The Association for Voice Interaction Design, or AVIxD

Jenni McKienzie, Peter Krogh

SLTC Newsletter, October 2009

AVIxD (the Association for Voice Interaction Design) recently hosted its 8th workshop on voice interaction design. These workshops are held once or twice a year as an opportunity for voice interaction professionals to come together, put companies and competition aside, and tackle issues facing the industry.

Customization: How Does Your App Adapt?

The Summer, 2009 topic was, Customization: How Does Your App Adapt? As is often the case in the speech industry, there is a lack of consensus on definition of terms. "Customization" and "personalization" are used interchangeably by some, and held as completely different by others. Some think neither term extends beyond greeting a caller by name. During the course of the workshop, participants identified three levels of customization:

Situational Awareness

This is customization that can occur without knowing the identity of the caller and without caller opting for custom behavior. A simple example is a hurricane affecting both the volume of and reason for calls to a travel company. Situational awareness does not depend on new technology but leverages existing knowledge about the organization, the domain, and caller populations.

Individualized call flow changes

Identifying methods for managing different levels of customization was the objective of this group's session. Adapting call flow functionality for a specific individual and incorporating their known preferences and past behavior can provide a highly-personal experience, but can cross the line into a sense of invasion. What are appropriate uses of known user behavior? What are the current best practices? What is the goal for using historical criteria?

Data-adaptive dialog

This group looked at a couple of different but related issues. First was the use of the large amount of data available from in-service dialogs to better optimize them. Second was how maintaining accurate estimates of uncertainty within a dialog can be used to make it more stable to potential errors.

Look for in-depth articles on each of these three topics to be published at AVIxD.org in the coming months.

The Maturation of VUI

Already published on the website are articles from the 2008 and 2007 workshops. The 2008 topic was The Maturation of VUI: What I Wish I Knew Back Then. Participants discussed many topics around how VUI design practices changed over the last 10 years and produced five articles:

Naturalness: How Closely Should IVRs Mirror Human-to-Human Speech

An early mantra of voice interaction design was that if a person wouldn’t say it, neither should an IVR. But IVRs aren’t people, and the same conversational techniques don’t necessarily work in both situations. It is also a mistake to discard wholesale the abundance of linguistic knowledge that may be utilized in designing effective IVRs. This article examines which things transfer well and which ones don’t.

Designing for the Overall User Experience

A single phone call to an IVR is only part of a customer’s interaction with a company. Designers must consider all possible interaction points a user may have with the company and design their application as part of that whole. This paper covers caller-agent interaction, repeat caller behavior, marketing campaigns, etc.

The Evolution of VUI Design Methodology

Various groups and companies have tried to come up with a design process methodology. Nobody has really found one that has taken hold. What’s been found is that there is a broad array of tools available to the designer, and that different situations require using different tools, and sometimes in a different order. Process patterns have emerged over the years and some of these are covered in this paper.

The Role of Data in VUI Design

There is more and more data available, and it should be leveraged. There is more research being done and published about specific design principles. Data is there to drive design decisions on new projects. Post-deployment production data is the most valuable, under-utilized source of data.

Positioning IVR Self-Service

IVR has not enjoyed a level of popularity commensurate with its promise or potential. The number of bad speech IVRs still outnumber the good ones. To ensure that a system falls into the good camp, the reward of using self-service must outweigh the effort. If this objective is not met, callers will "zero out" and end up disinclined to use IVR in the future.

In August of 2007, the workshop focused on the profession of voice interaction design. What does the job entail? What are the qualifications? If someone wants to design voice applications, what educational opportunities are available? These three papers are also available at AVIxD.org:

  • The Role of the Voice User Interface Designer in Speech Technology Projects
  • Education, Mentoring, and Training for Voice User Interface Designers
  • Where will this Career Lead? Voice User Interface Career Paths

In February of 2007, the focus was on error handling. This workshop led to a series of four articles published in Speech Technology Magazine, beginning in April 2008. The workshop has provided a valuable opportunity to further and refine the practice of voice interaction design. Watch AVIxD.org in the months to come for information on the next workshops. The possibility exists for one in London next May. The topic currently proposed for August, 2010 in New York is:

2010: Voice Interaction in a Multi-Modal, Multi-Channel World

As additional modes of interaction gain traction, voice interaction designers must understand and design to the concept of a "meta session" where time, space and memory are respected regardless of which medium is in use. This topic will be discussed in detail at the August, 2010 workshop in New York.

Watch AVIxD.org for details on membership. If you incorporate the voice channel in your work, or are interested in speech design methodology, you will want to join AVIxD!

Jenni McKienzie is Senior VUI Designer at Travelocity. Email: Jenni.McKienzie@travelocity.com. Peter Krogh is Director of Solutions Architecture at SpeechCycle. Email: peter@speechcycle.com


Add A Comment

This is a captcha-picture. It is used to prevent mass-access by robots. (see: www.captcha.net)

Code in the picture:
Title:
Your Name(*):
Email:
Notify me of any further comments to this thread:
Website:
Comment(*):