#2 In a Series of Voice Recognition Articles

Welcome to the 21st Century Eloquence Page, where we discuss new technological advances, enhancements and inventions toward Speech / Voice Recognition Technology. This page is updated regularly, so you may wish to revisit it every quarter or so.

Speech / Voice Recognition background

Speech recognition software is being used every day by hundreds of thousands of people. Have you ever tried to call with a calling card and get the message "Please say Collect, Calling Card, Third Number, Person to Person, or Operator now". If you haven't heard that message, maybe you heard a similar one? This is speech recognition.

Whether you know it or not, you have probably used speech recognition. It's a known fact that this technology is being implemented everywhere. It's been around for more than 50 YEARS! Why is it so popular now? There are two answers. The first answer is computer hardware is now available to take advantage of the technology, and the second answer is that now it's affordable. Previously computer hardware strong enough to run this technology was too expensive for the general public to purchase. But now, with the the advancements in Central Processing Units and Digital Signalling Processors, it has become not a dream, but reality.

The current market consists of different forms of Voice Recognition: These different forms are:

SPEAKER DEPENDENT
SPEAKER INDEPENDENT
COMMAND & CONTROL
DISCRETE SPEECH INPUT
CONTINUOUS SPEECH INPUT
NATURAL SPEECH INPUT

SPEAKER DEPENDENT - This technology requires users to participate in extensive training exercises that can last several hours. Once you are done "drilling the machine" the computer then begins several calculations from the data it has received from your exercises. After these calculations, the computer makes a voice profile that attempts to match your voice synthesizations.

SPEAKER INDEPENDENT - This technology, on the other hand, does not require a person to go through exercises. A user may begin using the Voice Recognition program upon installation.

DISCRETE SPEECH INPUT - This type of input requires the user to pause between words so that the computer may distinguish the beginning and ending of words. Although your speech has to be modified slightly, hence slowing your regular dictation, you can achieve well over 80 WPM, the speed of an advanced typist. Some have even reported speeds of up to 125 WPM.

CONTINUOUS SPEECH INPUT - This technology is currently available for very small vocabularies (2000 words) and numbers by very few vendors. This speech input requires the user to say only words that are known to the system. You are also limited to the expandability of the libraries. This technology is currently not useful for dictation, but is very useful for specific functions or programs, i.e. data entry systems. I estimate this technology being available around the fall of '97 or beginning of '98. However, I also believe that these systems will start at around $2000, and, I have also heard that you should start seeing this technology around the end of '96, but these systems will be specialized for specific purposes and they are going to start at more than $5000. I figure it will be affordable and usable by the year 2000. No, I'm not kidding. That's only 3 1/2 years from the time of this writing.

NATURAL SPEECH INPUT - This is the ultimate goal in Voice Recognition Technology. To be able to talk to your computer in no specific manner and have the computer understand what the user wants, then apply the commands or words (i.e. New letter to Michael Rooney from ABC Systems ). The computer will know to bring up a letter template with Michael's name and the inside address of ABC Systems and stop where dictation is to resume. This unfortunately is not yet available. Realistically, I don't expect to see this until the year 2002.

However, there are some systems in the market that are excellent today. Depending on what your goals are, you may be able to gain major benefits from choosing the right product.

The company I work for knows this and that is why they carry all of the popular major products that you would ever consider for your task. This way, they can give the right advice, and the right product for everyone's specific use.

What some people say about Voice Recognition Technology

I used to see many articles about Voice Recognition. Some of them said it's not ready. Since the writing of my first article 11/95, that has changed and many writers and journalists now use speech recognition software to write their articles. There are still some journalists who feel the technology is not ready, but if they typed no more than 25 wpm, and did not have the luxury letting someone else type their work, and have not received a repetitive strain injury (RSI), they would be using it too. [I GUARANTEE IT!].

Who's in the game

Articulate Systems

Articulate systems develops speech recognition products for the Macintosh. The PowerSecretary system uses the Dragon Dictate speech engine. So far, they are the only vendor to offer speech recognition for Macintosh platforms. Of course, Macintosh has the Apple Plaintalk (I'm sure there is a trademark somewhere in there). Plaintalk is great for some developers wanting to voice activate their programs, however, I do not see it as a solution for dictation (not since this writing).

Allows programmers to customize their applications with Voice Recognition Technology. Using continuous speech with switchable vocabularies, much can be done with this product.

