AI Powered Voice Assistant

The Hisense AI powered TV and Voice Assistant was designed to move beyond the current generation of Voice Assistants and provide a powerful yet intuitive feature set

My Role

AVP of Product Design, Product Owner, Project Lead, Interaction Design Contributor

The Challenge

The Goal

Hisense had a treasure trove of AI technologies that had been developed by research teams without any plan on how to utilize them in existing products. Hisense came to us to see if we could create a product to take advantage of their AI capabilities.

Take advantage of Hisense's AI capabilities to enhance the TV and provide a feature set that would help drive VIDAA sales and maintain Hisense's technology superiority in the China market.

The Process

It became clear very early on that coupling the AI and voice capabilities would provide us a formidable combination. This was further reinforced upon researching the current competition's approaches to TV Voice Assistants. For the most part, other voice assistants seemed to be a tacked on approach that still had much room to improve. Like many new features that were being added to smart TVs, it often felt incomplete and poorly implemented. It was an opportunity to differentiate Hisense's offering. So I proposed a Voice First approach to the design so that the design team could be unburdened from the constraints of a typical hybrid DPAD experience.

Hisense's technology provided us an opportunity to go beyond a typical voice assistant's capabilities. After analyzing all available data points, I created a strategy for the team to apply. It consisted of making decisions based on eight different data types that we could consider when optimizing the user experience. I coined the term "Multi-Dimensional Awareness", and set to work at making it a reality.

By leveraging the massive amount of data available to the platform, we were able to provide an AI powered experience that felt truly next generation. The AI algorithms powering the assistant become more accurate the longer the TV is used. Data is tracked and analyzed from a wide variety of sources and used to provide a seemingly magical experience.

Hisense's Jubao character was chosen by Hisense Management to be used as the Voice Assistant's avatar.

The Design

To minimize the most common voice errors, dynamic results and guidance are provided as the user speaks. This helps increase efficiency and guide the user to the desired results.

Before the user has even begun speaking, the AI assistant is already presenting the results that it anticipates based on all the available data. The more the user speaks the more refined the results become.

My approach of continually refining results as the user speaks allowed us to minimize incorrect results and dead ends.

By taking advantage of the many technologies and services available to us, we were able to provide an AI experience on the TV that was a generation ahead of similar experiences in North America at the time.

The template based design supported a massive amount of queries, and provided Hisense the ability to remotely add new data types at any time without updating the TV.

To further enhance the experience, the UI adapts to the method of control.

When using the remote, a classic DPAD style interface is presented. But when the TV is responding to voice, the UI seamlessly transitions to a Voice First UI that is fully optimized and designed for Voice Control.

The Result

The VIDAA Voice Assistant was included on the VIDAA 6 series TV at the end of 2018 and helped Hisense retain its number one position in China Sales for 2019. It has become an important part of the VIDAA ecosystem and was included in all subsequent VIDAA premier line TV releases. It is currently used by millions of happy customers.