Simulated Abnormal Sound Generation for Acoustic Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generating a suitable learning model for sound data processing is challenging due to the difficulty in acquiring large amounts of sound data with appropriate features, especially for abnormal sound detection, where existing methods may fail to detect subtle abnormalities.
Innovation Solution
A sound data processing method that generates simulated abnormal sound data and similar sound data based on normal sound data to augment learning data, allowing for machine learning-based abnormal sound detection using a sound data processing device with a processing unit configured to acquire and process sound data, including a simulated abnormal sound generation unit and a machine learning unit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large amount of sound data with appropriate features is acquired for machine learning, then the accuracy of abnormal sound detection is improved, but the difficulty of acquiring such data increases
Solution Approach 1:
The patent generates simulated abnormal sound data by copying and modifying normal sound data through signal processing techniques. The simulated abnormal sound generation unit creates artificial abnormal sounds from normal sound recordings, effectively copying the structure of real abnormal sounds without requiring actual abnormal sound acquisitions. This resolves the contradiction by providing sufficient training data (improving measurement precision) while avoiding the difficulty of acquiring real abnormal sound data.
Solution Approach 2:
The patent applies parameter changes by modifying sound data characteristics through various signal processing operations. The simulated abnormal sound generation unit changes parameters such as frequency, amplitude, and temporal characteristics of normal sound data to create simulated abnormal sounds. This transformation allows the system to generate diverse training data (improving accuracy) while using easily obtainable normal sound data as the base material.
2Quantity of substance
If normal sound data is used to generate simulated abnormal sound data, then the quantity of learning data is increased, but the complexity of data processing increases
Solution Approach 1:
The patent segments the data processing into distinct functional modules: a normal sound data acquisition unit, a simulated abnormal sound generation unit, and a machine learning unit. The generation unit further segments the processing into feature extraction, parameter modification, and data synthesis steps. This segmentation increases the quantity of learning data while managing complexity through modular organization, where each module has a specific function and can be independently optimized.
Solution Approach 2:
The system employs self-service by using the normal sound data itself as the foundation for generating simulated abnormal sound data. The normal sound data serves dual purposes: as actual training data and as the source material for generating additional training data. This self-service approach increases the quantity of learning data without requiring external data sources, and the processing complexity is managed through automated signal processing techniques that build upon each other systematically.
Data Source
AI summary
A sound data processing method of a sound data processing device, the sound data processing device including a processing unit configured to acquire sound data of a target by input and to process the sound data, the sound data processing method including: a step of generating, by using acquired normal sound data of the target, simulated abnormal sound data that becomes a simulated abnormal sound of the target; and a step of performing machine learning by using the acquired normal sound data and the generated simulated abnormal sound data as learning sound data, and generating a learning model for determining an abnormal sound of the sound data of the target to perform abnormal sound detection.