Call for a public-domain SpeechWeb

Abstract

Imagine using your cell phone to navigate through a network of hyperlinked speech applications as rich and diverse as the visual applications on the conventional Web. On holiday at the beach, you could use it to navigate through a vendor's site for information on the new camera you're using to take snapshots of your friends. In the evening, by the fire, you could verbally browse a library of electronic stories, picking the one most appropriate for the moment. Imagine, too, being able to build speech applications and add them to a public-domain SpeechWeb as easily as you build and add pages to the conventional visual Web. You could have a speech-accessible resumé. Information (such as directions on how to get to your home) could be made available for your guests to access through hands-free speech in their vehicles. Companies (with the expertise) could create natural-language speech interfaces to their products and services, making them available to anyone, anywhere using a handheld SpeechWeb browser. Moreover, a public-domain SpeechWeb would allow blind users to participate more fully in our knowledge-based society.

Mobile Voice Access in Social Networking Systems

“…Many VoiceXML portals provide developers with a variety of tools for creating telephone-based applications with voice interface. In addition to enabling applications to provide users with a natural language interface to access and retrieve content from the web [3,8], VoiceXML portals have been successfully used to implement a wide variety of desktop and mobile applications, from transcribing medical records [2] to assisting police officers communicate with computerized information systems hands-free during critical situations [4]. Mobile applications are positioned to gain the most from voice interfaces [9]; for example, a museum in Utah successfully used a VoiceXML solution to let their patrons take a self-guided audio tour [10].…”

MySpeechWeb

“…The architecture and speech browser have been described in previous publications, [1] and [2]. The new work reported in this paper, to be demonstrated for the first time at the ASSETS'08 conference, includes the complete open-source implementation of the architecture, the suite of exemplar speech applications, and a web site with extensive end-user and developer documentation.…”

“…Our open-source platform is based on the Local speech Recognition and Remote Processing (LRRP) architecture [2], in which a speech application consists of three components stored in a conventional web directory on a server connected to the Internet: 1) an executable text-in/text-out program written in any programming language supported by the server, 2) a file containing a grammar defining the application's input language, and 3) a file containing a small script written in a speechmarkup language such as VXML or X+V.…”

Realization of natural language interfaces using lazy functional programming

“…The database is also defined (within the program for prototyping or in external files). For example, The W/AGE environment has been used to create natural language databasequery interfaces which are hyperlinked in a Public-Domain SpeechWeb [Frost 2005] and which can be accessed through speech recognition interfaces running on remote lightweight end-user devices. Lavier [1990, 1993] developed a workbench in Miranda for experimenting with implementations of Montague-like approaches to natural language processing.…”