Public speaking can be terrifying. Before every TechCrunch Disrupt, moderators are required to go through mandatory speaker training to ensure we’re not flaming disasters on stage. Our bosses bring in a real-life human to train us and point out how often we use filler words or simply don’t make any sense at all.
SpeechCoach.ai, a project that came out of Disrupt SF 2017’s hackathon, takes a different approach by removing the need for a human-being, and instead using an artificial intelligence-powered platform. It listens to you speak, compares your speeches to pros like President Barack Obama and then gives you immediate feedback.
“Public speaking is one of the biggest challenges for engineers,” Chirag Mahapatra, co-creator of the project and engineer at Airbnb, said. “Most of us don’t have access to speech coaches or people to help us get better at public speaking.”
SpeechCoach.Ai takes into account your speaking pace, the number of filler words you used, sentiment analysis and grammatical mistakes. It then compares your speaking style to Obama, William Taft and Nelson Mandela.
“We kind of have aways wanted this but never had the time to do it,” Mahapatra told me. “This was a great platform to do it.”
In order to build this, Mahapatra and Karthika Purushothaman used the IBM Bluemix API to convert the audio input into text for analysis. Down the road, the team plans to integrate video in order to detect facial expressions. You can check out a demo of the app below.