ExamGPT

Conducts oral exam based on the topic provided by the User with the help of the GPT model.

Comment

Inspiration

The inspiration for the project came from the theme of the hackathon itself. Aimed towards "Protecting the Youth", the idea was to build a tool that would help students in their learning process. Since exams are a major part of a student's life, the aim was to help and make this process easier with the help of AI models. With ExamGPT, students can practice their answers or check their knowledge regarding a certain topic. It can be seen as a tool for mock oral examinations which will allow students to enhance their knowledge.

What it does

ExamGPT is a powerful tool that simplifies the process of conducting oral exams. It leverages the GPT-3.5 model from OpenAI to act as an interviewer, asking questions based on a provided topic. This project combines speech recognition, text-to-speech, and PDF handling to create a seamless oral exam experience.

Features

  • Topic-Based Questions: Input a PDF document on the desired topic, and ExamGPT will generate relevant questions for the oral exam.
  • Transcription: As the AI interviewer asks questions, the transcription of the speech will be displayed on the screen.
  • Recording Responses: Easily record your responses by clicking the microphone icon. When you're done, stop recording.
  • Text-to-Speech: ExamGPT uses the speechSynthesis library to convert text into speech for a natural conversational experience.
  • Speech-to-Text: The project includes react-speech-recognition library to transcribe spoken responses.

How I Built It

I used a combination of different technologies and packages to build the ExamGPT.

Frontend:

  • The front end was implemented using React, a popular JavaScript library known for its versatility and performance.
  • Packages like React Speech Recognition and the speechSynthesis API enabled real-time speech-to-text and text-to-speech capabilities.
  • The user interface was designed with a focus on simplicity and accessibility to ensure an intuitive experience for all users.

Backend:

  • For the Backend, I used Node.js, providing a scalable and efficient server environment.
  • The pdfjs-dist library was used to handle PDF documents, allowing extraction of text for question generation.
  • The core of the product is the response that is being generated using the OpenAI GPT 3.5 model.

Challenges I ran into

There were a lot of challenges that I ran into. From extracting text from pdf to ensuring correct response from GPT mode. However, the major challenge that integrating text-to-speech and speech-to-text functionalities. Ensuring correct transcription of text was necessary for the success of the project.

Accomplishments that I am proud of

I am happy and proud that I was able to complete this project on time. My initial plan was to participate with a couple of my friends which can't happen due to some reasons. At the end, I am happy to be able to build a working AI tool that acts as an examiner and can help students going forward.

What I learned

During this project, I learned and implemented many new things that I hadn't done before. I learned to extract useful information from pdf provided by the user. Along with this, I learned to implement speech-to-text and text-to-speech functionalities for the first time. Also, working with the OpenAI GPT model was a great experience, and overall, I can say that working on this project has been great for me in terms of learning.

What's next for ExamGPT

There are a couple of things that I would like to improve in ExamGPT. Starting from the front end, I would like to further enhance the user interface and improve the design and make use of loading and animations. Along with it, I would like to provide a way for the user to download their conversation for future reference. Maybe I will look to connect it to the database to store the chats which will allow users to save their conversation.

Built With

Share this project:

Updates