awscdk-app-automated-caption-placement-demo

AWS CDK app for deploying the necessary resources for the automated caption placement demo

Demo

Uploading a video file to S3 bucket triggers the following jobs:

• Extract captions from the video file’s audio track using Amazon Transcribe API
• Detect faces in the video using Amazon Rekognition Face Detection API
• Detect texts in the video using Amazon Rekognition Text Detection API
• Manipulate the captions file based on the geometry information obtained from the face/text detections
• Transcode the video/captions file into HLS streams using AWS Elemental MediaConvert

Install

1. Setup CDK environment (including Node.js, Docker)

2. Install this CDK app

   $ git clone https://github.com/kuu/awscdk-app-automated-caption-placement-demo.git
   $ cd awscdk-app-automated-caption-placement-demo
   $ npm i

Deploy

$ npx cdk deploy

Resources

The following resources will be deployed:

• AWS::S3::Bucket x 2
• AWS::Events::Rule x 1
• AWS::StepFunctions::StateMachine x 1
• AWS::Lambda::Function x 12
• AWS::SNS::Topic x 2
• AWS::SQS::Queue x 2
• AWS::CloudFront::Distribution x 1

Outputs

After the deployment, the following output will be printed:

• InputBucket: S3 bucket to upload the input file to
• OutputBucket: S3 bucket to store the output files
• EventBridgeRuleName: The EventBridge Rule to recieve S3 notifications and trigger the StepFunction state machine
• PlaybackUrl1: HLS playlist transcoded with the original caption
• PlaybackUrl2: HLS playlist transcoded with the styled caption

Cleanup

$ npx cdk destroy