Improve efficiency: Automate meeting summaries with Amazon Transcribe, SageMaker, and LLMs.

SeniorTechInfo
3 Min Read







Automatically Generate Meeting Summaries with AI


Automatically Generate Meeting Summaries with AI

The prevalence of virtual business meetings in the corporate world, largely accelerated by the COVID-19 pandemic, is set to continue growing. According to a survey by American Express, 41% of business meetings are expected to be hybrid or virtual by 2024. Keeping track of multiple meetings daily and managing ongoing topics can become increasingly challenging over time, impacting project timelines and customer relationships.

One solution to this challenge is automating the creation of meeting summaries using generative artificial intelligence (AI) and speech-to-text technologies. By automatically generating summaries at the end of each call, attendees can focus on the conversation knowing a transcript will be provided.

Solution Overview

Our solution automates the generation of meeting summaries from recorded virtual meetings using AWS services like Amazon Transcribe and Hugging Face. The process involves transcribing the meeting recording, processing it with a language model from Hugging Face on Amazon SageMaker, and delivering the summaries to meeting attendees via email notifications.

Prerequisites

  • AWS account
  • Basic knowledge of AWS and Python

Deploy the Solution

To deploy the solution in your AWS account, clone the GitHub repository, install the required dependencies, and use the AWS CDK to deploy the infrastructure. Be sure to specify the email address for summary notifications during deployment.

git clone https://github.com/aws-samples/audio-conversation-summary-with-hugging-face-and-transcribe.git
cd audio-conversation-summary-with-hugging-face-and-transcribe/infrastructure
pip install -r requirements.txt
cdk deploy --parameters SubscriberEmailAddress="<SUBSCRIBER_MAIL_ADDRESS>"

Test the Solution

You can test the solution by uploading a sample meeting recording to the project’s S3 bucket. The summary will be generated and sent to the specified email address. The entire process takes approximately 2 minutes for a 250 token input.

Conclusion

In conclusion, our architecture pattern enables the automated transformation of meeting recordings into insightful conversation summaries using AWS Cloud and Hugging Face AI technologies. By leveraging managed AI services and ML models, you can accelerate the development of generative AI applications for diverse use cases.

References

Authors

  • Gabriel Rodriguez Garcia
  • Jahed Zaïdi
  • Mateusz Zaremba
  • Kemeng Zhang


Share This Article
Leave a comment

Leave a Reply

Your email address will not be published. Required fields are marked *