Introduction

There has been significant development of voice technology over the last 20 years. Advancements have brought an increasing adoption, and today many various forms of voice technology are being used. Humans' exceptional ability to process and produce speech means voice technology could potentially disrupt many industries. Although voice technology has been vastly improved, there still exist several usability flaws. One such shortcoming is false expectations of a system's capabilities, which could potentially lead to users boycotting voice applications. I conducted the thesis for the UX Team at ASSA ABLOY Shared Technologies, the company R&D department located in Stockholm. Thus, the focus of the thesis was on voice technology in an access solution context.

Objective

The objective of the thesis was to investigate how synthesized voice attributes, such as gender and vocal cues, forms expectations. A Wizard of Oz-testing methodology was used to evaluate the effects of gendered synthesized voices and conversational strategies.

Design Process

Although the thesis followed a technical structure, the design process was treated as a separate part to avoid fragmentation. The complete process can be read in Chapter 3 in the paper available at the bottom of this page.

Competitor Analysis

Two separate competitor analyses were conducted to generate ideas and identify opportunities and weaknesses in voice technology and smart locks. Both products on the market and crowdfunding platforms were researched to investigate existing and prospective products. SWOT-analyses were then used to identify problem areas and possibilities.

Design Sprint

I conducted a design sprint following Google Venture's format after having identified a few hypotheses. I set out to investigate how users perceived conversational strategies. In the five days of the sprint, I went from idea to a functional Wizard of Oz-prototype. The critical elements of the sprint process included:

• Problem Definition
• Ideation using four-step sketches
• Storyboards
• Sample dialogues & Flow Charts

Second Iteration

With insights from the design sprint, I began a second iteration. Initially, I conducted interviews with experts on security and persuasive ethical design to gain further knowledge to develop personas, both for users and the voice user interface. After the interviews, a process similar to the design sprint followed to create a prototype.

Wizard of Oz-Testing

A scenery resembling a home was constructed in the UX Lab at ASSA ABLOY to create an authentic environment. The voice-controlled device that the users interacted with was mounted on the door. In the form of the "wizard", I controlled the synthesized voice outputs from inside a control room. The control room had one-way mirrors which meant the test participants could not see the wizard controlling the synthesized voice responses remotely.

Data Processing

From the tests, various data was collected. The data was gathered from both recordings and structured interviews from all sessions. All material, both qualitative and quantitative, were then made physical to allow processing and analyzation. After sorting the data in groups of Actions, Environment, Interactions, Objects, and Users, abstraction was used to remove noise. From there on, the data was able to be analyzed, and connections could be drawn.

Key Takeaways

Combining quantitative and qualitative data

All does not have to be black or white. In this project, I had the opportunity to combine the two different methods for gathering data. Putting the strengths and weaknesses of the two methods against each other provides truly valuable insights. Only using one approach can many times prove deceitful.

Ethical Design

Technologies, specifically new ones, can have several foreseen and unforeseen ethical implications to users. Although there is no existing framework for who should be responsible for such consequences, designers possess considerable power. Thus, it is crucial that designers not only consider user experiences but also look at the bigger picture and try to detect possible ethical implications.

Structuring big projects

I learned much about structuring and planning for big projects. The thesis was required to be both of a high scientific level as well as applying an evaluative UX approach. Focusing on one area at a time, separating the technical research and design process was tempting. However, the ability to work with the academic and design process simultaneously proved very useful for maintaining a structured project.