Voice User Interface Design

As part of my professional education, I took Voice Design classes. I like new techs and I think that AI, machine learning, voice are going to be used a lot in the next generation of applications. So I decided not to wait but instead to anticipate and to be at the forefront of these new technologies. During this course I learned the voice design process, and I created skills that I was able to test with the Alexa device.

The Challenge

The voice user interface design process is slightly different than the graphical user interface design process. The overarching problems of how to design your interface to best communicate the information you need to get across and how to create the best user flows remain the same. The information architecture of voice, however, among other aspects, is a bit different. So because I never designed for Voice before the challenge here was to put aside what I knew and to start with a open mind. 

Personas & Skill Map

As a UX designer, I'm already familiar with creating user personas for mobile, desktop, and web, and I've used some common user research techniques such as user interviews and surveys to gather information about my personas. Knowing how to design detailed personas helps me better communicate with my product team, leading to more effective product and business decisions. From now on, I’ll refer to these personas as user personas. Another type of persona I dealt with in voice design is the system persona—in other words, the identity of my voice interface.

My whole work was to build a new skill for Alexa called 5-minute recipe. You can access the project brief here and look at how Alexa skill is working there.

Based on limited data and assumptions, I was able to create the system persona and the user persona as you can see above. It was clear that only having the basic features would be the way to go about solving this challenge. After some research, card sorting, I decided to keep simple the features as seen in the relatively simple skill map below.

User Needs & User Stories

For any good design, I needed to make sure I've identified an unmet user need. First, I needed to determine how those needs and pain points relate to the skill requirements. Then, I could craft user stories to drive my design activities. See my findings.

Sample Dialogs

Sample dialogs are a lightweight way to design a successful conversation. They’re simple to write, but powerful in their ability to show whether a conversation is working or not. To write a sample dialog, all you need is a word-processing app or some pen and paper. A simple set of sample dialogs will get you a long way towards a usable voice experience.

User Flows for Voice

I'm already well aware of the value of user flows in building a visual design experience. For voice design, instead of thinking about individual pages, I needed to think about the different intents necessary for interacting with my app as seen below.

Voice Scripts

Another step in the Voice Design process is the necessity of creating scripts. Scripts include every line your system will read out to the user. They’re comprehensive and serve as the definitive source of truth for engineers as well as those who do quality assurance on your application. Scripts also often contain the utterances a user can say to obtain those responses. Scripts are incredibly important.

So I created scripts related to my recipes skill as shown below or here to have the breakdown.

Usability Testing - Testing my Skill

Testing a speech experience is, overall, very similar to testing a visual experience. My objective here was the same—to determine whether users can reliably and easily use my application. And, from that test, I could determine areas in need of optimization (points of friction) in order to ensure a smooth user experience.

Find my Usability Test Plan here

Building the Skill - Going The Dark Side

Once I had created my personas and skill map, I needed to implement this new skill in Alexa system. That is where things got more complicated. I programmed and tested my own version of the skill using the Alexa environment called AWS Lambda. While the skill itself was simple in terms of functionality, it gave me a basic understanding of how Alexa skills are built as seen below. 

Certifying my Skill

The Alexa Skills Marketplace is growing every day. All skills on Alexa have been meticulously tested by the experts at Amazon. If I wanted my skill to pass certification, I, too, needed to test it extensively before submission. I also needed to ensure I followed all of Amazon’s skill guidelines, especially in regards to its invocation name.

Once I felt my skill has been tested properly and met all of Amazon’s requirements, I submitted it for certification as seen below.

Conclusion

My skill was not certified at the first time. I made a few changes before it was validated. Now I'm able to use my skill on my own device and I must admit that it is pretty fun.

To sum up, the voice design process is very similar to a visual experience. I enjoyed learning new processes, tools and methods and I'm really excited to apply my voice design skills in my daily job.