Essentially a standard represents an agreement among a community about the meaning of a term or on a way of doing things. In some cases standards are arbitrary, for example, which side of the road you drive on, and in some cases standards represent an agreement on best practices. A standard might be enforced legally, such as building codes, or food safety, or it might just be an agreement within an industry on how to do things. In this section we present a classification of the different types of standards which will lay the groundwork for the discussion of specific standards in the later sections.
Architecture: Components and Communication
Architectural standards define the overall organization of a system, its components, their functions, and how they communicate. In the context of multimodality, we will discuss the World Wide Web Consortium’s Multimodal Architecture and Interfaces standard (Barnett, Dahl et al., 2008) and the older DARPA Communicator standard (Bayer, 2005) and describe their commonalities and differences.
Architectural standards describe how functions are allocated among specific hardware/software components and how they communicate. The goal of architectural standards is to ensure interoperability of components, even if they are developed completely independently. The World Wide Web is an excellent example of an architecture which supports independent servers, clients and applications with a high level of interoperability
Carefully-defined communication standards are critically important if components developed by different organizations are to interoperate. Communication takes place at several levels. The underlying protocols, such as TCP/IP and HTTP, will not be discussed here since they are not specific to multimodal systems, but do need to be referenced in multimodal standards in order to insure interoperability. Higher level communication protocols specific to multimodal systems, which we will discuss here, include the high level multimodal interaction life cycle events defined in (Barnett, Dahl et al., 2008) as well as standards defining the format of data payloads for the representation of user input. These include the Extensible MultiModal Annotation (EMMA) specification (Johnston et al., 2007) and InkML (Chee, Froumentin, & Watt, 2006) for representing stylus traces. We will also discuss the Media Resources Control Protocol (MRCP) (Shanmugham & Burnett, 2008). MRCP controls speech media servers which perform the functions of speech recognition, speech synthesis, and speaker recognition. Finally, we discuss some biometric standards, such as the BioAPI, being developed by the BioAPI Consortium.