Speechnet: a network of hyperlinked speech-accessible objects

The dominant technology for electronic communication and commerce, the telephone, does not discriminate between sighted and non-sighted users. The Web does discriminate. The Web is primarily based on text and visual scanning. Much of its functionality is inaccessible to visually-challenged users, or in applications where visual scanning is inappropriate. A solution is to augment the Web with a network of speech-accessible hyperlinked objects. Each of these objects has an associated grammar which defines the language that the object can respond to. These grammars are downloaded to "speech" browsers which use them to configure their speech-recognizers. This results in highly accurate user-independent continuous-speech interfaces to remote knowledge and functions. A prototype Speechnet has been constructed using IBM's Via Voice speech technology and a common Internet communication protocol.