Challenges we ran into

Our biggest challenge was integrating multiple AI and audio-processing technologies into a single workflow. We had to connect:

  • WhisperX for transcription and word-level alignment
  • Librosa for pitch, volume, and speaking-rate analysis
  • Llama 3.1 via Ollama for qualitative content evaluation
  • MongoDB for user accounts and leaderboard data
  • ElevenLabs for speech generation

Since many of these technologies were completely new to us, we spent a lot of time learning their APIs, handling compatibility issues, and making sure they worked together smoothly. Building a responsive full-stack application while coordinating all these moving parts was one of the most rewarding challenges of the project.

Accomplishments that we're proud of

We're proud that we built a complete AI-powered presentation coach from scratch.

  • Successfully combined speech transcription, audio analysis, local AI, cloud services, and data visualization into a single platform
  • Implemented privacy-friendly content analysis using Llama 3.1 running locally through Ollama
  • Built a full-stack application with authentication, data storage, and leaderboard functionality
  • Created actionable feedback that combines objective metrics with AI-generated coaching

Most importantly, we transformed a collection of independent tools and APIs into a cohesive experience that genuinely helps people become better speakers.

What we learned

This project pushed us far outside our comfort zones and exposed us to technologies we had never used before.

  • Building a full-stack application with Flask
  • Designing and interacting with a MongoDB database
  • Integrating external APIs such as ElevenLabs
  • Running and utilizing local LLMs through Ollama
  • Processing and analyzing audio data programmatically
  • Structuring backend APIs that connect AI models, databases, and frontend interfaces

By the end of the project, we had a much deeper understanding of how modern AI applications are built from end to end.

What's next for Pitch Prepper

We're excited to continue expanding Pitch Prepper beyond speech analysis.

  • Video integration to analyze eye contact, posture, facial expressions, and hand gestures alongside vocal delivery
  • Enhanced leaderboards with seasonal rankings, categories, and performance trends over time
  • Friend groups and private competitions so users can challenge classmates, teammates, or coworkers
  • Personalized progress tracking to help users visualize improvement across multiple presentations
  • More advanced coaching insights tailored to specific presentation styles such as pitches, debates, interviews, and classroom presentations

Our goal is to make Pitch Prepper the all-in-one platform for developing communication and presentation skills.

Built With

Share this project:

Updates