Overall architecture

This section provides an overview of deeply API architecture

Overall ArchitectureOverall Architecture of Deeply AI System
Overall Architecture of Deeply AI System

This diagram provides a high-level overview of the Deeply system, illustrating the connections and relations between the microphones, edge server, central server.

Introduction

Deeply system architecture is designed to monitor audio events in real-time, process them efficiently, and sound data. This architecture consists of multiple components working together seamlessly, including microphones, an edge server, a central server, and a dashboard.

Components

Edge Server

The Edge Server is a critical component in the system, responsible for initial processing of audio data received from the microphones. Its main roles include:

  • Audio Data Processing: Preprocesses audio data to reduce noise and isolate relevant signals.
  • Event Detection: Identifies specific audio events based on predefined thresholds and labels.
  • Data Transmission: Communicates processed data to the central server via API calls, ensuring efficient data transfer and reducing network load.

The edge server is strategically connected to the microphones to minimize latency and ensure faster sound data processing.

Event

The Event class represents an audio event detected by the system. Each event is identified by a unique label, confidence score, and timestamp, providing detailed insights into the detected sound. The key attributes of an event include:

  • ID: A unique identifier for the event.
  • Label: Classification of the audio event (e.g., "breathing heavily").
  • Confidence: Confidence score indicating the accuracy of the classification.
  • Audio File: Path to the recorded audio file associated with the event.
  • Device Information: Details of the device that recorded the event, including device ID, name, location, and status.

The event data is stored and processed by the central server, allowing users to view and analyze the events on demand.

Refer to event data for detailed event data

By combining real-time data processing with detailed event tracking, Deeply system architecture ensures efficient audio event monitoring and management, supporting both real-time and historical analysis.