Valfee: Decreasing Public Speaking Anxiety with AI-Assisted Observational Learning

Public Speaking
Date Completed

Public speaking anxiety prevents students from achieving their potential both professionally and personally. Current methods to reduce anxiety, both in the classroom and online, are not scalable and are not capable of personalizing feedback. Valfee, an app that uses artificial intelligence to help students learn how to mimic the speech patterns of role models, has the potential to significantly reduce public speaking anxiety at scale. Students open the app and choose a topic they would like to speak to. They then select a role model to emulate from avatars representing different fields. Next, they are given a question about their selected topic and asked to respond. Valfee uses three metrics— pace, frequency of pauses, and number of filler words—to compare the student to their role model. Students receive personalized feedback on how to modify their speech to better mirror that role model.

The goal of this research is to test if the use of Valfee over one semester [i.e., 4 months] will significantly decrease public speaking anxiety as measured by the Personal Report of Public Speaking Anxiety (PRSPA). PRSPA is a validated and reliable instrument created by James McCroskey, a professor at the University of West Virginia, in his work on public speaking anxiety. It consists of 34 items (e.g., “My thoughts become confused and jumbled with I’m giving a speech”), which students are asked to rate on a 1 (Strongly Disagree) to 5 (Strong Agree) scale. 

Lori Breslow, Nacho Nwana
Lab Name
Sloan School of Management