Product Reviews

There's nothing like a head-on battle for developing software, and Caere and Xerox have been trading blows in the OCR arena for several years now. The end result for consumers is improved products every year, and with the latest release of OmniPage Pro we're reaping the benefits of tenth-generation software.

Despite its maturity, OmniPage Pro 10 has a considerable list of improvements over OmniPage Pro 9 (reviewed issue 52, p198). According to Caere, it has revamped the OCR engine, giving this version a 30 per cent improvement over its predecessor in terms of errors.

Our figures are less conclusive. Overall, in our tests, OmniPage Pro 10 managed a 98.2 per cent word-recognition rate, the same as OmniPage Pro 9, while TextBridge Pro 9 (reviewed issue 58, p201) scored 97.9 per cent. However, OmniPage Pro 10 lost out to both its predecessor and TextBridge when it came to recognising small serif text. In a 55-word paragraph of eight-point text, TextBridge made 12 mistakes, OmniPage Pro 9 made nine and Pro 10 slipped up 17 times.

OmniPage Pro 10 made up for this by handling the rest of the document extremely smoothly. In marked contrast to its rivals, it reproduced fonts faultlessly, correctly spotting Arial and Times New Roman, and not making TextBridge Pro's blunder of deciding that all of the text was bold; and nearly all of the Web and email addresses were correctly spelt out.

As with TextBridge and OmniPage Pro 9, you have control over how the page is reproduced in your word processor. For exact positioning ('True Page'), OmniPage uses text boxes, which is fine if you're not going to be editing the document and just want the original reproduced for printing. If, however, you're going to be editing, the compromise option is to retain the picture positions and column layout but lose the exact positioning of text. True Page delivers mostly impressive results, but boxes aren't always as big as they should be, so the bottom line in the box often disappears.

Like all OCR software, OmniPage Pro 10 is much better at recognising normal fonts rather than italicised, tending to roll words into one, and it didn't cope well

Improvements aren't to character recognition alone. Caere has also concentrated on ease of use, with a much-simplified interface. For OCR novices there's a Wizard option for a step-by-step guide, but most of the time - if you're scanning simple documents - you just use the Auto OCR setting. Press the friendly green Start button and OmniPage will start your scanner and automatically recognise the areas of the page as text or images - all you need to do is name the finished document. It then fires up the destination program, ready for you to print it or make minor adjustments.

If you're an experienced OCR user, and want control over which areas of the page are going to be recognised, you can switch to Manual OCR. This allows you to manually select pictures and to tell the program whether certain blocks of text are reversed.

The proofreading and editing process is similar to that of the previous version, but as this was always a strong point for OmniPage this is a good thing. The Proofreader highlights any words it considers dubious, with a Context window displaying the relevant area of the scanned image. You also have easy access to the whole scan and the recognised text in the default side-by-side Window view. There's just one major new feature, namely a read-back facility, where the recognised text is synthesised into speech. This doesn't have many obvious uses, except perhaps for proofing long lists of figures or as a tool for the visually impaired.

Although OmniPage Pro 10 can convert scanned pages into HTML files, Caere also bundles a copy of OmniPage Web Personal Edition. This is an interesting program, which converts up to ten scanned pages into a formatted Web page. It also retains URL and email addresses as hyperlinks. The full, unrestricted version is aimed at serious users, as its £128 upgrade price shows.

Even without this inclusion, OmniPage Pro 10 moves ahead of TextBridge Pro 9 in most areas because of the improved format retention and character recognition. Of course, we fully expect Xerox to fight back with its next product, but for now OmniPage Pro 10 is the best OCR product on the market. At £67 for an upgrade from any other OCR software (including any bundle with your scanner), it may seem a lot to pay, but it's an investment well worth making if you want true reproduction of your scanned documents.