ArisTalkle

AI-powered debates. Smarter discourse. Stronger voices.

Comment

Inspiration

In today's digital world, meaningful discourse is often lost in text-based communication, leading to misinterpretation and polarization. Inspired by Aristotle’s belief in the power of thoughtful debate, we wanted to create a platform that fosters structured, engaging, and AI-driven discussions. Our goal was to help people refine their argumentation skills while making debates more accessible and interactive to make everyone more confident in their speaking abilities.

What it does

ArisTalkle is an AI-powered debate platform that enables users to engage in video debates with a human-like AI opponent. You argue, and Aristotle responds! It provides:

  • Personalized scoring and feedback to help users improve.
  • A practice mode for refining presentation and argumentation skills.
  • A personal dashboard to track past performance, scores, and areas for improvement.
    By combining AI-driven interaction with structured discourse, ArisTalkle enhances critical thinking and communication skills.

How we built it

  • AI Model:

    • We use the Gemini API to analyze the user's video file and generate a text-based argument.
    • The generated text is then passed to Cartesia API, which converts it into realistic speech.
    • The Sync AI model then synchronizes the generated audio with a virtual AI video speaker, creating a lifelike debate experience.

  • Backend:

    • Interacted with MongoDB using Next.js Server Actions to add and manipulate data in the database.
    • Built AI server using Python and Flask to manage AI model interactions and user data.
    • Stores debate history, scores, and personalized feedback.
    • Supabase was used to host .wav files, making them publicly accessible for Sync AI.

  • Frontend:

    • Developed using Next.js for a fast, server-rendered experience.
    • Styled with Tailwind CSS for a modern, responsive UI.
    • Utilized shadcn/UI for an accessible and aesthetic component library.
    • Handles video uploads, real-time feedback, and debate tracking seamlessly.

  • Database:

    • We used MongoDB to manage user accounts, track improvement metrics, and store debate performance.

Challenges we ran into

  1. Getting the Gemini API to work consistently

    • Initially, we faced issues with response delays and inconsistencies in AI-generated arguments.
    • We optimized API calls and implemented error-handling mechanisms to ensure smooth performance.

  2. Generating a publicly accessible .wav file for Sync AI

    • Sync AI required public URLs for the AI-generated speech files, but the output was locally stored by default.
    • We solved this by integrating Supabase to host and serve .wav files via public links.

  3. Integration between back-end and front-end

    • Since we were dealing with multiple data types (PDF, video, audio, text), ensuring smooth communication between the AI models, front-end UI, and Flask server was complex.
    • We developed a structured API endpoint system to handle the flow of data across the different components.

Accomplishments that we're proud of

  • Successfully integrated multi-step AI processing to create a realistic debate experience.
  • Developed real-time speech analysis and feedback to help users improve argumentation.
  • Solved the file hosting challenge with Supabase, enabling a smooth AI-driven debate process.
  • Built a modern, user-friendly UI with Next.js, Tailwind CSS, and shadcn/UI for an intuitive debate experience.
  • We were able to make a solution that analyzes a user's video logically and outputs a video in about the same time as the video generation AI Sora (sometimes even faster!).

What we learned

  • How to effectively integrate multiple AI models (Gemini API, Cartesia API, Sync AI) into a single pipeline.
  • The importance of reliable file hosting solutions for AI-generated content.
  • Best practices for back-end and front-end integration when handling multimedia data (video, audio, text, PDF).
  • How to fine-tune AI-generated responses to create structured and engaging debates.
  • Leveraging Next.js for server-rendered performance and shadcn/UI for accessible UI components.

What's next for Aristalkle

  • Expanding AI capabilities to support multi-person debates and different argumentation styles.
  • Integrating multilingual support to make debates accessible worldwide.
  • Enhancing feedback systems with deeper insights into argument strength.
  • Introducing debate tournaments where users can compete and track their progress.
  • Building partnerships with schools and workplaces to promote structured discourse and critical thinking.

Aristalkle is just the beginning—our vision is to create a world where meaningful debate is accessible to everyone! 🚀

Built With

Share this project:

Updates