UX Design Considerations For Voice Input (VUI)

UI/UX Design

An illustration of a head, with four speech bubbles coming from them. These are representing the various voice assistants a user can talk to, such as Alexa, Siri, Google, and Mercedes. This illustration is on a darker navy blue background.
An illustration of a head, with four speech bubbles coming from them. These are representing the various voice assistants a user can talk to, such as Alexa, Siri, Google, and Mercedes. This illustration is on a darker navy blue background.
An illustration of a head, with four speech bubbles coming from them. These are representing the various voice assistants a user can talk to, such as Alexa, Siri, Google, and Mercedes. This illustration is on a darker navy blue background.

Voice control for technology is nothing new - the first system was introduced in 1952. However, the introduction of Google’s Voice Search platform in the 2000s was the drive that really launched voice-control for technology into mainstream usage.

By using its vast network of data centres to collect speech recognition data and handle processing, Google’s voice search was able to learn rapidly and compile over 230 billion words. The 2010s saw competitors such as Apple’s Siri and Amazon’s Alexa launching, but Google remains the company to claim the lowest word error rate in speech recognition at just 4.9%.

But recognising speech isn’t the same thing as controlling technology with it. The recognition element only needs to be good enough to detect certain inputs - it’s then the product designer and technical team’s jobs to define how these inputs fuel interaction.

What is a VUI

This interface between you and the software is called the VUI (Voice User Interface). Voice commands are still a developing idea - there are new implementations occurring regularly. Capital One, for example, was the first bank to introduce a balance check skill to Alexa - which meant users could ask for their bank balance to be read to them.

The VUI is the system through which a user will interact with your voice-enabled software. For example, if a traditional interface is a series of buttons and screens, a VUI could be a series of clarifying questions that the software asks as you pursue a query.

Voice control issues

Unfortunately, there’s a clear issue with designing for voice input: errors. Voice commands can struggle to interpret certain tones, accents and even genders (female voices have 13% lower recognition than male ones). How can you help build user trust with a system prone to error?

In a PwC study, the main limitations around voice control were reported to be:

  • Lack of perceived capabilities: people are often unaware of what voice control can do - and they are less likely to experiment to try new things because of…

  • Lack of trust: because voice input often causes errors, consumers said they couldn’t trust their assistants to do basic tasks - which further limited their desire to try new features or capabilities.

  • Integration issues: consumers hesitated around spending on devices they didn’t know would properly integrate with their existing technology.

Despite this, for ‘low perceived value’ tasks like searching for something, checking the weather, setting alarms etc, consumers reported that voice assistants were preferable to other forms of input due to their speed and ease of use.

For higher-value tasks like banking and shopping, the lack of trust in voice assistants limited people’s desire to use them.

How a UX designer can build VUIs that engage

We’ve talked a lot recently about empathy in design thinking. This practice involves understanding users on an emotional level - which applies just as strongly to voice as it does other forms of interaction.

Bringing an empathy-driven UX designer to a VUI may feel counterintuitive since you often don’t need many graphical components, but the theory behind the design and approach can help make voice input more effective for the user.

In general, VUIs need to be simple and conversational. Like in traditional design, where the hierarchy of information prioritises importance at the ‘top of the page’, VUIs must prioritise importance at earlier conversations.

It’s also a good idea to build confirmations into your VUI, telling the user explicitly when your VUI has accomplished the required task.

Finally, anticipate errors and build some form of clarification and backup solution for when things go wrong.

Further elements that the UX design process can bring to the project include:

User research

Research is critical in UX - especially direct user research that can establish who users are, what their needs look like and how their emotions can impact their decisions.

Research can help you understand what role voice plays in your customers’ lives, what they want from future voice control options and even whether you need to integrate voice control at all.

For some users, designing a VUI might be a useless and costly task that has no real market value - but you need to do research to establish that.

Requirement gathering and project mapping

UX design teams are used to working on features that are based on user requirements. Following research, requirements can be crafted as user stories.

These stories utilise the formula “As an X, I want to do X so that I can X” - so, in a VUI project, it may be “as a driver, I want to use a voice assistant to update me on traffic so that I don’t have to look at 3my screen.”

Understanding intent

UX designers are driven by intent-based designs.

They know that a user is trying to accomplish a goal, and they must design to help them accomplish it. With voice search, intent becomes a little harder to interpret as it involves natural language, accents, colloquialisms etc. Where one user might give a command to: “Show me the time in Bangkok”, another might simply state “Bangkok time” and expect a similar answer.

It’s the VUI designer's job to anticipate this and build interfaces that can help clarify intent.

For example, for clear queries, the system can simply deliver the correct result, but for harder to interpret ones you might design a follow-up question/clarification stage.

Designing for conversations

Most VUIs we use every day now operate on a more conversational level. Google’s Assistant would ask you if you meant a certain word or tell you if it didn’t understand. However, it also displays text on the screen to help you clarify which side was incorrect.

VUI designers plan things out via conversation flowcharts, mapping dialogue as a graphical element. Imagine ‘Are you sure you want me to do this?’ as a ‘Confirm’ button and you’re in the right frame of mind.

Tracking results

Following a prototype and then launch, you can use tracking data to see which queries users are actually using. This can be used to refine your VUI further and to make it more accessible, delivering an experience more in line with how the user is actually talking to it rather than how your research suggested it might.

 

At KOMODO, we’re always at the forefront of innovation in tech. Of course, VUI is still a new field, so we’re not claiming to be experts - but our development process includes the user research and empathetic design thinking required to make intuitive digital products that fulfil rather than frustrate. Get in touch to see how we can help with a user experience project now.

Got an idea? Let us know.

Discover how Komodo Digital can turn your concept into reality. Contact us today to explore the possibilities and unleash the potential of your idea.

Sign up to our newsletter

Be the first to hear about our events, industry insights and what’s going on at Komodo. We promise we’ll respect your inbox and only send you stuff we’d actually read ourselves.