Readiris Pro 11 review

Optical character recognition (OCR) technology has improved dramatically since its early days, and Iris has launched two new products onto the market: Readiris Pro 11, the company's flagship OCR package; and IRIScan, a new slimline document scanner bundled with a copy of the previous version of Readiris Pro, confusingly numbered Readiris Pro 9.

We'll start with the hardware bundle. The scanner is very compact - it's the same width as the main block of keys on Apple's Pro keyboards - and while its design is unlikely to set your pulse racing, it's certainly smart. The USB cable is rather short and since the USB ports on an Apple keyboard don't supply enough power to drive the scanner, you may struggle to connect it without an extension cable or hub.

Once you've installed the software, inserting a sheet of paper prompts the software to draw it through for scanning before automatically analysing the page to work out where the blocks of text, tables and graphics appear. It takes exactly a minute to scan an A4 sheet, and another few seconds to analyse the page. Unfortunately, though, we found the review model gripped the paper quite loosely as it was being fed through, with the effect that a minor obstacle in its exit path or a fold in heavier stock resulted in the paper skewing and giving an unrecognisable result. As such, we had to babysit during the scanning process more than we would have liked.

Once the page is scanned, the software does a decent job of recognising the various different types of area to be processed, although you can turn off this automatic recognition if you'd rather do it manually. Either way, it's a simple matter of dragging a selection rectangle around the areas you want to scan and telling the software if you want to process it as text, a table or a graphic. These selection tools are still a little primitive, though - you can't, for example, select more than one at a time to manipulate or merge multiple selection fields. Moreover, while the software enables you to recognise text in more than a hundred languages (Hebrew and Asian scripts are optional extras), you can only set the language on a per-document basis, which is no use if you have a multi-lingual document.

The software can output your recognised document as text, RTF, HTML or PDF, and in each case there are a number of options, some unique to each format - in the case of PDF generation, for example, you can show the original scan and overlay the recognised text as an invisible, searchable layer.

The bottom line on character recognition is that Readiris Pro 9 does do a very good job, even with less-than-perfect originals. You do still need to edit the result, though. This is partly because the recognition of individual characters isn't perfect, but is more to do with the software's crude understanding of paragraph styles, which often leads to hard-wrapped lines and other glitches. Attempts to recreate a document with formatting and positions intact usually ended with a mess.

That said, the core function - scanning printed originals and generating plain text for later integration into other projects - does work well. It's a little too expensive for such a single-function product, but does fulfil a niche market.

The new OCR software, Readiris Pro 11, is compatible with the IRIScan and other Twain-compliant scanners. It has all the features of Readiris Pro 9 - and all the interface quirks, too. Iris claims it offers improved speed and accuracy, but in our tests we found these improvements to be negligible; the earlier version was already fast and reasonably accurate.