Speech Server to be Included in Communication Server, as Voice Recognition Flubs

Published: August 9, 2006

by Alex Woodie

Microsoft fleshed out its voice recognition and text-to-speech strategies this week at the SpeechTEK convention in New York City, including an announcement that Speech Server 2007 will be integrated into Microsoft Office Communications Server 2007, instead of being released as a separate product. The company also provided demonstrations of new speech technologies that will be included in Windows Vista. However, it appears that bit of technology could use a bit of work.

Earlier this summer, Microsoft unveiled the next-generation Communications Server 2007 product, which will serve as the real-time hub for a range of communication mediums, including e-mail, instant messaging, voice over IP (VoIP), and audio-, video-, and Web-conferencing (see "Microsoft Unveils Unified Communications Product Roadmap").

Communications Server 2007 was designed to work with the Speech Server product--which is used to deliver voice-enabled applications and has successfully answered more than a million phone calls--as well as with Exchange Server 2007. But the addition of Speech Server to the product proper will bring benefits, Microsoft says. Chief among these benefits is the integration of API sets, which will make it easier for developers to write voice-enabled applications for Communication Server.

Meanwhile, in a widely publicized demo of Vista's new voice-recognition software, things didn't go quite as planned. Microsoft's Shanen Boettcher was trying to dictate to Vista, but it seemed that Vista had other ideas in mind.

"Dear mom," Boettcher started. However, the computer heard "Dear aunt." He then tried to correct this mistake by saying "Fix aunt," a phrase Vista translated as "let's set."

Boettcher then tried to get Vista to backup by saying "delete that," three times, but the computer didn't do anything. He then told the computer to "select all." Instead of highlighting the screen so Boettcher could delete the offending rubble, the computer spit out "so double the killer delete select all."

The demonstration is available at several places on the Web, including YouTube.

Microsoft says the new speech technologies in Vista will be available in eight languages, and will benefit users lives by allowing them to dictate documents and e-mail messages, to fill out forms on the Web, and to command applications and the operating system. Judging by the current state of the technology, however, it probably wouldn't be wise to edit your registry with voice commands--not just yet, anyway.