Skip to Main Content
News Releases
2006
KDDI Announces World's First Distributed Speech Recognition Function with "Voice de Input" Carried on au Cellular Phones

KDDI Corporation
KDDI R&D Laboratories
Okinawa Cellular Telephone Company

2006/01/12

KDDI and Okinawa Cellular are pleased to announce that they will carry the distributed speech recognition function, "Voice de Input", in their au cellular phones. Beginning in early February of this year, this function will be offered in cooperation with Navitime Japan in the pedestrian navigation service "EZ Navi Walk", providing "voice new search" and "voice destination search" capabilities.

Customers using EZ Navi Walk "voice new search" and "voice destination search" by means of the "Voice de Input" system can change their search parameters or search for destinations by speaking into the phone without looking at the screen - when walking, in a hurry, or at other times when convenience and comfort are important.

The speech recognition function will be installed on both cellphones and servers. The system extracts features from the sounds spoken in the phone by users, transmits them via data network to a server which performs the speech recognition processing. Conventional cellphones have loadable speech recognition functions, but such functions can recognize only a few dozen words due to cellphone throughput restrictions. A distributed speech recognition function employs a highly efficient server which allows high-precision, large-scale speech recognition based on the newest information.

In addition to improving those aspects causing most of the dissatisfaction with conventional speech recognition, the "Voice de Input" function measures ambient noise levels, and its ability to perform sound analysis processing to maintain high recognition performance has proven useful in a variety of environments. The phone furthermore determines the causes of recognition errors and provides users with alternatives, making the speech recognition function easy-to-use and helpful.

"Voice de Input" was commercialized through improvements in noise-proofing, throughput, reliability, and other features based on the voice feature extraction software and speech recognition engine developed by KDDI R&D Laboratories (Location: Fujimino City, Saitama Prefecture. Director: Shigeyuki Akiba).

KDDI and Okinawa Cellular will continue to test individual services utilizing this function.

* The first DSR (Distributed Speech Recognition) speech recognition function in the world enabling speech recognition on cellphones other than Smart Phones (terminals carrying a general-purpose OS such as Symbian or Windows Mobile for small-packet transmission and reception) by distributing processing between terminals and servers.

News Releases > 2006
Return to Top