Dragon Systems demonstrates first PDA speech recognition technology on the Digital StrongARM Processor in Apple Newton MessagePad 2000.NEWTON, Mass.--(BUSINESS WIRE)--March 25, 1997-- World Leader in PC Speech Recognition Moves Onto PDAs Dragon Systems Dragon Systems, Inc., was the company that created DragonDictate and Dragon NaturallySpeaking. It was founded in 1982 by Drs. James and Janet Baker and bought by Lernout & Hauspie in 2000. , Inc., of Newton, Mass., has developed and ported high performance speech recognition to run on the new Digital StrongARM-based Apple Newton (computer) Apple Newton - A Personal Digital Assistant produced by Apple Computer. The Newton provides a clever, user-friendly interface and relies solely on pen-based input. MessagePad 2000. This Newton platform is designed as a powerful, compact highly mobile computing Using a computing device while in transit. Mobile computing implies wireless transmission, but wireless transmission does not necessarily imply mobile computing. Fixed wireless applications use satellites, radio systems and lasers to transmit between permanent objects such as buildings platform. Battery powered and weighing less than two pounds, the MessagePad 2000 contains the high performance, low power Digital StrongARM 32-bit RISC processor RISC processor [Reduced Instruction Set Computer], computer arithmetic-logic unit that uses a minimal instruction set, emphasizing the instructions used most often and optimizing them for the fastest possible execution. . The StrongARM microprocessor, under license from Advanced Risc Machines (ARM) in the U.K., runs at 160 MHz (MegaHertZ) One million cycles per second. It is used to measure the transmission speed of electronic devices, including channels, buses and the computer's internal clock. A one-megahertz clock (1 MHz) means some number of bits (16, 32, 64, etc. and dissipates less than 1/4 watt of power. Well-suited to many restricted vocabulary tasks, it features continuous digit string recognition in a variety of modes found in high volume applications. Such application environments span factory floors, healthcare environments, stock exchanges, wireless communications wireless communications System using radio-frequency, infrared, microwave, or other types of electromagnetic or acoustic waves in place of wires, cables, or fibre optics to transmit signals or data. , car interiors, and home living rooms. Typical tasks requiring this type of capability include: -- Voice dialing phone numbers -- Ordering item quantities, weights, and part numbers -- Specifying zip codes for sortation Identifying objects that are stamped with a bar code and routing them to the appropriate destination. Sortation is typically a high-speed process used in the transportation industry by companies such as Federal Express, UPS and others. See sort and bar code. -- Recording meter readings and mileage records -- Reporting quality control measures -- Reading Universal Product Codes (UPCs) -- Command/control of Network Computers (NCs) -- Changing channels on TV set-top boxes From technician to mobile business executive, the advent of low-power, handheld, belt-worn, and stand-alone information appliances offers faster and easier command/control, data entry, and information retrieval information retrieval Recovery of information, especially in a database stored in a computer. Two main approaches are matching words in the query against the database index (keyword searching) and traversing the database using hypertext or hypermedia links. . Features & Benefits -- "Small vocabulary" -- Continuous - The software can recognize phrases and sentences that are spoken continuously, i.e. in a natural way without pausing between words. The vocabulary and grammar can be context sensitive. "Small vocabulary" in this case means that the grammar perplexity perplexity - The geometric mean of the number of words which may follow any given word for a certain lexicon and grammar. is typically 50 or lower. The actual vocabulary size may be quite large (thousands of words); however, the number of words or phrases that are active at any one time must be 50 or fewer if a real-time response is required. Higher perplexity grammars can be used if a response time of a second or more is acceptable. -- Real-Time Performance - Subject to the restriction on grammar complexity mentioned above, the software will return a recognition result within one second after the user stops speaking. -- Whole word and Sub-Word (phonetic) Voice Models - The system can handle both whole-word and sub-word (phonetic) acoustic models. Indeed, a single application can use both types of models. Whole word models can be more accurate when the vocabulary is restricted to a set of words for which Dragon has previously built models. With sub-word models, however, the vocabulary and grammar are essentially unrestricted. -- Speaker-and Gender-Independent Voice Models - Both whole-word and sub-word models will deliver high accuracy without any training to most users, and without knowing the gender of the user. -- Speaker-Dependent, Whole-Word Models - The recognition software will also allow a user to build a custom, or "speaker-dependent" whole-word model by providing the spelling and a single spoken instance of a word. -- High Accuracy and Noise Robustness - The accuracy for recognition of complete five-digit strings (zip codes, for example) is typically in excess of 98%. Special processing algorithms make the software more robust against background noise. -- Small Memory Footprint Memory footprint refers to the amount of main memory that a program uses or references while running. This includes all sorts of active memory regions like code, static data sections (both initialized and uninitialized), heap, as well as all the stacks, plus memory required to hold - The static or non-volatile memory Refers to memory chips that hold their content without power being applied. It may refer to chips that are not changeable, such as ROMs and PROMs, or to chips that can be rewritten many times such as flash memory. required for a typical vertical application is approximately 1 MB. During recognition, the software requires an additional 250 KB ofstems, Inc., is a worldwide leader in PC speech , Germ an, Italian, Spanish and Swedish), enable users to create any text, issue commands, and enter and access data simply by speaking. The company also licenses its ns, and is pursuing leading-edge research and distered trademark. All other trademarks are of their respective companies. 1996 Dragon Systems, Inc. All designs, manufacturers and markets industry-lRISC Machines, Ltd. Mitsubishi Electric Mitsubishi Electric Corporation (三菱電機株式会社 Corporader in open client/server solutions from personal computing to integrated worldwide information systems. ions from business partners, help organizations compete in today's global marketplace. Digital, Digital Semiconductor and the Digital logo are trademarks of Digital Equipment Corporation, Windows NT is a trademark of Microsoft Corporation. Intel and Pentium are registered trademarks and Pentium Pro is a trademark of Intel Corporation. StrongARM is a trademark of Advanced RISC Machines, Ltd. CONTACT: Renee L. Blodgett Dragon Systems, Inc. (617) 965-5200 ext. 348 Reneeb@dragonsys.com |
|
||||||||||||||

Printer friendly
Cite/link
Email
Feedback
Reader Opinion