One of the first and most popular programs that was quickly used by users and programmers. I personally think that Dragon has the edge on their DOS product. It is very simple, fast and easy to customize. Dragon has what you call an OOPS buffer. This buffer holds the last ten word you've spoken into memory. This allows you to dictate fast, easily and quickly, go back and correct errors.

Dragon Dictate for Windows is almost identical to its DOS engine. Actually, I have had the opportunity to test their new 2.5 product which is NT compatible. Dictation is fast, accuracy is great, and functionality is awesome. With it's new release, Drag on has outdone themselves. I can't imagine what they'll do with their 3.0 version. I guess we'll have to see and I don't expect seeing it for another 9 - 12 months. So far, it will have to take my editors pick for general and practice specific dictation.

This is one of the fastest dictation systems using unique Trigram technology that quickly surpasses the speed of all the systems mentioned here, however, I must mention, Dragon Dictate is right behind it. With speed of up to 125 wpm, and accuracy above 95%. Great speed and great accuracy. It does however not let you "see what you are saying". There is a delay between words. The IBM system thinks while you continue your dictation. You can see it think too. If you want to dictate, and let someone e lse do the proofreading and formatting, this is an awesome system.

IBM Voicetype is available for OS/2 also, but if you are a true WARP user, the new WARP "Code named Merlin" includes dictation. (Thanks IBM)

IBM Continuous Speech Series is for the developers. This program allows programmers to generate up to 1000 word vocabularies for specific programs or functions. You can have several 1000 word vocabularies, but you cannot use them at the same time. My knowledge is limited to this product since I don't actually use it.

Kurzweil Applied Intelligence

Kurzweil Applied Intelligence leads the way for medical reporting using their patented VoiceReport technology, although, new continuous speech engines which should be out by the end of 1996 will challenge these products. Kurzweil's VoiceReport systems are speaker independent, and use a unique embedded structured type of reporting that gives their systems Quality Assurance and Risk Management features. You might say something like,

"New Patient"
"Patient Name" , "Juan Pedro" (you'll have to type that in) "Patient Demographics" .... "30", "Year", "Hispanic", "male"
"SOAP"
"Abdominal Pain"
"Routine"
"Take Defaults"

With these few short words, the Kurzweil system writes something to the effect of;

ROUTINE: [here, you will have a paragraph of information relating to default medical diagnosis.] Specific systems include;

A Speaker Independent system for Windows. Allows one to dictate into any windows application without having to train it. This system also has a continuous digit recognizer for easy data entry of addresses, spreadsheets and other miscellaneous tasks involving numbers.

Kolvox Communications, Inc. so far leads the way for 3rd party business solutions using voice recognition technology. Kolvox has recently merged with PureData and the new company's name is Wild Card. For details, visit the Kolvox Site (kolvox.com).

OfficeTALK
Speech solutions which give all types of users powerful voice word processing solutions.

LawTALK
Speech solutions which give attorneys the power of dictating without the costs of transcription or tying up the secretary.

Both of these systems use either the Dragon Engine, Kurzweil VOICE engine or the IBM engine. These products are tailored by adding thousands of macros and words to the dictionaries allowing a user the ability not only to dictate, but to do easy word processing and adding power to your voice engine. Things like customized marked forms for easy voice fill-in operation, an address book application to merge your addresses straight to your document and manipulating text such as "Bold Sentence" bolds a sentence and "Move Paragraph" moves paragraphs. These are things that the voice recognition engines do not come with and Kolvox has done a great job in filling in. Some engines currently offer some of the features of Kolvox's products, however, none of them have it all. And some, have none.

Who wants this technology

Several people wonder, is this for me? Do I really want, or need this technology?

Here is a quick summary of the types of people who are currently using the system, and how helpful it is for their daily work or play.

Physicians may use these systems to replace their transcription personnel or their transcription services. Using this technology, Physicians gain a lot. First is cost. It is very expensive for transcription services. Depending on the amount of work and the type of field a Physician is practicing, costs may range from $1000 - $3000 monthly per Physician. A Physician can spend more than $30,000 yearly on services or $20,000 yearly on a full time transcriptionist. This is not including the costs for unemployment compensation, insurance, Workers Compensation and payroll. Second is Risk Management. Using the VoiceReport system, a Physician only needs to say a few key words like "Chief Complaint", "Abdominal Pain", "Routine Checkup" and the VoiceReport system will bring up several different procedures that are usually performed when doing a routine checkup of abdominal pain. This will remind the Physician of the most common procedures that are to be performed when this type of examination occurs assuring a precise and complete examination. On the other hand, if the findings are all normal, you can say "All Normal" and the VoiceReport system will spew out a well documented report. Third will be Quality Assurance. All reports will be full, and wording will be correct and consistent to these types of examinations. Malpractice insurance may also be lower as a result of this technology. Currently, there are insurance companies that offer discounts for using systems such as this. Yet, with sophisticated configurations, a physician has the power of having automated faxing through the system, link this system with their current Medical Management packages, having automated prescriptions being printed out, and can even send ICD-9 Codes to their billing systems for automatic billing. There goes Medical Coders. There is much to be gained using this type of voice technology.

