August 1998



By Fred Henning


One of the Hot technology topics these days is the integration of Voice and Data. This is what I was doing when I began my computer career in 1960. We were researching ways to have a computer (UNIVAC) understand the spoken word. School children, on tours, could speak - 'one' - 'plus' - 'three' and have the computer type out 4.

Today there are a number of different variations of Voice and Data Integration. We have very sophisticated 'voice recognition systems' that can process continuous speech and translate it into text. Along the way they try to determine if it should be their or there. A variation of this is in 'medical transcription' where the goal is to have the physician or technologist speak fewer words and have the computer fill in pre-determined text. (Ask Dr. Moore about ABMC's experiment with this type of technology, in its early days, when he was Medical Director of ER.)

Voice response is the type of system you know best. You call the hospital and the voice says 'Press One' if you know your party's extension... Voice response is also used by banks, etc. to allow the customer the ability to 'look up' account information. Our Patient Financial Affairs office is anxious to install such a system.

The other side of the coin is data to voice. This can be the train at O'Hare talking to you or your computer reading you a text file. I have the Great Books on CD and have a program that will read them to me! Its a good way to get to sleep! The Patient Financial Affairs solution would also require this technology.

Voice to/from Data technology is popular today because computers are small, have fast powerful processors, have lots of fast memory and of course are relatively inexpensive. The problem is that there needs to be an 'interface' between the computer and the telephone or voice input system. As of now the 'interface' solution often costs more than the voice side of the system.

In the case of our AlexIS system, like many 'transaction' systems, access to data is not as easy as it would be with a 'Relational Data Base System'. In order for us to extract information from AlexIS we have three choices. Print a Report, write an SQL to gather the data or do something outside of AlexIS. In looking at how long someone would wait for a response it is obvious that the 'do something else' answer is required. We could provide an automated solution that would be just a little faster than a human looking up the data via AlexIS screens (MicroScript) or we could provide a repository - outside of AlexIS - that contained all of the data someone on the phone would need.

The latter is the best answer as long as we understand that the answers the caller receives are not 'as of now' but as of some specific time in the past. IE. If you call at 9 AM we may be able to say the data is as of Midnight. But if you call at 7 AM it would be the data as of yesterday at Midnight. This type of solution requires a new fast computer to hold all of the data that anyone calling to can retrieve. The next problem is obvious. Who will the callers be, what questions are we going to be able to provide an automated answer for, how many people are calling at the same time, how long will they wait for the answers, etc., etc. The answers are more obvious. The more you want and the faster you want it the more it costs!

Just as an FYI. Internet access to the same data would require the same 'new fast computer' and the same questions would need to be answered. The Internet, Voice Recognition and Voice Response are fast becoming the means by which we interact with computer based data. Many have said that 'Voice' will be the ultimate choice because it is natural, inexpensive and in most cases the 'user interface', the device we interact with, is the simple telephone.

Next time I will talk about the new 'Magic Box' that will brings us 100's of digital TV programs, super fast Internet access and the video telephone.