README.adoc

Amazon Transcribe and Comprehend using event sourcing

In this demo we are going to create a process pipeline to transcribe audio files and later on process sentiment analysis over the transcriptions. This demo uses the "Event Sourcing" pattern to provide a scalable and cost-efficient solution. The language for processing is spanish by default, you can change it in the SAM template.

Template parameters:

• ProcessBucketName - Name of the bucket to use for media and text storage.

• ProcessTableName - Name of the DynamoDB table used to store the process status and outcome.

• ComprehendLanguageCode - The language code for Amazon Comprehend, one of these languages.

• TranscribeLanguageCode - The language code for Amazon Transcribe, one of these languages.

Prerequisites:

• AWS Account

• SAM CLI - Install Guide

• IAM User with access keys and Adminitrator privileges.

Services used:

• Amazon Simple Storage Service

• AWS Lambda

• Amazon DynamoDB

• Amazon EventBridge

• Amazon Comprehend

• Amazon Transcribe

• Amazon Quicksight

Architecture:

Build

sam build

Deployment

sam package --output-template-file package.yaml --s3-bucket [ARTIFACTS_BUCKET_NAME] --s3-prefix transcribe-sentiment
sam deploy --template-file package.yaml --stack-name transcribe-sentiment --capabilities CAPABILITY_IAM

Where ARTIFACTS_BUCKET_NAME is an S3 bucket name used to upload the artifacts for deployment.