AI Story teller

• 

  Project logo

• 

  story listening page

• 

  Diagram

• 

  main page

Inspiration

At bedtime, I noticed my child had memorized all our books, and I kept forgetting to buy new ones. That’s when I decided to create my own stories using AI and ML.

What it does

The AI Story Teller generates personalized children's stories in seconds:

• Creates unique stories based on child's name, age, and theme
• Generates accompanying illustrations using AI image models
• Produces natural-sounding audio narrations
• Presents stories in an interactive, kid-friendly interface

How we built it

Frontend: React with TypeScript, styled-components
Backend: AWS Serverless Architecture:

• API Gateway + Lambda for REST APIs
• Bedrock for text (Claude 3 Haiku) and image (Titan) generation
• Polly for text-to-speech audio with long-form engine
• S3 for asset storage
• DynamoDB for metadata
  IAC: AWS CDK

Challenges we ran into

• Synchronizing async story generation across multiple AWS services
• Preventing audio interruptions during progressive content loading
• Optimizing Bedrock model prompts for child-friendly content
• Creating a UI that's both magical and intuitive for children

Accomplishments that we're proud of

Built a complete AI storytelling pipeline Built an event-driven backend application Nice kid-friendly UI, my daughter liked it

What we learned

Tested new AWS Polly engines, in the past I worked only with the Standard engine Worked with different Bedrock models, it's totally a new experience for me

What's next for AI Story teller

• Add video generation using Nova Reel
• Add more customization options
• Add speech to story supporting, to support the kid in describing what he wants and get a story
• Create a library of generated stories to allow list content generated in the past
• Add multilingual support
• Develop companion mobile apps

Built With

• apigateway
• bedrock
• cdk
• cloudfront
• dynamodb
• lambda
• polly
• s3
• sns