Meeting Genius

Inspiration

The inspiration for Meeting Insights Generator stemmed from the need to simplify and enhance the way teams process and analyze meeting data. Many professionals struggle with managing lengthy meeting recordings, identifying critical takeaways, and ensuring insights are effectively shared. By combining AI-driven transcription, summarization, and visualization, we aimed to create a tool that transforms meeting workflows into a seamless, efficient, and engaging experience.

What it does

Meeting Insights Generator is a multimodal application that:

• Converts meeting audio into accurate text transcriptions.
  
• Summarizes discussions into concise, actionable insights.
  
• Generates relevant visuals for key highlights to enhance comprehension.
  
• Provides audio summaries for quick reviews, catering to different user preferences.
  
• Securely stores all content, ensuring easy retrieval and sharing.

How we built it

We used cutting-edge technologies to develop this project:

• Frontend: React and Vite for a responsive, user-friendly interface styled with Tailwind CSS.
  
• Backend: Node.js and Express for robust server-side functionality, connected to MongoDB for efficient data storage.
  
• APIs:
  
  • AssemblyAI: For speech-to-text transcription.
    
  • Hugging Face (BART): For generating summaries from transcriptions.
    
  • Pexels: For retrieving images that visually represent meeting highlights.
    
  • Deepgram: For converting summaries into audio.
    
• Supabase: Integrated as a storage solution for generated content, offering scalability and accessibility.

Challenges we ran into

• API Integration: Handling asynchronous calls and ensuring smooth interaction between multiple APIs posed challenges, particularly with managing rate limits and response formats.
  
• Accuracy: Fine-tuning transcription and summarization models to achieve precise and contextually accurate outputs.
  
• Image Relevance: Generating visuals that align well with the summarized content required testing and optimization.
  
• Time Constraints: Balancing development and debugging within the hackathon's tight deadline was a significant challenge.

Accomplishments that we're proud of

• Successfully developed a fully functional, multimodal application that blends text, audio, and visual modalities.
  
• Achieved seamless API integrations, ensuring accurate outputs and user-friendly performance.
  
• Created a scalable architecture that can handle complex workflows without compromising efficiency.
  
• Delivered a visually appealing and intuitive user interface, ensuring accessibility for a broad range of users.

What we learned

• Multimodal AI Applications: Gained a deeper understanding of combining AI technologies across different data modalities (speech, text, and visuals).
  
• Team Collaboration: Improved our ability to work as a cohesive team under tight deadlines, focusing on shared goals.
  
• API Optimization: Learned best practices for optimizing API usage, including error handling and response management.
  
• User-Centric Design: Reinforced the importance of designing with end-users in mind, prioritizing simplicity and functionality.

What's next for Meeting Insights Generator

• Real-Time Transcription: Expanding the tool to support live meeting transcription and summarization.
  
• Language Support: Adding multilingual transcription and summarization capabilities to cater to diverse teams.
  
• Integration with Productivity Tools: Connecting with platforms like Slack, Microsoft Teams, and Trello for seamless workflows.
  
• Advanced Analytics: Introducing features like sentiment analysis and action item detection for deeper meeting insights.
  
• Enhanced Visualizations: Leveraging advanced AI models to generate more dynamic and personalized visual content.
  

Built With

• assemblyaiapi
• deepgramapi
• express.js
• huggingfaceapi
• javascript
• mongodb
• node.js
• pexelsapi
• react
• supabase