Once upon a time, there was an attorney who had a secretary. This attorney depended on his secretary for most of the office work. One day, there was a little friction in the office and she left the job. This was a two days before his vacation and there was a lot of work to be done. Well, he did everything that had to be done, but he couldn't type and his handwriting was bad. He knew if he sent it to transcription, they would all come back with errors. He decided to take his handwritten notes and spend an 2 hours a night on his vacation to type them in himself. Well, he finished and it only took him an average of 4 hours per night to finish.

Well to make the story short, he bought LawTALK for DOS and now does all of his documents by voice. He doesn't depend on anyone for anything now.

Attorney now ...................................... "Computer, print document and read"

Use this technology as great presentation tools, or office tools. You can simply tell your computer, "Smith & Company file" and watch the computer bring it up. You can call up reports, or say "Resume Draft" to continue editing a draft. You can also use this with your contact management program. Give it commands like "Find Company","21st Century Eloquence", "Print Profile", "Call Company". Pretty darn cool!

For Keyboard Impaired individuals such as people with Carpal Tunnel Syndrome or other Disabilities, this will allow you to be as productive as any other person who types. Let's face it, whether it is for college, work or hobby, we all type at some point. Dragon Dictate is the only "Hands Free" product in this roundup.

The most lucrative way to enhance your applications, and your programming. You can do voice-ready applications and your product will be state of the art. On the other hand, most programmers type fairly well, but can you imagine telling your program "For Next Loop" and have it bring in a long procedure written by you. You can virtually do voice-batch-files so you can call procedures, objects, libraries or functions instead of typing them in. Now that will save you programming time. On the other hand, make your applications a "voice-aware" option for easy navigation, data-entry, and functionality. Currently, you can play a Klingon game by voice and do lots of other things. Voice-aware applications will be the next generation. "Mark my words"

Which is right for you

Well, lots of people really don't know how to choose from all of the software, and truly, I don't blame you. There is a lot of software and each of them has different features. I'll try to give my best suggestions depending your needs. Look at my reviews for specific likes and don't likes, then refer to my software suggestion list broken up by profession.

Entity	1st Choice	2nd Choice	3rd Choice
--------------------	--------------------	--------------------	--------------------
Accounting	Dragon Dictate	Kurzweil Voice	Kolvox OfficeTALK
Corporate Entities	IBM Voicetype	DragonBiz	Officetalk w/Dragon
Developer	Dragon Tools	IBM VoiceType Application Factory	Dragon Dictate
Computer Guru	DragonTech	Dragon Dictate	Kurzweil Voice
Disable	Kolvox OfficeTALK w/Dragon	Dragon Dictate	Kurzweil Voice
Journalism	IBM VoiceType	DragonXtra	Dragon Dictate
Legal	Kolvox LawTALK	DragonLaw	IBM Voicetype
Medical	Kurzweil VoiceMED	DragonMED	IBM VoiceType
Sales	Kolvox OfficeTALK	Dragon Dictate	Kurzweil Voice
Teacher	Kolvox OfficeTALK	Dragon Dictate	Kurzweil Voice

Please be advised the the preceding article is based on the opinion of X-chief voice recognition specialist at 21st Century Eloquence. In no way does this article represent 21st Century Eloquence, Kurzweil Applied Intelligence, Kolvox Communications, Dragon Systems, IBM, MicroIntrovoice or any other company mentioned. Also, companies / products mentioned are Trademarks or Registered Trademarks of their respective entities.

This information may not be reproduced in ANY media unless written permission is granted.

I hope I have been helpful in guiding you through Voice Recognition Technology. Overall, all products do more or less what they're supposed to do. "Type while you dictate".

When you are ready to order, you may click here, or from anywhere in this site, click on the ORDER button.

Senior Technician on Voice Recognition Technology

TABLE OF CONTENTS Welcome

WELCOME

Speech / Voice Recognition background

What some people say about Voice Recognition Technology

Who's in the game

Articulate Systems

Kurzweil Applied Intelligence

Who wants this technology

Which is right for you