Whisper Processing AWS Stack

This is a full-stack processing pipeline for transcribing voice data using Whipser model.

It is built on top of AWS services, including S3, OpenSearch, Lambda, Step Functions, and Batch.

Development

Lambdas

We use AWS Chalice to develop lambdas.

Batch

The batch jobs are just plain-old Docker containers.

Deployment

We use AWS CDK to manage and deploy the entire stack.

Step Functions

The step function definitions are in infrastructure/stacks/processing_stack.py. Please make sure to edit based on Step Functions documentation.

Create ECR repositories

make deploy stack="whisper-processing-aws-ecr"

Deploy Lambda Docker Image

make build-lambdas

Deploy the Stack

make deploy stack="whisper-processing-aws-stack"