At a press briefing this week at its research facility here, IBM previewed several prototypes of speech recognition interfaces, including a speech console on the dashboard of a Chrysler that communicates with the driver about the condition of the car.
IBM also demonstrated a voice browser used to search for movies and showtimes on the Web; a robot head that can smile, frown, speak and move its eyes; and the Personal Speech Assistant, a conversational user interface for the PalmPilot III PDA that will be ready within a year.
The goal: 'transparent computing'
"More and more, we're driving technology into smaller devices," said Ozzie Osborne, general manager of IBM Voice Systems. "This technology is getting so small that it can go anywhere."
Osborne said IBM is creating the technological framework and partnering with application developers to achieve a "state of transparent computing," where users interface with computers "without knowing it."
One of IBM's partners, One Voice Technologies, demonstrated a "human interface technology" that allows users to conduct a two-way dialog with their PCs using natural spoken language. Company officials showed how users can search the Web, launch desktop applications and access files.
The OEM and consumer versions of One Voice's product, which runs on IBM's ViaVoice Millennium speech engine, will be available in the first quarter, officials said.
Asked why the computerized voice in some of these devices is poor, David Nahamoo, department group manager in the Human Language Technology Department at IBM Research, replied: "Because IBM hasn't found a way to make it good, and we're still learning how to do sound technology."
Nahamoo said the technology works well for industries such as banking and travel, where there are a limited number of voice responses.
