VoxML (Voice Markup Language) is a technology from Motorola for creating a voice dialog with a Web site in which a user can call a Web site by phone and interact with it through speech recognition and Web site responses. VoxML (TM) allows a developer to create a script of the conversation a user can have with an application program run by a Web server. The user calling in is connected to a client program called a voice browser. The voice browser in turn passes requests on to the Web server. The markup defined in VoxML is consistent with the Extensible Markup Language (XML), the strategic data definition language for the Internet. Using VoxML is intended to be no more difficult than writing a Web page using the Hypertext Markup Language (HTML).
Motorola is providing VoxML as an open interface and potential developers can download the development kit from Motorola's Web site. Motorola has already provided the technology to some early product developers, including The Weather Channel. Typical applications might include delivering weather information, flight information, and sports scores, as well as online product and service ordering.
How the User Uses VoxML
- The user would identify a Web site that offered phone interaction. For example, some e-tailing sites might allow users to order from the Web site by phone instead of by filling out a form.
- The company with the Web site would publish a phone number to call for their Web site.
- The user would call the number on a regular phone. (Initially, - that is, making the phone connection through the Internet itself - is not supported.)
- The user would get connected to a voice browser at the Web site.
- When connected, the user would hear a recorded voice that invited the user to interact with the Web site - for example, to place an order or choose some other option.
- As the user responded, the selected responses could trigger requests from the voice browser to a Web server application. The Web server might also send back new Web pages to the user's computer - for example, pictures of catalog items and order confirmations.
- The user and the VoxML application would interact until the transaction was completed. Then the user would hang up, ending the session.
How the Developer Develops a VoxML Application
- First, the developer designs and specifies the application, writing down the possible voice dialogs that will be possible.
- Working at a regular PC that has the development software installed, the developer uses VoxML, the user dialog statements, and the names of recorded audio response files to create a file that formally defines the dialogs.
- The developer (or someone else) records each possible voice inquiry, creating a speech file against which users' voice entries can be matched. The developer arranges for any predefined voice responses, which may be available from another application.
- The developer has the voice client on the PC so that the VoxML script can be tested.
- After testing, all files are put on the Web site server.
- The developer modifies the Web site to invite users to request the voice application.
- Once the VoxML application is set up, users can interact with the Web site through their mouse and keyboard as usual, and also, for invited purposes, over a telephone.