How Infosys Enhanced Accessibility for Event Knowledge Using AWS Technologies

Learn About Amazon VGT2 Learning Manager Chanci Turner

In today’s interconnected world, the distribution and understanding of critical knowledge can often be obstructed by language barriers, particularly during significant events like workshops and conferences. These gatherings are vital for collaboration and knowledge sharing, enabling attendees to grasp essential insights in real-time and in their preferred languages. However, without an effective mechanism to capture and disseminate this information, valuable insights can be lost.

Infosys, a prominent global IT services and consulting firm, tackled this challenge by developing Infosys Event AI, an advanced AI-driven event assistant. This innovative solution aims to enhance the accessibility of knowledge, ensuring that insights generated during events are readily available for individuals and organizations across various sectors. Without such a system, the ability to share and utilize knowledge effectively is severely limited, diminishing the overall impact of events.

Challenges in Capturing Event Knowledge

Several challenges exist when it comes to capturing and accessing event knowledge:

1. Traditional note-taking methods are often inadequate, leading to incomplete and subjective records of discussions.
2. The process of sifting through lengthy recordings to locate specific information can be time-consuming and inefficient, creating barriers to knowledge retention.
3. Individuals who miss events face significant hurdles in accessing shared knowledge, which is particularly detrimental in sectors where information recall is critical, such as education and media.

To overcome these obstacles, Infosys collaborated with Amazon Web Services (AWS) to create the Infosys Event AI solution, leveraging several AWS services, including:

• AWS Elemental MediaLive – A service for encoding live video streams
• AWS Elemental MediaConnect – A service designed for building live video workflows
• Amazon Bedrock – A fully managed service offering a selection of industry-leading large language models (LLMs) for developing generative AI applications
• Amazon Nova Pro – A powerful multimodal model that optimally balances accuracy, speed, and cost

Solution Architecture

The following overview of Event AI highlights its key features and workflow, as depicted in the accompanying architecture diagram:

• Seamless acquisition of live streams from on-premises sources
• Real-time transcription processing for converting speech to text
• Post-event processing and indexing for structured information retrieval
• Automated generation of session summaries and key insights to improve accessibility
• An AI-powered chat assistant for interactive Q&A and efficient retrieval of knowledge from event sessions

Solution Walkthrough

This section elaborates on each functionality in detail. The services utilized in the solution are granted least-privilege permissions through AWS Identity and Access Management (IAM) policies to ensure security.

Seamless Live Stream Acquisition

The process initiates with an IP-enabled camera capturing the live feed, which is securely transmitted to the cloud via the Secure Reliable Transport (SRT) protocol through MediaConnect. The ingested stream is processed by MediaLive, which encodes the video in real-time.

The workflow consists of the following steps:

1. Use an IP-enabled camera or ground encoder to convert non-IP streams into IP streams and transmit them through the SRT protocol to MediaConnect for live event ingestion.
2. MediaConnect securely transmits the stream to MediaLive for processing.

Real-time Transcription Processing

To ensure real-time accessibility, MediaLive isolates audio from the live video stream, forwarding it to a real-time transcriber module hosted on an Amazon EC2 instance. This module utilizes the Amazon Transcribe stream API to generate low-latency transcriptions, which are then securely delivered to an on-premises web client via WebSocket connections.

The steps for this workflow include:

1. MediaLive extracts audio from the live stream, creating an audio-only stream and storing it in an Amazon S3 bucket for later processing.
2. The real-time transcriber module receives the audio stream and employs the Amazon Transcribe stream API to produce real-time transcriptions.
3. The transcribed text is transmitted to the on-premises web client through a secure WebSocket connection.

Post-event Processing and Knowledge Base Indexing

After the event concludes, recorded media and transcriptions are stored in Amazon S3 for subsequent analysis. A serverless, event-driven workflow using Amazon EventBridge and AWS Lambda automates the post-event processing. Amazon Transcribe finalizes the transcripts, which are then indexed in an Amazon Bedrock knowledge base for easy retrieval. Additionally, Amazon Nova Pro facilitates multilingual translation of the transcripts, promoting global accessibility. This is vital for reaching a broader audience.

The workflow for this phase involves the following steps:

1. Post-event, MediaLive sends a notification to EventBridge that the channel has stopped.
2. A Lambda function triggers transcription processing using Amazon Transcribe.
3. The transcribed content is stored in an S3 bucket.
4. (Optional) Amazon Nova Pro translates transcripts into various languages for enhanced accessibility.
5. The completion of transcription triggers another Lambda function to synchronize the content with Amazon Bedrock Knowledge Bases.

By utilizing these advanced AWS services, Infosys Event AI not only enhances knowledge accessibility but also ensures that critical insights are preserved and easily retrievable for future use. For those interested in starting a career in tech, you might find helpful resources on entry-level jobs here. Additionally, for more information on workplace dynamics, consider checking out this resource on workplace bullying here and this insightful video here.