Hebrew. It will bring the full power of your computer to a world of material that was previously closed to it.

Rabbi Yitzchok Adlerstein scans the horizon for computer ideas for the Torah home when his is not busy directing an educational outreach program for the Simon Wiesenthal Center in Los Angeles, and serving on the editorial board of Jewish Action.