Talking to Your Troops:

Voice Recognition for
Computer Moderated Wargames

By Clinton Reilly

In this article I would like to explore the potential of voice recognition for computer moderated wargaming. In the past (issues 60 and 64 of The Courier) I have concentrated on the current status and potential of computer-moderated wargames. In this article I would like to concentrate on an aspect of computer moderation that is more future than present in its application. The short evolution times of PCs and PC software do however make these future developments "just around the corner." So this is something for you to keep it in mind as you shop for your next generation of hardware and software. Don't relegate it to the realms of science fiction - It's not really far away.

The Current Need

One of the main needs in computer moderated wargaming is to make the systems even friendlier and easier to use. They need to be less obtrusive and to flowinto the game naturally end easily. There are a number of ways of enhancing the performance of systems, such as using a mouse orproviding touch screens. These all help to make the interaction with the system faster and reduce the extent to which a computer can get in the way of the game.

However, they all involve being near the PC for part of the time so as to input data, and they don't leave your hands free to move troops on the table top while you are working on the PC. Also, if you are not a whiz on a keyboard or mouse, you may have to expend extra time and effort to get the hang of them as well.

So, what a boon it would be to be able to walk around the table and move the troops at the same time that you are passing information to the system on formation changes etc. This is the promise that voice recognition holds.

What is Voice Recognition?

Voice recognition by computers has been under development for about 20 years with a limited measure of success. The "Holy Grail" of computers kidding a natural conversation with people, as depicted in the science fiction film 2001, is still a long way off and may never be realized in practice. Until recently voice recognition was restricted to highly specialized areas such as military applications, where hands off computer usage was essential.

However, now there are signs of a maturation of technology to the extent that it is being applied to more and more to every day activities such as word processing.

Recently I attended a demonstration of IBM's commercially available voicerecognition product that worked with the word processing package AMI-PRO. The version of AMI-PRO demonstrated has been specially redeveloped to be "voice aware." Not all software packages are like this.

Having worked in the computer industry for over 20 years and been exposed to innumerable demonstrations dedicated to bowl me over, I have become somewhat case hardened and hard to impress. The salesperson's "magic mystery tour" usually leaves me cold. However on this occasion I was quite impressed!

The demonstrator was able to dictate a letter to the system word by word at a reasonably fast rate and have the system "type" it for him. This was not continuous speech as we use in everyday language, but instead involved slight pauses to seperate the individual words.

It sounded a little stilted but in spite of this was about as fast as normal dictation. Remember that when dictating you need to speak clearly and think as you speak, and this slows you down a little.

He could also give standard commands that allowed him to edit the document. For example, the term "up" was interpreted by the system as a directive to move the cursor up one line. A command was issued to tell the system that the user was now about to issue commands rather than dictate text for inclusion in the document. The software could interpret the difference between different words such as "to, too, and two."

The input device used was a microphone attached to a headset worn by the user and linked by a wire to the PC. The system could also "talk" and would read back a document when requested.

Strong Points

The key point is that is promises to allow us to use PCs without the intermediate step of using the keyboard and the mouse. Free hands and free movement. Some more practical trials will serve to prove whether the promise can be realized, but at this stage there is good ground for confidence.

The other major point is that there seems to be good support for current users and future development. IBM is a large organization and are not going to go away in the long term. They have spent about 20 years researching voice recognition and are not likely to give it up now that they are on a winner. So the push to make voice recognition happen is there. It is not a "flash in the pan."

Another point is that, contrary to my expectations with some large organizations, IBM have been very responsive to the application of voice recognition to computer moderated wargames. Frankly, I found the level of co-operation I have received to date to be quite staggering. I guess this may not be too surprising as the total market for all kinds of games (not just computer moderated) is large and games would be one area where voice input could gain a foothold for popular acceptance.

I first contacted IBM to find out how I would make my software "voice aware." They listened to my inquiry and offered to test a copy of my software at their Texas research labs. I thought they may relate more to my American C:ivil War game Rally Once More!, so that was promptly dispatched. The response was that the game would run best under their DOS version, which was under development. They offered to include me in the testing, so I naturally agreed. More about this in a later article.

The fundamental point is that computer moderated software can be made voice aware and the support is there to make it happen. There does not seem to be any real technical barriers. So I intend to proceed in the "voice aware" direction in the future.

Weak Points

There are some weak points as things stand at present but these should be resolved over the next one or two years as the usual technology enhancement and price reduction life cycle progresses.

Cost: The first one is cost. To use the software that I saw, you need a powerful PC with voice input board, microphone and speakers. The voice recognition software takes up a large amount of disk space. The total cost if you want to acquite this is about $10,000 Australian dollars. The cost may not be as much in the US, as an Australian dollar equals about 72 cents US, and I undetstand PCs tend to cost less in the US than they do here in Australia. Consequently the equivalent cost in US dollars may be less than $7,200. That is still a lot of money in anyone's currency!

Why would you bother to do this to run a wargame? The wargames software I sell costs $34.95 for one game and less per copy if you buy more than one - this is not quite a fair comparison as I sell my software at cost (maybe even a bit under) as it is really a hobby for me, and not my bread and butter. IBM and the hardware vendors are financing a serious business venture and so have to charge a lot more.

However, who in their right mind would buy $7,200 of software and hardware to run a $34.99 game? The simple answer is no one I know! However many people will be buying this sort of software and hardware to take advantage of voice activated word processors and spreadsheets for business use. The usual price reductions will then begin to take effect as the product becomes more widely used and economies of scale become available.

This in turn will bring the product within range of the SoHo. (Small Office HOme) market as well and the price reduction will snowball even more. It's the same old story. Wargames usage will then piggyback on this as PCs with more power and the appropriate software become available.

Operating Systems: The next one is the operating system. The version of the software I saw demonstrated runs under IBM's proprietary operating system OS/2. This is a very efficient operating system (for example, independent analysts claim it runs Microsoft's Windows faster than Microsoft's own operating system DOS), but not as widely used, especially in the home environment. Mostly (roughly 80%) home computers using DOC only (about 40%) or Windows (maybe 30%). So most of us cannot use the current IBM voice recognition software.

However, this will change in the near future. IBM has a Beta test versions of the software for DOS and Windows. In other words, the versions we can used at home are reported to be near at hand.

Voice Traniing: The system has to be "trained" to recognize your voice and that of any other user. It can recognize more that one user. This usually takes place over a three week period of use. The system has a huge vocabulary that can be added to. This will allow us to add those peculiar wargames terms that conventional dictionaries do not recognize. This training would occur when you first acquire the system and start using a conventional application like word processing.

My overall rating is that it is very promising but expensive at the moment.

Next Steps

I have been invited by IBM to take part in a trial of the DOS version of the voice recognition software. This has not yet commenced, so I will keep you informed. In future I hope to publish a sample game using the software. This should provide a more detailed view of the operation of the system. I would prefer to try the software before venturing into more detail, but I expect the experience will be rewarding.


Back to Table of Contents -- Courier #67
To Courier List of Issues
To MagWeb Master Magazine List
© Copyright 1995 by The Courier Publishing Company.
This article appears in MagWeb (Magazine Web) on the Internet World Wide Web.
Other military history articles and gaming articles are available at http://www.magweb.com