ralphw@C7.IUS.CS.CMU.EDU (Ralph Hyre) (10/08/88)
[I presume this suggests that some of the CMU speech software will be ported to the thing. Speech I/O probably means the rumored AT&T DSP 56000 chip onboard, with appropriate connectors. One member of CMU's Speech group has apparently done some work with the machine, but he can't provide details.] Don't expect to be able to turn on your NeXT and talk to it in real time for a little while longer, at least. The Sphinx system is one of the major successful speaker-independent systems here, right now it takes a 3 systems with 68020s and Weitek floating point chips. (ie it would eat bare Sun-4's for breakfast) There's also a TMS 32010/20 DSP chip for signal processing. [disclaimer: I do not speak for the CMU speech group:-). My information may be old or outdated.] 07-Oct-88 10:06 "Maria J. Jones" NeXT Computer System From: "Maria J. Jones" <mj11+@andrew.cmu.edu> (The following is the text of an Oct. 5 news release from the Public Relations office.) Two important computer technologies developed at Carnegie Mellon University have been incorporated into the NeXT* Computer System. The machine is the first to offer an integrated speech input/output capability. And it also features Mach, a Unix-compatible operating system base for parallel and distributed computing. "The NeXT Computer System's advanced speech input and output features make it easy to port speech recognition and response software without special purpose attachments," says Professor Raj Reddy, director of Carnegie Mellon's Robotics Institute. "While I expect that all workstations in the 1990s will routinely have speech capabilities, NeXT provides us with the first integrated solution to this problem." "Mach provides features and facilities not available in other operating systems," says its chief researcher, Computer Science Associate Professor Richard F. Rashid. "It furnishes a basic kernel onto which a number of different operating system environments can be layered." Rashid says Mach's unique features include the ability to execute multiple activities or "threads" within a single program. It integrates communication with virtual memory management, letting the user send large amounts of data between programs at very little cost. Mach also permits creation and management of memory objects by user programs. These memory objects can be mapped into the address spaces of user programs and managed by server processes, allowing efficient implementation of system services such as transaction processing. Mach research at Carnegie Mellon began in 1984 as part of a Defense Advanced Research Projects Agency (DARPA) project targeted to the development of advanced parallel computing. The system was first brought on line early in 1986. Since then, the university has licensed and distributed Mach software to more than 100 manufacturers, universities and corporate research laboratories. Reddy says Carnegie Mellon is working with several computer companies to develop portable software libraries for effective use of speech technology by non-experts. Carnegie Mellon has been a leading speech research center for the past 20 years, and is a major contractor in DARPA's speech understanding program. To provide rapid technology transfer from this program to computer manufacturers and vendors, the university has established a speech consortium. NeXT is one of the members of the consortium. Carnegie Mellon, one of NeXT's original investors, owns one percent of the company's stock. ###