Skip to content

Latest commit

 

History

History
72 lines (52 loc) · 2.84 KB

README.md

File metadata and controls

72 lines (52 loc) · 2.84 KB
Raw

Azure IoT Edge Solution

Overview

This is a solution for Azure IoT Edge. It is based on the Azure IoT Edge.

Architecture

This solution is based on the following components:

  • GPT Skills
    • This module is responsible for generating text based on a prompt, where there are multiple skills (diferent prompts) that can be used to generate the text.
    • This uses the GPT3.5 model from Azure OpenAI.
    • This module is developed in Python after doing experiments in Azure ML PromptFlow.
  • Orchestrator
    • This module is responsible for orchestrating the GPT Skills and human interaction.
    • This module is developed in Python.
    • There are three primary block in it
      • Hearing the human voice via microphone and converting it to text using Azure Cognitive Services Speech to Text.
      • Generating text using GPT Skills.
      • Speaking the generated text via speaker using Azure Cognitive Services Text to Speech.

Getting Started

To get started with this solution, you need to do the following:

Prerequisites

  • Azure Subscription
  • Azure IoT Hub (Optional if you want to use Azure IoT Edge)
  • Azure Container Registry (Optional if you want to use Azure IoT Edge)
  • Azure OpenAI Service
  • Azure Cognitive Services Speech to Text and Text to Speech
  • Azure Bing Service (Web and News Search)
  • Open Weather Map API Key
  • Raspberry Pi 4 Model B
    • This solution consumes ~25% (1 core) of the CPU and ~1.5 GB of RAM.
  • Microphone attached to the Raspberry Pi
  • Speaker attached to the Raspberry Pi
  • GPIO touch button or press button attached to the Raspberry Pi
    • Two required, one for starting the conversation and one for stopping it.

Development

Standard Python development tools are required and any IDE can be used (Visual Studio Code, PyCharm, etc).

Configuration

Create a .env file in this directory by copying the .env.template file and filling in the required values.

Deployment (Azure IoT Edge)

To deploy this solution to Azure IoT Edge, you need to do the following:

  • Create an Azure IoT Hub
  • Create an Azure Container Registry
  • Create an Azure IoT Edge Device
  • Create an Azure IoT Edge Deployment
  • Build and push the Azure IoT Edge Modules
  • Deploy the Azure IoT Edge Deployment to the Azure IoT Edge Device

Deployment (Systemd)

To deploy this solution to a Linux machine, you need to do the following:

  • Install Python 3.9
  • Install the required Python packages
  • Install the required Linux packages
  • Create a systemd service
  • Start the systemd service

You can perform the above steps manually or use the Makefile to do it for you.