Vehicle Abnormal Sound Diagnosis Using Voice-Marked Spectrograms
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sound vibration analysis devices struggle to accurately diagnose abnormal vehicle sounds due to interference from non-abnormal sounds, particularly human voices, which affect diagnosis accuracy.
Innovation Solution
An abnormal sound diagnostic system that utilizes spectrograms to visually distinguish between time zones with and without human voice, allowing operators to select or set diagnosis ranges effectively, and incorporates machine learning for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the complete sound data acquired during vehicle traveling is used for diagnosis, then all sound information is available for analysis, but the diagnosis accuracy deteriorates due to interference from non-abnormal sounds such as human voices
Solution Approach 1:
The voice recognition unit extracts and identifies human voice components from the sound data, and the display control unit separates the utterance time zones from the spectrogram display. This extraction principle removes the harmful human voice interference from the diagnosis process while preserving the useful abnormal sound information for accurate diagnosis
Solution Approach 2:
The system performs voice recognition and identifies utterance time zones before the actual diagnosis process. By preliminarily marking and separating the time zones containing human voices, the system prevents users from accidentally selecting these interference regions during diagnosis, ensuring accurate diagnosis results
2Measurement precision
If the spectrogram displays the complete time range including utterance zones, then all time information is available, but the user may accidentally select noisy ranges for diagnosis
Solution Approach 1:
The display control unit uses visual differentiation (color changes or shading) to distinguish utterance time zones from normal time zones in the spectrogram. This visual cue system helps users quickly identify and avoid selecting noisy ranges containing human voices, improving both measurement precision and ease of operation by making the selection process more intuitive
Data Source
AI summary
The abnormal sound diagnostic system of the present disclosure includes an abnormal sound diagnostic system including a diagnostic device for diagnosing abnormal sound generated in a vehicle based on data of sound emitted from a vehicle and inquiry information about abnormal sound generated in a vehicle, the abnormal sound diagnostic system including an calculation processing unit for acquiring a spectrogram indicating a relationship between time, frequency, and sound pressure from the sound data, a voice recognition unit for extracting a vocal time zone that is a time zone including a human voice from the sound data by speech recognition, a display unit for displaying a spectrogram, and a display control unit for causing the display unit to display a spectrogram so that the vocal time zone extracted by the voice recognition unit is visually distinguished from a time zone that does not include a human voice.