Funetic
Team: Sanjana Melkote, Shreya Puli, Amrita Basin, Kevin Yim, Isabelle Marsh
Discipline: User Research, User Experience (UX), Interaction Design
Duration: Jan ‘21 - Aug ‘21
Tools: Figma, Sketch, Adobe Illustrator
Methods: 6-3-5 Brainstorming, Value Opportunity Analysis (VOA), Competitor Analysis,Weighted Matrix, Empathy Map, Reiteration
Overview:
This project was the sole focus of UC Berkeley’s Jacobs Institute for Design Innovation course “Design Methodology”. This course combines practices from design, business, and engineering to expose students to a multitude of design methods. Not only did we learn a wide breadth of tools, we also applied these to our own project tackling an unmet need.
Since most of my group had family members who struggled to use voice AI, myself included, we decided to address this problem with a new app that allowed users to “teach” the AI how to pronounce non-Western names.
My Role:
Conducted 5 user interviews
Created the Value Opportunity Analysis (VOA)
Conducted in-depth competitor analysis
Identified product requirements that expand accessibility
Led and organized meetings
Prototyped the final wireframes
Reiterated design based on user feedback
How Existing Solutions Fall Short
Current AI voice recognition struggles to understand and pronounce non-Western names.
Although Apple’s Siri currently provides the user with 3 options to select which name pronunciation is most accurate, the user cannot provide direct feedback on how to correctly say a name.
Google Home and Alexa have the same problem.
The Challenge…
How might we make speech recognition more inclusive and representative?
User Research Insights
After conducting interviews with stakeholders, our team was able to identify where existing solutions fall short.
Siri failed to understand names that are not spelled according to English phonetics.
Users interact better with AI that has a screen
Interviewees mispronounce names so that Siri could understand the request.
Empathy Map for Data Synthesis
By identifying larger trends across our user research, our final product will be linguistically receptive to different accents, dialects, and languages. It will also have a strong sense of confidence, reliability, and social impact.
Product Requirements
Value Opportunity Analysis
Concept Generation
Our team used 6-3-5 Brainstorming method to generate 75 original concepts that fulfill our product requirements.
We then used dot voting and a weighted matrix to filter and narrow down ideas.
Prototyping
Three rough and rapid prototypes that I generated of our final concepts.
Mid- Fidelity Wireframes
Funetic achieves 3 essential goals -
Create voice-to-text links for users
Compile links for personal contacts and public dictionary
Share contact cards with other users
Future Projections
Our team plans on gathering more user feedback on our final product to improve user experience.
Our hope is that our public database of users’ pronunciations will not only help improve our app, but also serve as a foundation for existing AI products that lack diverse testing. Non-western users have been long-neglected in the technological industry and our team’s mission is to make voice recognition software more equitable for and representative of all users.
Reflection
Working on this project was a rewarding experience that allowed me to apply new design methods to an ongoing interest of mine: social impact. It was inspiring to visualize the importance that this project would have on the lives of underrepresented groups, had it been brought to fruition. And to work with such talented team members helped fuel the imagination and critical thinking that dictated many of our creative choices. This project has directed me towards the type of designer I want to be, one that aims to empower others.
