TherapyAI | Devpost

TherapyAI logo

🧠 Inspiration

Mental health is something that is constantly overlooked, and a lot of people in today's world do not have resources to get the help they need to maintain positive mental health. TherapyAI is an application powered by Cohere's Natural Language Processing and Autocomplete technologies to listen to the user's concerns and give them feedback based on its knowledge.

？What it does

TherapyAI uses SpeechRecognition package to convert speech to text using google's speech transcriber. After the speech is converted to text it is analyzed by Cohere's Natural Language Processing to detect if the user's mood is positive (happy), negative (sad), or neutral (neutral). After the tone of the speech is detected by the Natural Language Processing it is inputted into a prompt which goes as "bot is an emotion support voice assistant who wants to help the user feel better because user is {feelings}" {feelings} representing the tone detected by the NLP. Using this prompt we use Cohere's autogeneration to generate an appropriate response which is broadcasted by the computer using the "os" package implementing text to speech.

🧱 How we built it

We built this using python, cohere's NLP and Autogeneration technologies, tkinter for the Graphical User Interface, we used SpeechRecognition package for speech to text, and os package for text to speech. By making this project we wanted to give back to the community to give them a partner someone they could talk to about all their worries without fearing being judged or not being able to afford services which should be accessible to anyone. We hope this will application will do good to communities and decrease overall depression and anxiety rates in young teenagers as it is currently a major ongoing issue in today's world.

🐛 Challenges we ran into

Main issue we ran into was when we were trying to manipulate strings we had to do a lot of trial and error which took away a good amount of time from us. We were also greeted with variable used before referenced error often not knowing the exact reason for it so us having to resort to using workarounds. But by far the most challenging part of this project was to think about what we should add to it and how to train the models, mental health is a serious issue and we are not professionals so we had to research for what kind of features we could implement in this type of application which definitely limited our creativity but gave us a guiding path for what was appropriate for a project of this caliber.

🎉 Accomplishments that we're proud of

Our team is still pretty new to python development environment and we are proud that we managed to use python along with other technologies such as cohere in our project. We are also very proud that we were able to implement cohere, at first it looked intimidating setting up cohere's autogeneration but as we started reading the documentation and the helpful examples it became clear for us what to do and we were able to move onto other features

🧮 What we learned

Although we have used Cohere's services in the past we have not used it to the same degree as we have had in this project. In the past we have used NLP but have not trained the model as well as we did this time and we were not aware of the text autogeneration at the time. We plan to keep using the text autogeneration for more projects in the future as we believe it will prove handy in other projects other than TherapyAI

😲 What's next for TherapyAI

We would like to make the speech recognition faster and understand different accents so it can be used by a variety of people who may not even speak english. We would like to change the voice of our voice assistant We would like to convert this application into an executable so other people can use it on their local systems without being connected to the internet