SummitAI – AI that helps you “summit” your meetings with clarity.

AI-Powered Meeting Transcription & Summarization Platform

SummitAI is an end-to-end meeting intelligence system that automatically transcribes, analyzes, and summarizes meetings. It leverages Whisper for speech recognition, Gemini for abstractive summarization, and a FastAPI backend integrated with MongoDB and Tebi for efficient data storage and retrieval.

---

🚀 Features

• 🎧 Automatic Speech Recognition (ASR) using OpenAI Whisper
• 🧠 Summarization powered by Google Gemini for concise, context-aware meeting overviews
• ⚡ Audio Preprocessing with FFmpeg for robust media handling
• 🧩 FastAPI Backend for RESTful endpoints and async data processing
• 🗃️ MongoDB + Tebi Integration for scalable storage of transcripts, summaries, and metadata
• 📊 Performance Benchmarking utilities for ASR and summarization workloads

---

🏗️ System Architecture

                    ┌──────────────────┐
                    │   Audio Input    │
                    │ (.wav/.mp3 file) │
                    └───────┬──────────┘
                            │
                            ▼
                     ┌──────────────┐
                     │   FFmpeg     │
                     │ Audio Preproc│
                     └───────┬──────┘
                             │
                             ▼
                     ┌──────────────┐
                     │   Whisper    │
                     │  Transcribe  │
                     └───────┬──────┘
                             │
                             ▼
                     ┌──────────────┐
                     │   Gemini     │
                     │ Summarization│
                     └───────┬──────┘
                             │
                             ▼
                     ┌──────────────────┐
                     │ FastAPI + Mongo  │
                     │  + Tebi Backend  │
                     └──────────────────┘

---

⚙️ Tech Stack

Component	Technology
ASR	OpenAI Whisper
Summarization	Google Gemini API
Audio Processing	FFmpeg
Backend	FastAPI
Database	MongoDB, Tebi
Benchmarking	Custom Python Scripts

---

🧪 Benchmark Results

Tested on a 1-minute WAV file processed end-to-end.

Metric	Value / Range
Avg CPU Usage	~49.8%
Peak RAM Usage	~64.1%
Disk Read (MB)	Negligible (~0.08 MB total)
Disk Write (MB)	~15 MB (peaks around 5 MB/s)
Avg Processing Time	~25 seconds for 1 min audio
Throughput	~2.4× real-time

Interpretation: SummitAI maintains a stable CPU footprint (~50%) and consistent memory usage (~64%), with minimal disk I/O. This indicates efficient streaming and in-memory processing suitable for scalable multi-session workloads.

---

🧰 Setup Instructions

1. Clone the repository

git clone https://github.com/prabhsuratsingh/SummitAI.git
cd SummitAI

2. Build Docker container

docker build -t server .

3. Run the Container

docker run -p 8000:8000 server

---

📈 Future Roadmap

• 🔊 Real-time streaming ASR pipeline
• 🗣️ Speaker diarization and emotion tagging
• 📅 Meeting analytics dashboard (insights, action items)
• ☁️ Multi-cloud deployment (Tebi, GCP, AWS)

---

🧑‍💻 Author

Prabhsurat Singh Linkedin: @prabhsuratsingh