Business AI Meeting Companion 🚀

Hi! I'm Garbii, and this is my personal project: Business AI Meeting Companion.

---

📝 About This Project

This is a personal project I built to experiment with AI-powered meeting tools. The app captures meeting conversations, transcribes them using OpenAI's Whisper, and then summarizes the transcript and extracts key points using IBM WatsonX with Llama 3. The interface is built with Gradio for easy use.

---

🎯 What I Learned

Working on this project helped me:

• 🧑‍💻 Write Python scripts that use large language models (LLMs)
• 🗣️ Integrate OpenAI's Whisper for accurate speech-to-text
• 🤖 Use IBM WatsonX (Llama 3) to summarize and extract key points from text
• 🖥️ Build a user-friendly web UI with Gradio
• 🔗 Orchestrate LLM prompts and workflows with LangChain

---

🛠️ Core Technologies

Technology	Purpose
Whisper	Speech-to-Text (ASR)
IBM WatsonX (Llama 3)	Language Model for Summarization
Gradio	User Interface
LangChain	Prompt Orchestration
Python	Programming Language

---

⚙️ How to Run It

Tip: I recommend using a virtual environment for Python projects.

1️⃣ Set Up Your Environment

pip3 install virtualenv
virtualenv my_env
# On Linux/Mac
source my_env/bin/activate
# On Windows
.\my_env\Scripts\activate

2️⃣ Install Dependencies

pip install transformers==4.36.0 torch==2.1.1 gradio==4.23.0 langchain==0.0.343 ibm_watson_machine_learning==1.0.335 huggingface-hub==0.20.1

3️⃣ Install FFmpeg

• Linux:

  sudo apt update
  sudo apt install ffmpeg -y

• Windows: Download from ffmpeg.org and add to your PATH.

---

🚦 Usage & Demo

1. Test Speech-to-Text

Run:

python3 simple_speech2text.py

2. Try the Gradio Transcription App

Run:

python3 speech2text_app.py

Then open http://0.0.0.0:7860 in your browser.

3. Test Llama 3 Summarization

Run:

python3 simple_llm.py

4. Full Meeting Analyzer App

Run:

python3 speech_analyzer.py

Then open http://0.0.0.0:7860, upload a meeting recording, and see the AI-generated summary and key points!

---

📄 License

This project is open source under the MIT License. The code and models for OpenAI's Whisper are released under the MIT License. Everything here is for educational and personal use.