 |
BEYOND
THE BLEEDING
EDGE:
21ST CENTURY "CLIPPY" |
B Y
N E I L P E R L I N
Boston Chapter
Ten years ago, most online help appeared as white ASCII text on a black screen and WinHelp and context-sensitivity were cutting edge technologies. Today, of course, it’s hard to tell what’s cutting edge from one day to the next. But one thing that’s as common today as it was ten years ago is the complaint that PCs are too hard to use. That complaint is at the heart of one of the most interesting areas of interface design – the search for a friendlier, or "social" interface.
Microsoft’s Bob was the first attempt at a social interface and a resounding failure. Mentioning "Bob" in a speech is a sure-fire laugh-getter. Clippy, the animated paper clip interface to the help in Microsoft Office 97 applications, was the second attempt. And, while Clippy and its brethren – Einstein, the Genie, et al – are loathed by experienced users, new users like them, even referring to Clippy as "he." But whether you love or hate Clippy, the important thing to remember is that it’s simply the interface to underlying natural language recognition and search technologies. As those technologies improve, it’s reasonable to expect that the interface will improve as well, and what better interface than a person? Like an always-open line to tech support…
There aren’t enough people to serve as help interfaces, but what about the next best thing – virtual people? It may be years before virtual people actually provide the interface for advanced help systems, but a brief look at some of today’s technologies suggest what’s possible.
Many virtual people are really virtual bodies. For example:
- The Land’s End site (www.landsend.com) lets you build a Personal Model Ô (for women only at this time) on which to try on clothing. It’s an approximation, but better than looking at a jacket on the screen and using your imagination.
- "Transom Jack" from Transom in Ann Arbor, MI (www.transom.com), is a virtual body and face development system used to test positioning and motion in machine and assembly line design.
- Paraform (www.paraform.com) in Palo Alto, CA offers virtual people and development tools.
But a real virtual human interface needs a face, and there are many attempts at creating one. For example, Big Science Corp. in Dallas, TX (www.bigscience.com) offers the Klone (TM) that uses a real human face with different expressions tied to different responses. See Figure 1.
Figure 1 – Meet a Klone
Figure 2 – After telling it I have a cold (change "health" to "hardware" to imagine a tech. support Klone)
| A Klone can express emotion, but it’s static. What about a live face? A little research revealed interesting developments, most having to do with controlling mouth and lip movement to simulate real speaking. |
Figure 3 – The dialog continues |
|
Live mouth movement for animation has already been done, albeit primitively, as far back as the 1950s in the Clutch Cargo cartoon series (Figure 4). Clutch was a writer and pilot who flew around the world on assignment, accompanied by his friend Spinner and a dog named Paddlefoot. The show used a technology called Syncro-Vox that superimposed film of actors’ lips speaking the lines over the cartoon characters’ faces. Compare the lips in Figure 5 to those in Figure 4. You’re seeing the superimposed actor’s lips in Figure 5. The technology cut animation expenses but has two problems from our point of view. One is the odd effect of a static face speaking emotional lines. More important is the fact that the character can only say the lines spoken by actors, whereas the help interface’s virtual human would have to generate words in response to unpredictable questions. What we need is the ability to generate lip and mouth movements synchronized to the words being spoken. And this too is under development. |
Figure 4
– The opening panel from the Clutch Cargo show Figure 5 – The superimposed lip |
|
Figure 6 – The LipsInk model |
A recent column about Y2K technologies in Mobile Computing Magazine touched on research being done by speech technology vendor Lernout & Hauspie on synchronizing lip movement and mouth shape to the phoneme data stream – that is, make the character look like it’s really talking. HyperVision, in Britain, offers a similar product and service called LipsInk. To see a demo, go to www.hypervision.co.uk/lipsink/ and click the See LipsInk At Work link. Figure 6 shows the demo’s face, which has realistic mouth movements. The result resembles a talking Barbie doll. To complete this picture, we need to animate the rest of the face as well. Search the web and you’ll find many sites dealing with this, most written for programmers. But one example, a paper dated 1996(!), written by a Moroccan researcher named Al Hassan Moubaraki from a research lab outside Kyoto, Japan, shows an example of full face animation. See Figure 7. |
Figure 7 – Moubaraki’s example of full-face animation
Within a few years, these technologies should be merging into one unified virtual person that can run on PCs with commercially realistic processing and memory requirements. The result might be help systems represented by a virtual face that pops up in a corner of the screen and talks users through some task. The development of these virtual people is likely to be the job of the programmer but, hopefully, it will be the job of the technical communicator to design and create the virtual human’s lines – to give it a virtual life.
Neil Perlin has twenty-one years experience in technical writing, with fifteen in training, consulting, and development for various types of online documentation including WinHelp, HTML Help, and some now known only in legend. Neil writes about online documentation and is a popular speaker before the STC and other professional groups. Neil provides training, consulting, and development for online documentation through Hyper/Word Services of Tewksbury, MA. You can reach him at nperlin@concentric.net or www.hyperword.com.
Copyright © 2000 Neil Perlin submitted to the STC for use in Hyperviews:Online.
Volume 3, # 1