Meeting Summarizer 📝

A Streamlit-based web application that generates concise summaries from meeting transcripts using OpenAI's GPT-4o-mini model. The app supports both text transcripts and audio files (transcribed via Whisper) and allows you to specify a custom meeting name.

Features 🚀

• Upload TXT transcripts or audio files.
• Whisper-based transcription for audio files.
• Custom meeting name via a text box.
• Summaries include key discussion points, takeaways, and action items in Markdown format.
• OpenAI API key validation (sk-proj-...) for security.
• Clean, user-friendly Streamlit interface.

Prerequisites ⚙️

• Python 3.x
• OpenAI API key (starts with sk-proj-)

Installation 📦

1. Clone the repository:

   git clone https://github.com/alejandroarguello/meeting_summarizer.git
   cd meeting_summarizer
   

2. Install dependencies:

   pip install -r requirements.txt
   
   

3. Create a .env file in the root directory:

   OPENAI_API_KEY=sk-proj-your-key-here
   

Usage ▶️

1. Run the app explicitly using the Python interpreter from your active conda environment:

   python -m streamlit run app.py
   
   

2. In the web interface: • Enter the meeting name (e.g., “Weekly Marketing Sync”). • Choose Text Transcript or Audio File. • Upload either: • A .txt transcript, or • An audio file (.mp3, .wav, .m4a, .ogg). • Click Generate Summary.

3. View the generated summary in Markdown format.

Example 🎯

In the transcripts folder you have a file named 'sample_transcript.txt', upload it and set the meeting name to “Product Roadmap Review” to see the summarized result. You also have an example audio meeting named 'denver_extract.mp3' to try the Whisper transcribing capabilities.

Configuration 🔧

File	Purpose
app.py	Main application logic
.env	Stores OpenAI API key
requirements.txt	Lists Python dependencies

Troubleshooting ❌

❌ "API key invalid" → Ensure your key starts with 'sk-proj-' and has no whitespace.

❌ File upload errors → Verify the TXT file is properly formatted.

❌ Audio transcription issues → Make sure the uploaded file is a valid .mp3, .wav, .m4a, or .ogg. Large files may take longer.

❌ Meeting name not showing up properly → Double-check you’ve entered text in the “Enter the meeting name” field.

Note 📌

•	This app uses OpenAI’s GPT-4o-mini for final summarization.
•	Audio is transcribed via OpenAI Whisper.
•	The app works with general meeting transcripts in .txt format.
•	Meeting name input personalizes the summary to reflect your actual use case (e.g. “Board Meeting,” “Team Sync,” “Denver council,” etc.).

Interface example 🖼️