StageMind

Analyzing Performance
Homepage
Generating the Scenario
Confirming the Scenario
Feedback
Zoom Meeting

Inspiration

Our journey began with a simple goal: to help students learn in a meaningful, impactful way. Initially, we considered teaching a "dumb" AI, but our vision quickly evolved into something more dynamic, creating a platform where users can engage in real-time discussions with intelligent AI agents. This shift was driven by the understanding that communication is a powerful skill—one that plays a critical role in job interviews, negotiations, and everyday interactions. By learning from AI agents, users can practice and improve their communication skills, building confidence and expertise in a safe, supportive environment.

What it does

StageMind is an interactive platform that helps you boost your communication skills by having real conversations with multiple AI agents. Whether you're practicing for a job interview, a presentation, or just improving your speaking confidence, you choose the topic or scenario and StageMind guides you through it.

How we built it

StageMind uses Next.js and Python to create real-time, AI-powered conversations.

The user speaks about a chosen topic—captured through Speech-to-Text (STT) with OpenAI API.
An LLM generates smart replies from AI agents with OpenAI API.
Responses are converted back to speech via Text-to-Speech (TTS) for a natural flow with OpenAI API.
The user's answers are analyzed for sentiment (e.g., confidence, excitement) using HuggingFace wav2vec2 emotion-msp-dim model.
An evaluation model scores the user’s performance, highlighting strengths and areas to improve with Perplexity Sonar.
Each AI agent offers personalized feedback and tips for better communication with Perplexity Sonar.

Challenges we ran into

One major challenge was implementing realistic and responsive AI agents, and finding the right services to power them. We also faced difficulties in connecting the backend to the frontend smoothly, ensuring the conversation flow, feedback, and visuals were all displayed correctly in real time.

Accomplishments that we're proud of

We successfully built multi-agent, real-time conversational AI that responds intelligently and naturally to user input. We also integrated a sentiment analysis and evaluation system that provides meaningful, personalized feedback, helping users understand their communication style and improve with every session.

What we learned

Through this project, we gained valuable insights into building real-time AI interactions and the complexities of integrating multiple systems, from speech recognition to emotion analysis and feedback generation. We also learned the importance of seamless frontend-backend connectivity to ensure smooth, user friendly experiences and the challenges of fine-tuning AI agents to respond naturally and effectively.

What's next for StageMind

Looking ahead, we plan to enhance AI agent capabilities, making them even more adaptive and context-aware to provide richer, more personalized conversations and more detailed, actionable insights to help users improve specific communication skills. We’ll also focus on scaling our platform and introduce subscription models to provide users with premium features, including advanced analytics, personalized coaching, and exclusive content