Describe your project

Seekho - Gen-AI Powered NPTEL Course Creation Platform

Seekho is a generative AI platform designed to streamline course creation for educators. It automates key tasks like generating course layouts, notes, quizzes, and concept animations, while offering AI-driven doubt-solving for students. Seekho allows professors to focus on making video-based content, automating the rest with tools like content translation and multiple difficulty-level note generation.

In-Scope

1. Course Layout Generator: Automatically creates course structures based on input.
2. Notes Generator: Provides notes for different proficiency levels (beginner, undergraduate, expert).
3. Quiz Generator: Generates weekly quizzes to track student progress.
4. Animated Concept Explanation: Uses doodle animations to explain complex topics for various learning levels.
5. Socratic Doubt Solver: Encourages critical thinking by resolving doubts through a Q&A approach.
6. Content Translation: Translates videos and notes into multiple languages.
7. Voiceover Support: Provides multilingual voiceovers to enhance accessibility.

Out-of-Scope

• Real-time human-led instruction beyond the AI-powered features.
• Video content creation, which remains the educator's responsibility.
• AI-driven adaptive learning and grading systems.
• Infrastructure for video recording or editing.

Future Opportunities

Seekho has potential for further growth, including the addition of AI-driven adaptive learning based on student progress and enhanced analytics for educators. Improvements in language synchronization for videos will further broaden accessibility. Collaborations with universities could scale Seekho to a nation-wise platform, making it a comprehensive tool for online education. It can also leverage the current NPTEL platform based course creation to a much more interactive and time-efficient tool.

Challenges we ran into

One significant challenge we encountered while developing Seekho was achieving precise audio synchronization for videos translated into multiple languages. The issue arose when the audio from the voiceovers did not align correctly with the lip movements of the speakers in the original videos. This desynchronization negatively impacted the user experience and the effectiveness of the educational content.

To tackle this challenge, we implemented Silero Voice Activity Detection (VAD). This technology allowed us to detect and segment audio into manageable chunks based on speech activity. By breaking the audio into smaller pieces, we were able to analyze the timing and pacing of each segment individually.

After segmenting the audio, we utilized an API to adjust the pace of these audio chunks. This adjustment ensured that the speed of the voiceovers matched the visual cues in the video, resulting in a more cohesive and synchronized viewing experience. We tested this solution across various languages to ensure that the synchronization was consistent, regardless of linguistic differences.

While this approach significantly improved audio alignment, it remains an area of ongoing refinement. We are continually exploring ways to enhance the accuracy of our synchronization methods, ensuring that Seekho delivers a seamless and engaging learning experience for students across different languages.