de(talks) | Devpost

Inspiration

I noticed that in many work environments, such as the office or within a school, there are cases where it can become toxic because of other people. These kinds of environments make it so people begin to hate their jobs or school, which reduces productivity, mental health, and overall low group morale. It is also an issue that isn’t talked about as often as other issues but it would greatly benefit everyone if people simply began feeling more comfortable in these kinds of environments.

What it does

I developed a website called de(talks) which is like a joke for detoxification, that basically detects speech when a conversation begins and then analyzes it for any negative or hurtful comments. It also features a live graph that saves toxic sentences and shows them on a scale from 1-10. Over time, it develops a sort of trend and you can begin to see how a conversation can start to escalate.

So one of the solutions to toxicity on this app is to have a sound play that reminds people they should talk deep breaths when things get too heated. Additionally, if this technology is used within an environment where there exists a supervisor such as a boss or a teacher, a phone number can be added to it and an sms message regarding the toxic phrase can be sent to them so they can deal with it personally.

How we built it

I built the website suing HTML, CSS, and javascript. For the functionality of the toxicity points, I made my own filter and level determiner using just javascript. I used twilio for the SMS messaging feature as well as openai for a surprise easter egg within the code. Finally, I used chart.js to display the live graph for my data.

Challenges we ran into

It was my first time using a lot of these technologies such as Twilio. It was difficult especially as it required me to make a major change to my code near the end of the hackathon.

Accomplishments that we're proud of

Creating a filter was quite difficult as well as getting it to display live on a graph. I was also proud of the fact that I managed to make such a minimalist website all on my own.

What we learned

I learned how to use Twilio as well as create and run a server. I also learned how to make openai API calls as well as learn about the outrageous number of models they have.

What's next for de(talks)

Hopefully I will be able to expand and develop the filter functionality as now the methods used was quite barbaric and messy. I hope to be able to incorporate another API like the one from perspective into my code.