Advanced Voice-Command Processor Chip Offers Flexible Platform for Applications Development; Dual-Mode Speaker-Independent/Dependent DSP Processor Combines Latest Speech Recognition and Noise Adaptation Algorithms to Meet Demands of Wireless Industry and Other Markets.

CUPERTINO, Calif.--(BUSINESS WIRE)--Sept. 11, 1995--DSP Communications, Inc., (NASDAQ:DSPC) today announced a dual mode, speaker-independent and -dependent, voice-command chip with sophisticated capabilities representing the company's newest achievements in speech recognition and accuracy.

The company's advanced technology has brought speaker independence and dependence to a single digital signal processor (DSP) chip that achieves as high as 97 percent recognition rates. This dual-mode processor enhances DSP Communications' wireless product line, positioning the company as one of the comprehensive suppliers of DSP solutions for the personal communications market.

Integrating multiple capabilities, DSP Communications' new DVC306 Advanced Voice Command processor provides a cost-effective base for developing highly flexible applications with simple user interfaces. Not only does the chip offer speaker-independent and -dependent voice recognition, it also offers built-in voice synthesis, which enables vocal feedback for user guidance, prompting and verification.

The new DVC306 achieves high recognition rates and quick response times even in demanding noisy environments such as automobiles. This noise-robustness should make the chip well suited for implementation in a variety of wireless voice-command applications, including hands-free voice dialing, car navigation, audio control, personal digital assistant (PDA) operation, and others.

"The new Advanced Voice Command Processor should be an ideal building block for developing next-generation hands-free wireless products that require high-performance, voice-actuated user interfaces," said Shaul Berger, vice president of marketing and sales for DSPC. "The chip's new capabilities emerge at a time when mobile communications manufacturers are seeking a mature portfolio of features to enhance product evolution and strategically position themselves ahead of the increasing competition."

"The new DSPC chip offers developers an opportunity for quick development of voice-driven consumer products," said William Meisel, editor of Speech Recognition Update, an industry newsletter. "The market for such products is growing rapidly, and the mixed speaker-dependent and -independent recognition will make such products easier to use."

Along with the voice-command processor, DSPC is offering a Windows-based development and evaluation kit to support integration and prototyping. DSPC also provides technical support to help developers quickly and efficiently integrate the chip's voice-command capability into their products.

The voice-command processor, presently priced below $10 at 100K volumes, will be packaged in a 100-pin TQFP. Samples will be available in Q1 1996.

Features and Use

The DVC306 provides a flexible and cost-effective platform for voice-command processing. The single-chip DSP solution integrates low price, small size, low power and high performance.

Unlike software-based solutions, this chip is designed to minimize host burden and complexity. The chip is controlled by a simple 4- or 8-bit external host microcontroller and is easily integrated into voice-command systems. The host controls the chip's operation mode, commands the chip to play synthesized phrases and reads status information such as recognition results.

The DVC306 recognizes a word out of a group of predefined speaker independent or user-defined speaker dependent words. Speaker-independent words are stored in external ROM during product development, while speaker-dependent words may be added or deleted to or from the external SRAM at any time during the end-user training process.

With both speaker-independent and -dependent recognition, the DVC306 Advanced Voice Command Processor, eliminates the need for training prior to end-user operation, yet enables the creation of personalized directories. In addition, end-users with accents may replace speaker-independent words with speaker-dependent words to improve the recognition rate.

During recognition, the DVC306 compares spoken words with the word templates stored in memory. The results for the best match -- along with the input signal level, SNR and error codes -- are transferred from the chip to the microcontroller. Acceptance/rejection thresholds for recognition can be modified, depending on specific application requirements.

The DVC306 provides an integrated voice command solution supporting audio cancellation, key-word spotting, speaker phone, memory pad, battery charger, and DTMF (dual tone multifrequency) generation. Support for multiple users and variable length vocabularies further enhances operation.

Speech-Recognition Technology

Speech recognition technology increasingly is being employed to enhance applications that traditionally require a user to enter commands on a keypad or keyboard. Complex applications can be accessed and used much more efficiently with verbal commands, on everything from personal computers to automated attendants.

New developments in speech recognition have helped the technology move forward. Processor speeds have multiplied while costs have dropped, speaker-independent capabilities have been added to speaker-dependent features, and recognition rates continue to climb. The company believes that these advances make the market ripe for rapid expansion.

DSP Communications, Inc.

DSPC, based in Cupertino, Calif., is a leading developer of chipsets for the wireless personal communications market. The company develops and markets integrated circuits and algorithms for the analog and digital cellular, wireless data communications, and voice recognition markets based on DSP technology and proprietary application-specific integrated circuits (ASICs). The company's products include integrated baseband chipsets for digital cellular voice and data, and voice-recognition solutions that provide high recognition rates in noisy environments. -0-

Note to Editors: Windows is a trademark of Microsoft Corporation.

CONTACT: DSP Communications Inc.

Patrick Corman, 415/326-9648

[e-mail]

Lisa Croel, 415/326-0487

[e-mail]