NEW YORK IBM Corp. and Nokia, the mobile phone giant, announced an agreement this week to collaborate on long-term research and development of speech interfaces and speech-recognition technology for mobile devices.
In addition, Nokia said it will participate in the development of the VoiceXML programming language standard.
The IBM-Nokia agreement extends an agreement between the companies announced earlier this month at Telecom 99 involving the promotion of the mobile Internet and a collaboration on wireless application protocol (WAP) solutions. WAP is a global standard that allows mobile devices to connect to the Internet.
Like the companies' earlier WAP pact, the speech collaboration effort is designed to move a mobile Internet forward. It may also lead to better speech interfaces for numerous devices in the long term, because Nokia, with its strong position in communications platforms, has already tackled the challenge of noisy environments for mobile systems. Allowing consumers to access information in natural conversational language will be key to enabling next-generation portable devices, the company said.
"Nokia is committed to useability and user-friendly access and we see the voice interface as a complementary element to simplify user interfaces," said Petri Haavisto, director of Nokia's speech and audio labs, which operates from Tampere, Finland.
Haavisto said speech interfaces are especially useful in situations where eyes and hands are busy, displays are inconvenient, or users need a quick shortcut to access information.
Speech-recognition technology is part of Nokia and IBM's vision of the mobile Internet, and both companies see its possible application across a range of products.
"We both foresee where the business is going that speech will be an integral part of mobile products," said W.S. "Ozzie" Osborne, general manager of IBM's Speech and Pen Systems (West Palm Beach, Fla.). "And we believe we'll be able to satisfy that faster by working together."
Under their joint development agreement, speech researchers at IBM will work with Nokia researchers on speech interfaces for mobile devices, and will research multi-modal Internet browsing and other speech-enabled applications. Nokia's Haavisto declined to say when devices using IBM speech technology would be available, but described the types of devices Nokia will design.
"One of the trends we see is mobile devices that will have such a natural speech interface that the user won't be able to tell whether the speech technology is in the device or in the network," he said. "We expect to see VoiceXML implemented in the network. That's why Nokia is interested in VoiceXML."
The agreement gives Nokia access to IBM's speech technology portfolio, including the ViaVoice Millennium desktop dictation engine; directory dialer, natural language technology and a voice recognition engine for telephones; smaller speech engines for embedded products such as cells phones and cars; and a voice-enabled Internet browser.
Nokia has already licensed IBM's ViaVoice Directory Dialer, a voice dialer technology now used in Nokia's corporate network and available for licensing to Nokia customers.
Nokia has made its own forays into speech technology. Company researchers have developed speech-recognition technologies and well as speech coding and other techniques for manipulating speech in noisy environments, Haavisto said. This technology has been used to implement voice dialing in several small handsets for GSM and for Japan's mobile phone system.
But Nokia wants to go beyond voice dialing and voice control to create a more natural conversational speech interface for its mobile communication devices, Haavisto said. Nokia believes IBM's portfolio of speech technologies and expertise in natural language interfaces will support this goal, and help Nokia "deliver the Web in your pocket," said Haavisto.
In the belief that speech technology will be critical to next-generation mobile devices, Haavisto said Nokia is lending its support to the VoiceXML Forum, which is designing a voice programming language standard. Nokia will work with forum cofounders IBM, AT&T, Lucent Technologies and Motorola on the effort.
"Voice is an additional access method and we believe whether it will happen will depend on the development of open standards," said Haavisto. "We want to influence VoiceXML so that it is compatible with mobile Internet and mobile communications."
The VoiceXML Forum recently released its 0.9 specification for comment, and expects to receive approval for the spec from the World Wide Web Consortium (W3C) in the first quarter of 2000.
At the announcement of its agreement with Nokia this week, IBM demonstrated a speech-enabled Web browser that uses the VoiceXML 0.9 specification. The browser, which can be download for free from the IBM Web site, uses IBM's ViaVoice Millennium speech engine, Websphere application server for e-commerce, interactive voice response and other IBM speech technologies.
The browser will be licensed to various IBM partners and will be available to developers wishing to write applications to it. IBM also plans to license the technology to other OEMs, Osborne said.
In addition to the Nokia pact, IBM announced that the Voice Technology Initiative for Mobile Enterprise Solutions (VoiceTimes) has released its first specification for the design of digital recording and mobile devices that use deferred speech recognition. The specification provides developers a standard interface for implementing voice in mobile enterprise environments.
The VoiceTimes specification includes audio hardware and signal specifications, a microphone, analog signal processing, a codec, audio signal processing, digital data compression, restored pulse code modulation and electromagnetic interference.
The VoiceTimes initiative, announced in April 1999, intends to have products incorporating the spec available by the end of 1999. Participants in the initiative include IBM, E Digital, Intel), Philips, Olympus, Dictaphone Corp. and Norcom Electronics Corp.
The VoiceTimes spec can be downloaded from the Web.