Speech Technology: WordPerfect Working on Windows Speech API

Since the first practical implementation of the typewriter by Christopher L. Sholes in 1867, a large portion of how we think about communication has been defined by this device. However, as time moves forward and computing devices increase in sophistication, the keyboard is quickly forming an information bandwidth logjam.

Speech recognition will certainly form an integral part of future stylus-based computers, but not until the technology conquers the desktop. WordPerfect Corp. is teaming with several leading application and speech developers to create an open speech recognition application program interface (API) for Microsoft Windows.

The participating companies, listed in the sidebar in this article, are working to develop a standard that will enable programmers to seamlessly integrate speech recognition technology into Windows applications, while at the same time enabling speech recognizers from multiple vendors to work with the same application.

Mark Calkins, vice president of product and strategic marketing for WordPerfect Corporation expressed the goal of the project as follows: “Speech recognition will become a standard productivity tool that will allow users to enter task directions verbally and even create text and data by speaking into their computer instead of typing or using a mouse.”

WordPerfect has conducted usability studies that show that the level of integration between the application and the speech recognition engine has a great effect on how natural the user interface appears. WordPerfect and its partners hope that speech-enabled applications relying on a common API will create a more consistent environment for users and permit certain tasks to be completed more naturally using voice.

Support for Speech

The Speech Recognition API will include support for both continuous and discrete command control capabilities. In addition, the API will also enable people to input text and data, again using continuous and discrete dictation.

Discrete dictation requires each spoken word to be separated by a brief pause. While this is unnatural for normal data input, such as dictating a letter, it is reasonably well suited for navigation and command control within an environment such as Windows.

Examples of these types of commands include “File, Open” and “Cancel.” An example of a continuous command would be something like “print two copies of this page.”

WordPerfect will offer the Speech Recognition API as part of its PerfectFit Technology, which forms the basis for all WordPerfect Windows applications. The technology will also be available to third-party developers using WordPerfect’s PerfectFit Software Developers Kit.

The Speech Recognition API with the PerfectFit SDK will be available in the fourth quarter of this year. WordPerfect also plans to include the Speech Recognition API in its Windows applications by the early part of next year.

Dragon Systems, Inc.
Contact: S. Ming Lee
(617) 965-5200
(617) 527-0372 (fax)

IBM Corporation
Contact: Walt Nawrocki
(407) 982-4033
(407) 443-2778 (fax)

Kolvox Comm. Inc.
Contact: Isaac Raichyk
(416) 218-3100
(416) 218-3110 (fax)

Kurzweil Applied Intelligence
Contact: Mark Flanagan
(617) 893-5151
(617) 893-6525 (fax)

Lernout & Hauspie
Contact: Bart Verhaeghe
(617) 932-4118
(617) 932-9209 (fax)

Novell, Inc.
Contact: Jessica Kersey
(408) 577-8739
(408) 577-5104 (fax)

Philips Dictation Systems
Contact: Jamie O’Donnell
(415) 923-1660
(415) 923-9863 (fax)

WordPerfect Corporation
Contact: Deborah Hendrickson
(801) 225-5000
(801) 228-5077 (fax)