Inspiration
So much of our everyday communication happens through texts and messages. We wanted to create more opportunities to practice speaking out loud, especially when we have to respond without time to prepare. Speak Up grew from that idea: make speaking practice something people want to come back to.
What it does
Speak Up is a gamified speaking trainer built around debate-style impromptu speeches. Users choose a prompt and a side, then make their case out loud. After each speech, they receive a score and targeted feedback to help them improve their next attempt.
How we built it
We used AI agents to help build the interface and gave the platform a distinct visual style. For speech grading, we developed a deterministic scoring algorithm with a rubric informed by published research. AI helps analyze the speech, while code calculates the scores.
Challenges we ran into
Our biggest challenge was implementing the backend for community reactions. Connecting that feature to the rest of the platform took more work than we expected.
Accomplishments that we're proud of
We're proud of building a platform that gives users concrete feedback they can apply to their next speech. Turning a recording into specific suggestions made our original idea feel useful.
What we learned
We learned how to guide AI agents toward an interface with personality and a consistent design. We also learned how game mechanics can make a task like speaking practice more engaging and give users a reason to try again.
What's next for Speak Up
We want to add computer vision to analyze hand gestures and posture during speeches. Using research on nonverbal communication, we hope to offer feedback that helps users speak with greater confidence and impact.
Built With
- css
- dockerfile
- html
- javascript
- plpgsql
- python
- typescript
Log in or sign up for Devpost to join the conversation.