| About InfoWorld : Advertise : Subscribe : Contact Us : Awards : Events : Store |
|
||||
|
|
||||
|
Speech vendors target developers with multimodal tools By Ephraim Schwartz July 15, 2002 7:10 am PT DUELING SPEECH TECHNOLOGY announcements will come to the fore on Monday when the SALT (Speech Application Language Tags) Forum ships Version 1.0 of its multimodal specification and IBM reveals its X+V technology for multimodal applications.
Multimodal refers to a developer's ability to mix both a voice and visual interface into a single application. The IBM technology, X+V, an acronym for xHTML plus Voice, will ship in October and will be included as an extension to Big Blue's WebSphere Everyplace Access in the first half of next year. "A brokerage customer requesting trading positions uses his voice, but he doesn't want to hear the results because they can be fairly long. With multimodal, the request is verbal and the response is visual," said Sunil Soares, director of product management for IBM's Pervasive Computing initiative, in Somers, N.Y. IBM submitted its specification late last year for X+V to the W3C and it has been acknowledged -- terminology for the standards body's willingness to consider a specification as a standard. The W3C has, in fact, set up a working group to focus on multimodal, Soares said. "The spec lets you operate across three modes. It lets you operate in a multimodal environment, but if you want to use just visual it ignores the VXML tags and vice versa," Soares said. In October IBM also will ship a set of tools for developers to write multimodal applications using VXML. Despite appearances of dueling standards, the W3C will most likely come up with a single solution, according to Bill Meisel, president of TMA Associates in Tarzana, Calif. "VXML will be broken into pieces and some of the same ideas and concepts in SALT will be applied to those pieces," Meisel said. However, there is one non-technical difference between the two proposed standards that could lead to problems down the road. While the SALT specification is royalty-free the VXML specification is not, Meisel said. "The technical people on the VXML Forum wanted it royalty-free, but they were over-ridden by their companies. W3C could never get the VXML members to drop all of their IP requirements," Meisel said. However, it is highly unlikely that any company will try and sabotage the VXML initiative by claiming someone is violating its patent, Meisel added, although the possibility does cloud the issue. The release on Monday by the SALT Forum is the first fully functioning specification suitable for developers to take and build applications around, said Peter Gavalakis, marketing manager, at Intel in Parsippany, N.J. Gavalakis also sees the possibility that the standards organizations might take a look at the different paths toward multimodal and decide to converge them. However, there is another option. "They might provide multiple options, one more high-level using XML -- and that would be VXML -- and a more low-level solution using SALT," Gavalakis said. As to the long-standing disagreement between SALT and VXML proponents, it appears that it still continues. Soares called SALT "verbose." "From a developer perspective, SALT requires you to write a boatload of Java script to manage voice dialogues," said Soares, adding that VXML, on the other hand, has an FIA (forms interpretation algorithm) that automatically manages voice dialogues. Salt's representative, Gavalakis, didn't disagree but had his own counterpunch. "He's [Soares] right. The strength of VXML was engineered to enable telephony applications and having the FIA embedded provides a faster development time for certain types of applications that are system directed, that use prompts. But Web applications and multimodal applications are more user-directed and FIA limits your flexibility," Gavalakis said. TMA Associates' Meisel had the final word: "The W3C will sort it out." InfoWorld Editor at Large Ephraim Schwartz is based in San Francisco. SPONSORED WHITE PAPERS
SPONSORED LINKS
|
||||||||||||||||||||||||||||||||||||
|
||||||||||