Yes, You |
 |
by R. J. Bullock |
Voice recognition is not new, but it is getting more attention these days because of major software improvements. Many HAL-PC members are new to voice recognition, and have some basic questions.
Voice recognition is software that converts your speaking voice into computerized words and commands. The three most notable companies that make this software are Kurzweil, Dragon, and IBM.
IBM just announced a consumer version called Simply Speaking. IBM has several professional voice recognition products, but this is the first time they have sold an introductory consumer version.
Simply Speaking works well. IBM programmers have worked for over 30 years on the program to recognize the human voice. The core program in any speech recognition software is called the "speech engine." The speech engine does the speech recognition work. A recent InfoWorld review found that the IBM speech engine was noticeably better than the other two competitors. Simply Speaking has the same speech engine as the professional version.
All speech recognition software fails to correctly recognize words sometimes, particularly if you have a thick accent or mmumbblle—excuse me, mumble. The vast majority can talk to the software. The new speech recognition software is "speaker independent", which means about 85% of the population can speak to it successfully right out of the box.
The old software required you to train it first; however, with the new software, and speaker independence, you don't. The training capabilities are still there, and will improve your accuracy, but most of us don't have to train Simply Speaking before using it. If Simply Speaking has trouble understanding your accent, just train it.
The professional IBM version (VoiceType Dictation) is used by doctors, lawyers, and other professionals who prepare reports and case notes. In the past, doctors had to wait hours or days to see their dictated report. (Now that's a problem in the Emergency Room!) With VoiceType Dictation, turn-around is instantaneous, 24 hours a day. Lawyers use it to prepare contracts, dictate court notes, and manage legal paperwork. Professionals can use it to prepare any type of report. Realtors can use it to document properties. Secretaries and transcribing services can use it to save time and money. Journalists can use it to write newspaper and magazine articles. Anyone who produces lots of text can use it to speed up turn-around, and reduce report costs.
Simply Speaking software is probably not the version a doctor or lawyer would use. The professional version, VoiceType Dictation, has the same top-rated speech engine as Simply Speaking, but adds productivity features that save hours of time in report preparation and file management. Most professionals would want to graduate to VoiceType Dictation fairly soon after starting on Simply Speaking.
A regular computer user with voice recognition software could dictate letters, prepare student reports, document recipes, make notes on hobbies, or even use voice to interact with games and entertainment programs. With more and more software coming out voice-enabled, the list is endless.
Speech recognition can certainly be a great advantage by handicapped people, and the same convenience speech recognition has for handicapped people is useful to others too. The good news is that the price on these products has dropped dramatically, and is widely affordable for the first time ever!
To run Simply Speaking you need 100 MHz 586 Pentium, 16 meg RAM, a standard sound card and Win95. You also need a high-end active noise-cancellation microphone to filter out background noise, but IBM has included one of those in the box (a high-end microphone alone normally sells for about $70!).
Voice recognition comes with OS/2 Warp 4.0, but not just the Simply Speaking consumer version. OS/2 Warp 4.0 has the professional version called VoiceType Dictation, with macro capabilities, navigation for windows, templates for business forms, and the option to add tailor-made professional vocabularies. You can now run VoiceType Dictation on either Win95 or OS/2 Warp 4.0.
Simply Speaking recognizes 22,000 words out of the box, with capabilities to add up to 64,000 words. And when you consider that most of us only use about 5,000 words in everyday speech, that's a lot of horsepower! You can add regular words, jargon used in your profession, proper nouns like names of people, even acronyms like HAL-PC.
It's easy to get started. And it's also an amazing experience to see your voice instantly converted into words. But to really use speech recognition effectively requires extra time in training and support. Simply Speaking is a consumer introduction to the professional world of voice dictation. There's lots to learn about how to use it effectively. That's why we formed a SIG.
Come to our new HAL-PC SIG on voice recognition and internet telephone to learn more about this. The SIG is called "You Can Voice". We will have demonstrations, discuss applications, share user experiences, compare products, and learn to use it to best advantage.
RJ Bullock, the leader of You Can Voice SIG, is a trained and certified voice recognition specialist. By the way, this report was prepared by voice dictation.
E-mail me at webmaster@hal-pc.org with any comments you have and tell me what you want to see here.
Back to the Magazine Home Page