Advanced Speech Recognition Enables Gamers to Interact with Game Characters

Using Voice Commands

PEABODY, Mass.--(BUSINESS WIRE)--Jan. 13, 2005--ScanSoft, Inc. (NASDAQ:SSFT

- News), a global leader of speech and imaging solutions, today announced

that Nintendo has implemented a ScanSoft® speech solution for the Nintendo

GameCube(TM) game entitled Mario Party® 6. Using ScanSoft's speech

recognition, players can control and interact with game characters, adding

an exciting and natural dimension to game play.

As the sixth version of Nintendo's popular Mario Party series, Mario Party 6

is the first to provide speech recognition technology to control certain

mini-games. Using ScanSoft technology, the game will accurately recognize

the nuances and unique elements of a wide array of voices, including the

voices of children. The game, which comes bundled with a microphone, allows

players to use their voices in a variety of quizzes and select mini-games

within the world of Mario Party 6. Using the Mic controller, users can bark

orders at their racing character for motivation, or answer quizzes to earn

extra stars to spend on game extras like Magic Books, hidden levels and game

secrets.

"ScanSoft has a reputation for adding a level of creativity to a variety of

applications," said Genyo Takeda, general manager, Integrated Research and

Development at Nintendo Co., Ltd. "The software was extremely easy to

integrate into Mario Party 6, and adds an incredible level of real-life

interaction to the experience."

"Speech adds a whole new element to the user experience for games," said

Alan Schwartz, vice president, Embedded Speech, SpeechWorks Solutions from

ScanSoft. "By making our Speech Games SDK available on the Nintendo

GameCube, developers will be able to easily add speech to the user interface

for many different genres of games, as evidenced in Mario Party 6. We look

forward to working with Nintendo and its developers to offer this exciting

new feature in future titles."

SpeechWorks Solutions from ScanSoft are powering a new generation of games

and applications that uses the power of speech recognition to provide

advanced command and control of the gaming environment, as well as real-life

interaction for gaming enthusiasts. Gamers can perform complex operations

simply by speaking, and add real dialogue to game characters, bringing them

to life. With speech technology, virtual characters can express their

emotions, share their needs and answer questions, becoming unique,

personalized lifelike creatures.

Integrated within Mario Party 6, SpeechWorks' Solutions from ScanSoft VoCon®

Games Speech SDK leverages ScanSoft's renowned ASR 1600 engine, designed to

ensure a smooth, natural interface for users of all ages. SpeechWorks

Solutions from ScanSoft VoCon Games Speech SDK for Nintendo GameCube is

available for all Nintendo-certified developers, and is the first

recognition engine specifically designed to work with children's voices.

Advanced speech recognition engines use proprietary signal processing

techniques to ensure accuracy, while a scalable vocabulary ensures

flexibility in game development.

Highly Accurate Recognition

The VoCon Games Speech SDK features a phoneme-based speech

recognition engine that uses proprietary signal processing

techniques and advanced speech recognition algorithms to

provide highly accurate voice recognition in different

environments.

Scalable Vocabulary

Game developers can easily create new vocabularies or enhance

existing ones using a structured grammar format (BNF) featured

in the VoCon Games Speech SDK.

Speaker-Independent Acoustic Models

With VoCon Games Speech SDK, the end user can simply plug in

the microphone and start playing with no training or system

enrollment required.

Children's Voice Support

Japanese and American English acoustic models have been

trained with children's voices as well as the standard male

and female adult voices.

Multiple Recognition Modes

Game developers can easily associate actions with recognized

words or phrases. End users can initiate speech commands in

"push-to-talk" (or "intercom") mode or by talking directly in

the microphone.

Support for Customization

Full customization of the speech recognition engine parameters

is available to optimize performance and CPU and memory

consumption--including reduction of the memory footprint for

simple applications.