Remembering Scott McGlashan
Voice-enabled self-service on the telephone is so ubiquitous that it almost seems it must have been inevitable. But the voice industry would not exist in its present form without the inspired leadership of those individuals who had the vision to develop early voice browsers and voice browser standards. Outstanding among these was Scott McGlashan, the dynamic and brilliant editor of the foundational VoiceXML 2.0 specification. We lost Scott this year after his short struggle with pancreatic cancer. He will be deeply missed, not only for his technical work, but also for his leadership and spirit. Scott's life reminds us of the impact that a single individual can have.
In the late 1990s, the growth of the World Wide Web and the founding of early commercial speech recognition companies inspired IBM, AT&T, Motorola, and Lucent to collaborate on developing the idea of a voice browser. A voice browser would be based on a high-level XML-based markup language, analogous to HTML for the visual Web. This language would separate the voice browser platform from specific applications, letting application developers focus on the user experience. Moreover, if this language were a standard, speech applications could be platform-independent, developers would only have to learn one set of development skills, and the foundations would be laid for a whole ecosystem of voice applications. The companies offered their work, VoiceXML 1.0, to the World Wide Web Consortium (W3C) to serve as the basis of a new official W3C standard for voice browsers. The W3C launched the Voice Browser Working Group in 1999. Scott was an early participant, joining Jim Larson as cochair in 2002. Scott's standards work had a firm grounding in science. He received master's and Ph.D. degrees in cognitive science from the University of Edinburgh and participated in the European VERBMOBIL and SUNDIAL projects that laid the technical foundations for much of today's speech standards and technology.
Scott led the VoiceXML team and was responsible for managing the progress of VoiceXML 2.0 through the huge number of new feature suggestions, bug reports, and change requests that the group received. The Voice Browser Working Group membership was large, and included people from many companies and cultures, all with different ideas about how the standard should take shape. Scott's job was to create consensus on the standard from this diverse group. His technical knowledge and skilled leadership enabled the group to make progress many times when the discussion seemed to be at an impasse.
Voice XML has been widely implemented. Scott's company, PipeBeach (later acquired by HP), built an early VoiceXML platform. This and other early implementations greatly contributed to informing the standard with actual implementation experience. As VoiceXML 2.0 was becoming finalized, the W3C collected 18 testimonials from companies that had implemented VoiceXML systems, evidence that, even in 2004, the standard was serving a valuable role. These companies included PipeBeach, BeVocal, HeyAnita, Tellme, Unisys, VoiceGenie, Loquendo, Vocalocity, and Voxeo. Since then, there have been many mergers and acquisitions involving companies in the speech industry, but all the major players continue to use the VoiceXML standard.
Scott played a significant role in other important speech standards as well. He was a coeditor (with Andrew Hunt) of the Speech Recognition Grammar Specification, the standard for representing speech recognition grammars, another foundational standard that helped make today's voice industry possible. He contributed to a number of other voice specifications, including VoiceXML 2.1, CCXML 1.0, SCXML, SSML, and VoiceXML 3.0.
Paolo Baggia, editor of the Pronunciation Lexicon specification and a colleague of Scott's from the W3C Working Groups, says, "I most loved and remember the talks during the night in casual places like the subway of Budapest, or pubs and restaurants, with all those handkerchiefs scribbled with diagrams and plans. I'm sure none of the many people attending W3C VB and MMI will forget Scott's leadership and profound technical expertise with a touch of humor and lots of humanity." Dan Burnett, coeditor of VoiceXML and now WebRTC, comments, "Scott McGlashan taught me how consensus-based standards should work. The hard part is forging the team—the rest is just details."
Scott's contributions to the standards out of which our industry grew are enormous. His energy, extraordinary leadership, and enthusiasm for speech applications are unique and irreplaceable. He will be greatly missed.
Deborah Dahl, Ph.D., is principal at speech and language consulting firm Conversational Technologies and chair of the World Wide Web Consortium's Multimodal Interaction Working Group. She can be reached at dahl@conversational-technologies.com.