Speech Technology Magazine

 

Sensory and CSR Team to Offer Speech-Powered Bluetooth Chip

Known for making embedded speech recognition chips and software, Sensory expanded its offerings today with the announcement that it will provide chips for wireless Bluetooth headsets manufactured by CSR.
By Lauren Shopp - Posted Sep 18, 2007
Page1 of 1
Bookmark and Share

Known for making embedded speech recognition chips and software, Sensory expanded its offerings today with the announcement that it will provide speech recognition software for CSR's Bluetooth silicon chips. Sensory's FluentSoft Bluetooth Suite is the latest addition to CSR's eXtension Partner Programme, which provides enhancements to its BlueCore-Multimedia platform. The speech-enabled chip allows product designers to integrate speech control and command functions to their Bluetooth headsets and hands-free car kits.

The headset, which works with all mobile devices that have Bluetooth technology, is the first of its kind to use speech recognition to execute commands. In addition, the device is interactive, as it responds to commands such as, "Check battery power," or "Am I connected?"

Todd Mozer, president and CEO of Sensory, says the partnership between CSR and Sensory represents a turning point within the current headset market. "A big problem with Bluetooth headsets is the user interface," he states. "There's no room for a display, and the buttons are really small."

Creating a chip that would enlarge the headset was not an option, so Sensory opted for speaker-independent speech recognition technology  while making the headset even smaller. "The hardest part in making a robust speech technology was that size constraints limited the amount of RAM," Mozer says.

Sensory and CSR also included new commands and features to the chip, including: battery check, voice dialing with pre-programmed numbers, and constant access to Google's mobile search application, GOOG-411. The chip's new features represent a change from the steps required to operate older Bluetooth headsets. Instead of pressing a series of buttons together, alone, or concurrently to perform functions, users need only speak their command.

Mozer says that background noise interference with spoken commands was tested and that the headset's performance was, "not that different than other speech recognizers." While users may still use the device while driving in the car or listening to music, talk-radio tends to cause difficulties. Mozer also says background noise levels may reach 70 to 80 decibels and the chip will still understand spoken commands.

The headset will enter the retail market with a price in the $75 to $150 range. Sensory and CSR's product will be released soon.

Page1 of 1