SafeSpeech

Menu page
Harmful words are highlighted, advice is given for a different approach.
Pass the scenario
Post scenario critique

Inspiration

I was inspired to create this project because I am a direct victim of gun violence, and I understand how difficult it is to talk about these events, especially with children. Though my experience I noticed that not everyone knew how to discuss these topics in a way that was emotionally sensitive and empathetic.

What it does

SafeSpeech generates realistic scenarios where a child might be exposed to or ask about gun violence, allowing the user to practice responding in a simulated conversation. After the user responds, the system evaluates the reply, highlighting words or phrases that could be harmful or insensitive and offering guidance for more empathetic communication. By providing iterative feedback and practical tips, Safe Speech empowers adults to discuss difficult topics with care, ensuring children feel supported and emotionally safe.

How we built it

We built it using C#, with WinForms as our UI framework.

Challenges we ran into

There are many Gemini calls in our application, so managing the asynchronous flow was a large challenge. We also had trouble with UI state relying on those async functions, but ultimately we were able to iron out the process and created a smooth experience.

Accomplishments that we're proud of

We are proud of creating a tool that provides real world benefit for children, parents, and teachers. We are also proud of how quickly we were able to develop features with minimal bugs.

What we learned

Separation of responsibilities is critical for rapid development and short term scaling. It helped us to separately develop features and easily link them together when ready.

Game changing AI

Gemini is used to generate realistic scenarios, highlight harmful language, determine if the conversation was productive and successful, and provide feedback to the user. ElevenLabs is used for speech-to-text functionality so the user can naturally talk during the scenario.

AI Usage: ChatGPT was used to generate code bits for JSON parsing, audio reading, and creating python processes.