Intelligent Sound Classifier for Home Intrusion Detection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current home intrusion detection technologies, such as video monitoring and motion detection, lack the intelligence to differentiate between normal household activities and suspicious events, failing to accurately alert homeowners to potential threats.

Innovation Solution

The Intelligent Sound Classifier (ISC) system, integrated with Internet-connected devices, uses sound classification and machine learning to identify audio sources within a home, creating Sound Profile Models that can be shared across devices to differentiate between known and unknown sounds, providing accurate alerts based on spectral, time, and amplitude characteristics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional video monitoring and motion detection devices are used to detect home intrusion, then the device can monitor entrance points and windows, but it lacks the intelligence to differentiate between normal household activities and suspicious events, resulting in inaccurate alerts

Engineering Contradiction:
Improveaccuracy of threat detectionVSAvoidability to differentiate sound types
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The patent segments sound detection into distinct spectral, temporal, and amplitude characteristics. The sound classifier circuitry divides the audio signal analysis into multiple independent feature extraction processes, allowing precise identification of different sound types (e.g., glass breaking, shouting, normal household activities) without requiring complex adaptive learning for each scenario.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transforms the sound detection approach by analyzing multiple parameters simultaneously - spectral content (frequency distribution), temporal characteristics (time patterns), and amplitude features (intensity variations). This multi-parameter analysis enables accurate differentiation between threatening and normal events without requiring the system to adapt to each specific environment.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If sound classification circuitry analyzes spectral, time, and amplitude characteristics to identify audio sources, then the system can accurately differentiate between normal and suspicious sounds, but the device complexity increases due to multiple processing components

Engineering Contradiction:
Improvesound source identification accuracyVSAvoidnumber of processing components
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent merges multiple sound analysis functions into a single integrated sound classifier circuitry component. This unified circuitry simultaneously performs spectral analysis, temporal pattern recognition, and amplitude evaluation, achieving high identification accuracy without the complexity of multiple separate processing modules.

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The sound classifier circuitry is designed as a universal component that handles multiple types of sound analysis (spectral, temporal, amplitude) within a single device. This multi-functional approach eliminates the need for separate dedicated circuits for each type of analysis, reducing overall device complexity while maintaining high precision.

Inventive Principle:
Principle #6Universality (Multi-functionality)

3Speed

If the system stores local sound profiles for comparison, then the device can quickly identify known sounds, but the sound profile memory requires significant storage capacity to maintain comprehensive audio source databases

Engineering Contradiction:
Improvesound matching speedVSAvoidmemory storage capacity
Core Design Contradiction:
SpeedVSQuantity of substance

Solution Approach 1:

The patent extracts only the essential characteristics of sounds (spectral, temporal, and amplitude features) for storage in the sound profile memory, rather than storing complete audio recordings or complex waveform data. This extraction approach enables fast comparison and identification while minimizing the storage capacity required in the sound profile memory.

Inventive Principle:
Principle #2Taking out (Extraction)

4Adaptability or versatility

If the device communicates with remote sound profile exchange to obtain unknown sound profiles, then the system can expand its recognition capabilities, but the loss of time occurs during network communication and profile retrieval

Engineering Contradiction:
Improverecognition capability expansionVSAvoidnetwork communication delay
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by pre-processing and characterizing sounds into compact spectral, temporal, and amplitude feature sets before storage and exchange. This preliminary characterization allows for extremely fast local comparison operations, minimizing the time penalty associated with network communication and profile retrieval, as the actual matching process operates on simplified feature data rather than raw audio.

Inventive Principle:
Principle #10Preliminary action

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

The ISC system effectively differentiates between normal household sounds and potential threats, reducing false alarms and enhancing home security by accurately identifying the origin of detected sounds, even in varying acoustic environments.

Implementation Method 1

converting, by a microphone, sound waves to electrical signals

Methodology Applied
Scientific EffectAcoustic transduction:

Data Source

PatentEP3483851B1Intelligent sound classification and alerting
Publication Date: 2020.10.14 HONEYWELL INTERNATIONAL INC
  • EP3483851B1 patent drawingFigure 1
  • EP3483851B1 patent drawingFigure 2
  • EP3483851B1 patent drawingFigure 3

AI summary

A device can include a microphone, sound classifier circuitry to analyze signals from the microphone and determine a profile for each of the spliced segments, the profile indicating one or more spectral, time, or amplitude characteristics of the signals, a sounds profile memory to store sound profiles of audio sources in a building in which the device resides, and sound profile manager circuitry to determine whether the profile sufficiently matches a sound profile of the sound profile memory and, in response to a determination that the sound profile does not sufficiently match any sound profiles in the sound profile memory, provide a communication to a sound profile exchange remote to the device, the communication requesting the sound profile exchange to determine an audio source to associate with the sound profile.