March 23, 2005

New speech recognition programs such as Dragon Naturally Speaking 8 Professional(ScanSoft, Windows 2000/XP with 1.6 GHz Pentium III/IV, Professional Edition $710; Preferred Edition $169.99; Standard Edition $87.99; features vary with edition; check www.scansoft.com for details) produce transcription accuracy of up to 95% or better and are filled with a host of usability features that make it easier than ever to have your speech directly converted to text or commands.

Faster computers, more memory and improved speech recognition programming make it possible for these programs to recognize "continuous speech," which means you can now talk in a conversational mode or even faster, since the program can transcribe up to 160 words per minute, as long as you speak clearly. The program types your words as you dictate at speeds faster than that of most skilled typists, and there are no misspelled words, since all the words come from the program’s dictionaries. If you want, the program can also automatically add periods and comas to your sentences.

We started with a short 15-minute tutorial and were off dictating with the resulting transcription showing amazing accuracy. If Dragon Naturally Speaking recognizes a word incorrectly for example "two" instead of "too," just say "select (word)," and it’s selected--along with a menu of suggested changes. If none fit, say or type the word you want to replace it with, and it’s replaced. A host of formatting and cursor-moving commands make navigating your transcription easy, once you get the hang of it. You can also select phrases to edit, just say "select (phrase)" and if you want it bolded, for example, just say "bold that." If we couldn’t remember the commands, we just said "What can I say?" and a list of voice commands appears on the screen.

The program also provides other voice command functions, so that without ever touching the keyboard or mouse (except to turn the microphone on), you can start a program, open a file, dictate text, save files or send e-mail just by speaking the commands. For example say "run word" and word opens, or say "open doc dot doc" and word starts and opens doc.doc. Or use any combination of voice, mouse or keyboard. You can even turn off the microphone to stop voice recognition, but you cannot use voice to turn it back on. You’ll need your good old mouse for that.

Sounds like the future is here now, and it is. Voice recognition isn’t absolutely perfect, but we found it’s extraordinarily good and can benefit many people--for example, those who need to reduce their use of keyboard or mouse, don’t have secretarial help, are unable to or have difficulty using a keyboard or mouse, or those who would just like to join the future, now.

Dragon Naturally Speaking 8 has become better and better in the business of continuous speech recognition, now sporting a vocabulary of 160,000 active words (250,000 total word vocabulary). You can add custom or unusual words or have the program search your documents for words to add that are not in its dictionary. The program has expanded its language and vocabulary support, with five variations of English: US English, UK English, Indian English, Asian English or Australian English and also provides general, custom or teen (how about that) vocabularies. Multiple users on a single computer are supported as well.

Dragon’s proprietary programming language allows you to customize just about anything for voice commands. Cursor control is especially flexible. We could move the text insertion point by lines, words or letters or use their ‘Mousegrid’ feature to move the cursor anywhere on the screen; we could also click, double click or drag, all with only voice commands. Dragon Naturally Speaking 8 can also dictate into almost any Windows based application such as Microsoft Word, Corel WordPerfect and other word processors, Outlook, Outlook Express and other e-mail programs, Excel and many others. The program can be further customized to fill in blank forms. A new feature is the ability to insert graphics or blocks of text with voice commands.

Direct recognition of recorded speech is supported using PC dictation devices. Text-to-speech and playback of recorded dictation can be used to help correct your transcriptions as well. Or use the text-to-speech feature to have any of your documents or e-mails read back to you.

Though the program may not initially recognize your pronunciation of certain words or unusual words, with additional training and the Acoustic Optimizer feature, it will learn. The Acoustic Optimizer stores your corrections and additional training in special files. Then makes use of this information to update your user files for further improving the program’s speech recognition accuracy.

The more you train, the better the recognition gets, but it does take some time, patience and training (both you and the program) to get used to this new way of controlling your computer and transcription. There is extensive on-line help within the program as well as a 206-page printed User’s Guide.

Oh, and if you want that smiley face at the end of your e-mail, just say "smiley face," and it appears.

Click Here to Return to the Main Column Archive Page

Click Here to Return to the Home Page