Day 2 Meeting Minutes from Audio with Free Open Models | Whisper + Llama 3.2, by The Kurta Coder on YouTube. 5 min 9 s.
- Video length
- 5 min 9 s
- Time to do it
- 45 minutes
- Difficulty
- Medium
- Steps
- 10
What you need
Steps
Open the Colab notebook linked in the video description and sign in with your Google account.
In Colab, go to Runtime, then Change runtime type, and select T4 GPU as the hardware accelerator.
Install the needed libraries in a cell: transformers, bitsandbytes, accelerate, torch and datasets.
Load Whisper medium.en through a Hugging Face pipeline for automatic speech recognition and pass your meeting audio file to it.
Save the returned transcript text to a variable so you can review it before writing minutes.
Load Llama 3.2 3B Instruct with 4-bit quantization using bitsandbytes so it fits on the free T4 GPU.
Write a prompt that asks the model for a summary, decisions, action items with owners and due dates, and a quality check.
Run the prompt with the transcript and print the structured minutes.
If you get the CUDA is required but not available error, check that the T4 GPU runtime is selected and restart the session.
Ask the model to translate the minutes into French or Spanish by adding a translation instruction to the prompt.
Tips and common mistakes
Use a short clip first, since a full hour of audio takes a long time to transcribe on a free GPU.
Keep the transcript and minutes in separate variables so you can rerun the Llama step without transcribing again.
Check the transcript for names and numbers before trusting the action items.
Free Colab sessions disconnect after idle time, so save your notebook and outputs often.
Questions
Do I need a paid GPU for this?
No. The pipeline is built to run on a free Google Colab T4 GPU using 4-bit quantization for the 3B model.
Which languages does it support?
The demo covers English, French and Spanish. Whisper medium.en handles English, so use a multilingual Whisper model for other languages.
Can I use this for real company meetings?
The creator calls it a learning project, not production ready. Review and edit the minutes yourself before sharing them.
This guide was drafted from the video's details and general task knowledge. It has not been checked against the full video. Check the video and relevant product instructions before following a procedure. Report a correction.
