Design for AI Product Project
Instructor: John Zimmerman, Qian Yang
Time Span: 2 Weeks
Contribution: Concept, Research, UI/UX Designer
Team Members: Meng Wan, Ethan Ye, Neha Agarwal, Pinxu Ren
An accent analysis game
Accent.AI is an accent analysis game which motivates the crowd to provide their voice and profile for accent recognition data acquisition and system self-development. I worked as an AI product designer taking care of research and design in the team with other 3 cross-functional teammates.
Design AI computing system based on valuable activities that uses a crowd of people as a proxy
DO YOU HAVE ACCENTS WHEN YOU SPEAKING ENGLISH?
U. S. Census Bureau. we learned that more than adults. Fromin the United States are not native English speakers. Besides ESL, almost everyone has a bit of 35 million regional accent.
Huge Economic and Social Values Exists in Accents Recognition
By Rick Aschmann
IMPROVING PERSONAL ASSISTANT
As Alexa is used in more and more places around the world, the number of accents is growing, too.
Accent data can be used for more accurate speech recognition and to make personal assistants feel more familiar.
Accents are another piece of demographic data about users and potential consumers
Advertisers can better target and relate to consumers with this additional insight about their linguistic background
The system implements interactive machine learning. It would begin with developers to create an MVP and then with enough crowd data and input, which could tune the model to be more accurate on its own.
User learns about Accent.AI on Facebook Messenger and give Accent.AI profile info.
To play with Accent.AI, user reads through two pieces of script to microphone and system will collect user's voice data.
Accent.AI will reflect user their accent makeup with different country tags and percentage.
User could also play with another accent challenge mode to compete with friends who is better on specific accent mimicking.