Groqnotes is a streamlit app that scaffolds the creation of structured lecture notes by iteratively parsing and generating notes from transcribed audio lectures using Groq's Whisper API. The app mixes Llama3-8b and Llama3-70b, utilizing the larger model for generating the notes structure and the faster of the two for creating the content.

Demo of Groqnotes fast transcription of audio and generation of structured notes

• 🎧 Generate structured notes using transcribed audio by Whisper-large and text by Llama3
• ⚡ Lightning fast speed transcribing audio and generating text using Groq
• 📖 Scaffolded prompting strategically switches between Llama3-70b and Llama3-8b to balance speed and quality
• 🖊️ Markdown styling creates aesthetic notes on the streamlit app that can include tables and code
• 📂 Allows user to download a text or PDF file with the entire notes contents

As with all generative AI, content may include inaccurate or placeholder information. Groqnotes is in beta and all feedback is welcome!

To use Groqnotes, you can use the hosted version at groqnotes.streamlit.app

Alternative, you can run groqnotes locally with streamlit.

First, you can set your Groq API key in the environment variables:

export $GROQ_API_KEY = gsk_yA...

This is an optional step that allows you to skip setting the Groq API key later in the streamlit app.

Next, you can set up a virtual environment and install the dependencies.

python3 -m venv venv

source venv/bin/activate

pip3 install -r requirements.txt

Finally, you can run the streamlit app.

python3 -m streamlit run main.py

• Streamlit
• Llama3 on Groq Cloud
• Whisper-large on Groq Cloud

Groqnotes may generate inaccurate information or placeholder content. It should be used to generate books for entertainment purposes only.

Improvements through PRs are welcome!

Coming soon!

• Create summary version of transcript, batching into sections of n characters.
• Allow upload of multiple audio files