Inspiration
The idea for equiLearn came to us while browsing YouTube, where we stumbled upon a Django tutorial in Hindi. Even though we didn’t fully understand the language, we realized the content was valuable but difficult to follow due to the language barrier. This made us think about how much important knowledge is inaccessible to people worldwide because of language differences.
That’s when the concept for equiLearn emerged—a platform that would make learning from any source, in any language, both easy and accessible. Our goal was to create a solution that could translate educational content in real-time. With equiLearn, users could input various learning resources, get summarized notes like Quizlet flashcards, and take interactive quizzes. It would also feature an AI tutor to guide users through challenging topics, ensuring a seamless learning experience for everyone.
At its core, equiLearn stands for the belief that knowledge should be available to all, no matter where they are or what language they speak. By harnessing technology, it empowers learners globally by turning educational content into a universal resource.
What it does
- Accepts inputs like video URLs, PDFs, .mp4, and .mp3 files.
- Offers innovative AI dubbing, providing language translations and voiceovers (a unique feature).
- Generates comprehensive notes based on the content.
- Creates flashcards to help users retain information.
- Features an AI tutor, powered by ChatGPT, that creates quizzes, answers questions, and provides feedback using the supplied content.
- Generates interactive quizzes to reinforce learning.
How we built it
We built the frontend with React to ensure a clean and user-friendly experience, while the backend processes were handled with Flask and Python. YouTube videos are downloaded using the YouTube API, and audio transcription is done through Assembly AI. The text is translated using Google Translate, and dubbed audio is created using Microsoft Azure Text-to-Speech (TTS). We synchronized the dubbed audio with the original video using moviePy.
This process was also applied to podcasts. From the transcripts, we generated notes using OpenAI GPT-3.5 Turbo API and created flashcards in JSON format. Additionally, we developed a GPT-based AI tutor to provide quizzes and answers based on the transcript. For PDFs, we utilized Convert API to extract text, which followed the same process as videos.
Challenges we ran into
- Properly formatting flashcards into JSON structure.
- Effectively syncing audio with video using moviePy.
- Integrating the React frontend with the Flask backend.
- Finding cost-effective APIs, like Microsoft Azure TTS, to manage our resources efficiently.
- Managing English-to-English video processing while handling storage issues.
- Planning for scalability, such as efficiently storing user-generated videos.
Accomplishments that we're proud of
- AI Dubbing Innovation: Created a pioneering feature that translates and dubs content in various languages.
- Comprehensive Learning Platform: Successfully integrated notes, flashcards, and an AI tutor to create a holistic educational experience.
- Smooth Backend-Frontend Integration: Effectively connected React with Flask and Python to ensure the platform runs seamlessly.
- Effective API Use: Skillfully leveraged APIs like Assembly AI and Microsoft Azure TTS for high-quality, low-cost functionality.
- Efficient Workflows: Streamlined the transcription, translation, and content generation processes.
- User-Centered Design: Developed a clean, accessible interface for a smooth user experience.
- Problem-Solving: Overcame technical challenges, such as JSON formatting and API integration, with persistence and creativity.
- Vision for Growth: Laid out a roadmap to improve video processing, data handling, and deploy custom machine learning models.
What we learned
We gained experience integrating advanced technologies and APIs while optimizing workflows to create a platform that solves real-world educational challenges. This project taught us the importance of user-friendly design and how to scale innovative solutions efficiently.
What's next for equiLearn
- Optimized Video Processing: Work on reducing processing times so that a 30-minute video can be transcribed and translated within seconds using custom machine learning models.
- Efficient Data Pipelines: Develop advanced data pipelines to handle multimedia content faster and provide translated materials in near real-time.
- Edge Computing: Explore edge computing options to allow for faster transcription and translation, even in low-bandwidth conditions.
- Streamlined UI: Redesign the user interface to simplify navigation, minimizing steps for users while maximizing efficiency.
- Performance Monitoring: Implement real-time system tracking to improve user feedback and system performance, ensuring consistent, high-quality service.
- Custom Machine Learning Models: Build proprietary models to improve translation accuracy and speed, reducing reliance on external APIs.
- Deployment: Use Docker and Google Cloud (leveraging $300 in free credits) for scalable, cloud-based deployment.
Log in or sign up for Devpost to join the conversation.