🌟 About the Project: Debate AI Coach

🎯 Inspiration

As active learners and tech enthusiasts, we noticed that the art of debating is an incredibly valuable skill—helping students improve their critical thinking, logical reasoning, and public speaking. However, debate training often lacks personalized, real-time feedback outside of formal sessions or tournaments.
We thought: “What if an AI could act as your personal debate coach, offering instant feedback on your delivery, clarity, and argument structure?”
Thus, Debate AI Coach was born—a virtual mentor for debaters to practice anytime, anywhere.

🔨 What We Built

• A simple web app using Streamlit that lets users record their debate speeches.
• The system uses speech-to-text processing (via OpenAI Whisper API) to convert audio into text.
• Natural Language Processing (NLP) algorithms (spaCy + custom GPT model prompts) analyze the logical flow, clarity, and filler words.
• Basic speech analysis tools check:
  • Speech Speed (WPM)
  • Confidence Level (tone clarity)
  • Number of filler words like "um", "uh", "like".
• The app returns an easy-to-understand feedback report to help the debater improve their performance.

📚 What We Learned

• How to integrate speech processing APIs (like Whisper) with NLP models.
• Using Streamlit to quickly prototype AI tools.
• How to analyze human speech characteristics (speed, tone, clarity) computationally.
• Challenges in quantifying "good reasoning" in debates using machine models.
• Balancing technical complexity with user simplicity in real-time feedback tools.

🚧 Challenges We Faced

• Speech-to-text accuracy: Ensuring that technical debate language or unusual pronunciations are transcribed correctly.
• Designing simple yet effective logic analysis prompts for GPT models.
• Handling noisy or low-quality audio recordings.
• Making sure feedback is clear, actionable, and not overwhelming for the user.
• Managing API usage limits for Whisper and OpenAI within a free-tier budget.

🚀 What's Next?

• Adding real-time visual speech indicators (confidence meters, filler alerts).
• Gamifying debate practice (levels, rewards).
• Extending to multi-user debate simulations.
• Building an Android/iOS mobile app version.

Built With

• github
• natural-language-processing
• python
• sarvam
• spacy
• streamlit