Operational Sound Segmentation for Privacy-Safe Abnormality Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for detecting abnormalities in apparatuses by analyzing emitted sounds risk leaking customer privacy information due to the recording of ambient conversations or voices, which may contain sensitive data.
Innovation Solution
An information processing apparatus that divides sound data into sections where word detection is not possible and transmits specific section sound data along with operation information to an external server for analysis, ensuring that privacy information is protected by avoiding the specification of conversations or words.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If sound data is recorded and transmitted for abnormality detection, then the ability to detect apparatus abnormalities is improved, but customer privacy information may be leaked through recorded conversations or voices
Solution Approach 1:
The sound data is divided into multiple sections based on operation information timing. Only specific sections corresponding to actual apparatus operations are selected and transmitted for analysis, while other sections containing potential privacy information (conversations, ambient sounds) are excluded. This segmentation approach maintains abnormality detection capability for operational sounds while preventing privacy leakage from non-operational periods.
2Measurement precision
If all recorded sound data is transmitted for analysis, then abnormality detection accuracy is improved, but data transmission volume and processing load increase
Solution Approach 1:
The system extracts and transmits only the specific sections of sound data that correspond to actual apparatus operations, as indicated by operation information. This extraction process removes unnecessary data (conversations, ambient sounds, non-operational periods) while retaining the critical operational sound data needed for accurate abnormality detection, thereby reducing transmission volume without compromising detection accuracy.
Data Source
AI summary
An information processing apparatus includes a processor configured to: acquire sound data representing a sound emitted by an apparatus; generate plural pieces of section sound data by dividing the sound data into sections where detection of words is not possible; and transmit specific section sound data among the plural pieces of section sound data, and operation information that indicates an operation of the apparatus and corresponds to the specific section sound data, to an external apparatus that analyzes the sound data of the apparatus, in association with each other.