Bob and Joy
                                      
 
     By Bob and Joy Schwabach
                                                                        

Home (947 bytes)

Columns  (947 bytes)

Internuts (947 bytes)

  Bob's Bio (947 bytes)

Email (947 bytes)

 

Home

Columns

Internuts

 About Us

Email

 
                                                                                                               


 

December 2001, Week 3 -- Scanning the Printed Page

 

 

 

 OmniPage

 

   A new version of OmniPage and a fine new import from Russia has us looking at OCR software.

   OCR stands for Optical Character Recognition and it simply refers to software that reads type. That's different from scanning a page into the computer. Because just scanning a page does nothing but take a picture of the page. Unlike text, pictures can't be searched for key words.

   Some businesses depend heavily on OCR software, and most businesses have some use for it. Any business generates thousands of documents (think of a law firm, or insurance company, for examples) and you can either hire people to go through them or you can store them all on disks and find what you want more or less instantly. At least that's the ideal. Reality is running a little behind.

 

   OmniPage from ScanSoft is far and away the market leader here, and they have just come out with version 11. I can remember reviewing version one. Things have progressed a lot since then, but not to perfection. Accuracy has risen to a claimed 99 percent. All the OCR programs out there claim around 99 percent and most of the time they hit it. But you should put that in perspective. On a typical 250-word double-spaced page of typing that results in 2-3 errors a page. Scan 100 pages and you have 250 errors. Scan 100 single-spaced pages and you have 500 errors. That's a lot of errors. And that's just the beginning of your problems.

 

   If you scan magazine or newsletter pages for archives, you should preserve the pictures and the formatting. Think of an ordinary page from Newsweek or Time. Not only are there pictures and unusual page layouts but some text is printed on colored backgrounds, some on white. It's a tough problem and OmniPage handled it well. It handled pages from Time magazine nearly perfectly and simply by using an automatic "wizard."

  Fine Reader

   OmniPage is the market leader in sales, but "FineReader," a new contender from Russia is drawing lots of favorable comment. It seems to be slightly more accurate than OmniPage but didn't handle complex magazine pages as well. The American distributor for FineReader is ABBYY USA and the web site is www.finereader.com. The web site for OmniPage is www.scansoft.com.

   There were several things we liked about FineReader, not least that it started out with a clear tutorial that explained its strengths and limitations and gave you immediate choices based on the type of pages you were scanning. It also told us what resolution our scanner was set on and what we should use for best results. As a general rule scanners have a default setting of 100-150 dpi (dots per inch>. You will get much better results if you raise that to 300-600 dpi. Scanning will slow down but accuracy will improve.

 

   Both programs have so-called wizards to simplify the selection of tools suited to magazine pages, though OmniPage did this better than FineReader. We tried both programs on a typed page that had no columns and no pictures. OmniPage rearranged some of the lines and added spacing. FineReader handled it accurately but changed the type size, which was odd but less troublesome.

 

   This is still a technology in process. OmniPage has a feature that can read the text aloud, which some people may find useful. Both programs come in regular editions for $99 and professional editions with more features for more money. OmniPage Pro is $500; FineReader Pro is $400.

 

Making music

 Plasma

 

   The "Plasma" mixing studio from Cakewalk lets you create new songs or customize existing ones by adding and mixing loops from a library of sound clips. The library even includes the turntable scratches used in hip-hop music. You can use Cakewalk's loops and/or add your own. The loops can be dragged and dropped into place with the mouse. Simple.

   Plasma is $49 for Windows, from <www.cakewalk.com>, or $29 if you own other Cakewalk software. For another $19 you can add a "Plasma FX Pad" that lets you alter the sound effects in real time, providing an active disc-jockey function for things like hip-hop. Lots of fun here and the potential to create new and interesting music. The music can be saved as MP3 or Windows WAV files and recorded to disk.

 

Internuts

 

-- www.wannalearn.com  Free lessons on many subjects, including business, arts and crafts and languages, from Albanian to Vietnamese. How about "Norwegian in five minutes a month?" (Note: Barnes & Noble, booksellers, also offers free courses.)

 

-- www.pinkmonkey.com  Free downloads of summaries of 347 classic works of literature. Online textbooks in 15 subjects, preparations and sample testing for the SAT (Scholastic Aptitude Test), used by many colleges for entrance exams. Message boards, research links and 1,800 tests in many subjects. Free online, $3.95 for printable copies.

 

-- www.gameboomers.com  A totally great new site for walk-throughs and cheat codes for around 500 PC games (no game machine games). By "walk-throughs" we mean they tell you just about everything you need to know to move through and win a computer game. Site also has trivia quizzes for baby boomers.

-- http://tesla.liketelevision.com  Streaming videos of old TV shows, shorts and movies: the Beverly Hillbillies, Dragnet, Bonanza, The Seven Samurai, etc. Typical charges are $5 per show.

NOTE: Readers can search nearly four years of columns at the "On Computers" web site: www.oncomp.com. You can e-mail Bob Schwabach at bobschwab@oncomp.com  or bobschwab@aol.com.

 

[00google.htm]