Tired of scrawling notes on your Compaq iPaq? By early September, you may be just talking to it. That's when Microsoft Corp.'s Research (MSR) division will release a free speech dictation application that runs on the popular handheld.
The dictation application was one of several research projects shown off by Chairman and Chief Software Architect Bill Gates Tuesday during his keynote speech at the International Joint Conference on Artificial Intelligence (AI) under way here.
While the dictation application will be free, you'll need an 802.11b wireless modem -- called "Wi-Fi" -- for your iPaq. You'll also need a matching wireless modem on your desktop PC, because that's where the heavy lifting of speech-to-text processing will actually take place. The iPaq lacks the power to handle such a big task by itself. Finally, you'll need either Office XP, which has a speech engine built in, or Microsoft's Speech API software developers kit, which also contains the speech engine.
And don't expect your voice transcriptions to be entirely error-free. During his demo during Gates's speech, researcher Derek Jacoby said, "If I get good recognition, then it's promotion time." The application, however, translated it as, "If I get good recognition, menace promotion time." To which Jacoby replied, "I guess not," causing the audience of AI researchers to erupt into laughter.
The free speech dictation application -- dubbed an "MSR Power Toy" -- will be available in late August or early September at Microsoft's site.
Gates's speech focused on research attempts to improve computer user interfaces and how AI may be brought into play to accomplish some of those changes.
"We need these advances," Gates said. He pointed out that there are no significant hardware obstacles to future improvements, so software research will be key to making computers easier to use.
Later in Gates's speech, researchers from MSR's adaptive systems group demonstrated user notification technologies that will be included in Microsoft's .Net Web services initiative. The .Net Notification platform will govern how and when a user receives information such as e-mails, voice calls, and instant messaging, and whether they're directed to the user's PC, handheld, or cell phone. It's one of four initial "HailStorm" building-block services that Microsoft intends to supply early next year.
Researcher Eric Horvitz showed off a prototype notification system that combined his personal preferences as to when to interrupt him with a message or call, with information regarding which of his many devices was best to deliver the message. Gates has given less developed demos of the technology in recent months. But in a new twist, Horvitz's group has added a vision system and an audio processing system that recognized from the crowd noise and his demeanor that he was in the process of giving a presentation -- showing how a notification platform could figure out not to interrupt him at that particular moment.
"We're very close to integrating key aspects of this technology into .Net," Horvitz said.
Another demo performed data mining on TV-watching patterns and displayed relationships between users' viewing habits. For instance, the AI-laced analysis concluded users who watched Friends were likely to also watch Seinfeld, whereas viewers who liked Murder She Wrote were more likely to watch 60 Minutes.
Following his keynote, Gates announced at a brief press conference that Microsoft has donated US$7.2 million toward a new $70 million computer science building at the University of Washington in Seattle. Microsoft executives and employees have come up with an additional $4 million.
The move echoes similar contributions Gates and his company previously made to Massachusetts Institute of Technology and Stanford University.