At a Glance

Name: AI meeting companion: From voice to insight
Price: Free CAD
Rating: 4.7 (15 reviews)
Author: sina_nazeri, fateme_akbari, vicky_kuo

Create an app to capture audio (like lectures) and summarize it. Build the app using OpenAI Whisper (text to speech), then summarize it with an open source LLAMA 2 LLM hosted by IBM watsonx. You deploy the app in a serverless environment using IBM Cloud Code Engine.

Imagine that you're a student who records your teacher's lectures. But, what if you need to turn that recorded lecture into text? There's a handy AI app that can do that for you, accurately. It also summarizes the lecture and highlights the main points.

In this project, you use OpenAI's Whisper to transform speech into text. Then, you use IBM watsonx AI to summarize and find key points. This stage couples with prompt engineering through PromptTemplate in Langchain. You'll make an app with HuggingFace Gradio as the user interface.

The output from the LLM not only summarizes and highlights key points, it also corrects minor mistakes made by the speech-to-text model, ensuring a coherent and accurate result.

A look at the project ahead

Following are the key objectives for your project:

Speech-to-text conversion: Use OpenAI's Whisper technology to convert lecture recordings into text, accurately.
Content summarization: Implement IBM watsonx AI to effectively summarize the transcribed lectures and extract key points.
User interface development: Create an intuitive and user-friendly interface using HuggingFace Gradio, ensuring ease of use for students and educators.
App deployment: Learn and apply the skills necessary to deploy the application online by using IBM Code Engine, making the tool accessible to a wider audience.

What you'll need

To work through this guided project, you need a general knowledge of Python a well as a browser.

Offered By: IBMSkillsNetwork

AI meeting companion: From voice to insight

At a Glance

A look at the project ahead

What you'll need