Skip to content

Proof-of-concept for deploying Amazon Nova multimodal models on AWS. Includes hands-off, reference architecture using Amazon Bedrock and AWS services for text, image, and speech tasks. Ideal for customers exploring scalable, production-ready Nova integration.

License

Notifications You must be signed in to change notification settings

doitintl/amazon-nova-architecture-reference

Repository files navigation

Nova Sonic: Speech-to-Speech with Amazon Bedrock

Nova Sonic is a speech-to-speech conversation application powered by Amazon Bedrock and Amazon Nova Sonic. It enables real-time voice conversations with an AI assistant, featuring WebRTC for audio streaming and a restaurant booking capability.

Key Features

  • Real-time Speech-to-Speech Conversations: Engage in natural voice conversations with an AI assistant powered by Amazon Bedrock
  • WebRTC Integration: Low-latency audio streaming for responsive interactions
  • Restaurant Booking Capability: Demonstrate practical use cases with a restaurant booking feature
  • Real-time Event System: AppSync Events API for real-time updates and notifications
  • Scalable Architecture: Deployed on AWS with auto-scaling capabilities
  • Multiple Deployment Options: Support for both ECS and EC2-based deployments

Architecture Overview

Nova Sonic Architecture

The Nova Sonic application consists of several key components:

  • Frontend Web Application: WebRTC-enabled interface for user interactions
  • Backend API Service: Processes audio streams and manages conversations
  • Amazon Bedrock Integration: Powers the AI conversation capabilities
  • DynamoDB: Stores conversation history and restaurant bookings
  • AppSync Events API: Provides real-time publish/subscribe functionality

For a detailed explanation of the architecture, see ARCHITECTURE.md.

Quick Start

Prerequisites

  • AWS Account with access to Amazon Bedrock
  • AWS CLI configured with appropriate credentials
  • Node.js 14.x or later
  • AWS CDK v2 installed
  • Docker for building container images

Deployment

  1. Clone the repository:

    git clone https://github.com/yourusername/speech-to-speech-amazon-nova-sonic.git
    cd speech-to-speech-amazon-nova-sonic
  2. Configure environment variables:

    cd nova-sonic/api
    cp .env.example .env
    # Edit .env with your settings
  3. Deploy the CDK stacks:

    cd ../../cdk
    npm install
    cdk bootstrap
    cdk deploy --all

For detailed deployment instructions, see DEPLOYMENT.md.

Documentation

License

This project is licensed under the MIT License - see the LICENSE file for details.

About

Proof-of-concept for deploying Amazon Nova multimodal models on AWS. Includes hands-off, reference architecture using Amazon Bedrock and AWS services for text, image, and speech tasks. Ideal for customers exploring scalable, production-ready Nova integration.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published