Designing for the voice interface. Part two

Ux for VUI

In this, the second part of the article I will start to explorer the Ux design considerations for the voice user interface (VUI)

User Research

Probably the most important element of the Ux process is user research and should be carried out at regular intervals throughout the life of the product. It is especially important in the initial scoping phase where some of the fundamental needs related hypothesise are being tested. This usually takes on two forms, questions and observations.

Participants for the research are best chosen from the existing customers/user base. If the existing users are unreachable or don’t exist (as in the case of a startup product) then they should be selected by user segmentation and demographics to be approximately representative of the early adopter user/customer that the client would be aiming for.

A set of standard questions should be drawn up that tests the assumptions that the clients have about how the customers use or will use their digital product should primarily explore the users goals and needs The survey should have questions that examine the mental model the users have of how the task required could be carried out. attention should be paid to which channels (web, app, email, phone etc) the users would expect to use to complete the task (the research might suggest that it may not be limited to just one channel. Questionnaires can be sent to a large group of candidates. This research could be carried out face to face or remotely. The responses received could indicate candidates that would be ideal for, and could be recruited to form, a subset of users that could be called upon for further observed testing.

Users should initially be observed performing the task they need to achieve that will eventually be done by the system that is being designed. These tasks are best observed within the environment that they would usually be performed. At a later stage those same users could be revisited and observed performing those same tasks with a prototype (wizard of OZ or higher fidelity) of the new system and comparisons made. This testing should be carried out for every iteration of the design (unfortunately fund rarely allow for this). If the task requires long term and regular use of the product then users could be asked to keep a diary.

When observing the users performing the task or using the digital product researchers should record carefully the nature of the environment that the user is performing the task in and where appropriate make audio, and photographic records.

Much of these processes are not unique to VUI design so special attention and thought should go into those that require a specialist knowledge and analysis.


The UX designer's role puts them in a slightly uncomfortable position where they have to equally satisfy the client (business) and the customer (user). Unfortunately these two groups may have radically different needs.

Obviously, the needs of the business are important but It must be remembered that we are designing for the user and, we are designing for speech. As M. H. Cohen, Giangola, & Balogh (2004)01 put it, “spoken language skills are learned implicitly at a very young age rather than through an explicit educational process the VUI designer must work on the user's terms”

The design techniques that have been used to develop some of the better IVR systems that we have seen over the last two decades are a useful start point, although Cathy Pearl (2016)03 points out that “some of the design strategies from the IVR world also apply to mobile VUI design (as well as VUI for devices), mobile VUIs also present a unique set of challenges (and opportunities)”

In order to be able to design for the VUI we first have to understand the how users relate psychologically with the technology that they are interacting with. Voice interaction is seen as a more human like interaction than graphical interaction and, as such, the social needs of the interface must be understood. Reeves and Nass (1996)04 explored these issues through several lenses. They looked at manners, personality, emotion, social roles and form.

The designer must evaluate whether the system should be voice only or offer a multimodal approach. For example large hierarchical menus presented verbally are very hard for the listener to remember, which is why it may be better to design these for multimodal input/output. Utilising the mobile device’s screen in addition to verbal/auditory signals will give a much richer user experience. In addition, Cathy Pearl (2016)03 thinks that “this visual component can allow the user to continue at a more leisurely pace.”

Alternatively an intelligent VUI with enough vocabulary and linguistic programming could actually improve the navigation of hierarchical menus. As Cohen and Oviatt (1995)02 state, “Natural Language Interaction can short cut the navigation of a menu hierarchy to invoke known commands like picking out an individual object, a set of objects, time period, and so forth.” Such a system would need only a small amount of coaching of the user to be fully effective Thus reducing the need for a multimodal approach unless it was to give the user a choice.

The designer must also determine whether the system should adopt a one turn approach of a question and then an answer or be more of a conversational between the user and the system. It should be pointed out that this should be decided by the sophistication of the AI underlying the speech system used. As Cathy Pearl (2016)03 reminds us “The level of frustration is about where we’re at with many VUI systems today. Despite the many recent advancements of speech recognition technology, we're still a long way from simulating human conversation.”

It also has to be remembered that the current crop of voice recognition digital assistants are currently most suited to handling one-off tasks, such as providing the answer to a search query, adding an appointment to your calendar or launching a playlist future digital assistants will be able to anticipate further and related tasks.

Cohen et al (2004)01 suggests that “testing comes at the end of the development stage. Like with all other interaction design processes the user journey has to be carefully considered and all iterations thoroughly tested.” he refers to it as “the detailed design phase” and traditionally in UX design it would utilise such techniques as, Wizard of Oz testing and wire framing, user interviews and observation and scenarios. This approach is similar to the waterfall design process, but current UX design thinking would disagree with this suggestion as designers are seeing the benefit in an agile design strategy and that user testing should be done early and often. As Anthony Vivano explains it “we seek quick customer validation or early failure.” (Viviano, 2014)06

The main objective for any digital product design whether for GUI, VUI or for multimodal use must the creation of a positive user experience. As Darla Tucker (2003)05 asserts, “Users who have had a positive experience using an automated speech system feel valued by the company deploying the system, and are more likely to continue to do business with such companies in the future.”

In part 3; I shall be looking at the persona and dialogue of the interface

  1. P., & ().Voice: User interface design (2nd ed.). Boston, MA: Addison-Wesley Educational Publishers.
  2. , & . (). The role of voice input for human-machine communication. Proceedings of the National Academy of Sciences, 92(22), 9921–9927. doi:10.1073/pnas.92.22.9921
  3. . (). Designing Voice User Interfaces United States: O’Reilly Media, Inc.
  4. ., & . (). The Media Equation: How People Treat Computers, Television, and New Media like Real People and Places (2nd ed.). United Kingdom: C S L I Publications/Center for the Study of Language & Information.
  5. . (). Voice user interface design – purpose and process. . Retrieved from
  6. . (). The lean UX manifesto: Principle-driven designsmashing magazine. Retrieved January 14, 2017,
Please follow and like us:

Also published on Medium.

Leave a Reply

Your email address will not be published. Required fields are marked *