Speech Assist

Inspiration

Many people fear public speaking. But if they had someone to talk to before their speech, one or a group of audience that could check his/her speech and give advices, they might fare better off in a public speaking situation. This is what motivated us to build this app- a speech and debate practice app that allows Big Brother and Big Sister members to help out each other's public speaking fear, by pairing them up and let them practice their speech and their debate skills with each other

What it does

What makes the app unique is that we have a match making system which pairs people up into real time video/ audio chat, in which they can practice speech and debate with each other, and also that we have an AI, trained to catch common stutter words such as "uh" "ah" "and" so that even when offline, users can practice speech and debate using the AI.

How I built it

We built the AI with python and CMUSphinx we built the match making backend and frontend with node.js , webRTC, and websocket. The match making is built with websocket and the video/audio chat is built with webRTC.

Challenges I ran into

Researching webRTC, and all the technology stacks used for a peer to peer real time video/audio chat. Researching to build an AI that can catch stutter phrases. We settled for CMUSphinx eventually.

What's next for Speech Assistant

We will subdivide it into more categories, such as political debates, speech practice, impromptus, etc. And students will be leveled according to their skill, and they will be matched with people of their own level.

Built With

• javascript