Noise-Augmented Mapping for Vehicle Sound-Cause Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for identifying the cause of sounds in vehicles using machine-learned mappings are hindered by the variability in noise removal accuracy due to superimposed noise components in sound signals, leading to unstable identification of sound causes.
Innovation Solution
A method involving signal correction and machine learning is employed, where a noise signal is superimposed on the sound signal to correct it, and the corrected signal is used as training data to update the mapping, enhancing the accuracy of identifying sound causes by incorporating background noise components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If noise removal processing is performed on the sound signal, then the clarity of the sound signal is improved, but the accuracy of sound cause identification becomes unstable due to varying noise removal effectiveness
Solution Approach 1:
The patent converts the harmful noise components in the sound signal into beneficial training data by intentionally adding various noise signals to the training sound signals. This allows the mapping to learn and adapt to noisy environments, transforming the previously problematic noise into a useful element that improves identification stability under real-world conditions.
Solution Approach 2:
The patent performs preliminary noise addition to the training data before the mapping is deployed for actual sound cause identification. By pre-exposing the mapping to noisy training signals, the system prepares the model to handle real-world noise variations, ensuring stable performance when deployed in actual vehicle environments with background noise.
2Measurement precision
If the mapping is trained with clean sound signals only, then the theoretical accuracy is maximized, but the practical performance degrades when background noise is present
Solution Approach 1:
The patent changes the parameters of the training data by systematically varying the noise signal characteristics (type, level, frequency content) added to the training sound signals. This exposes the mapping to a wide range of noise conditions during training, enabling it to maintain high identification accuracy across diverse real-world acoustic environments.
Solution Approach 2:
The patent makes the mapping universally adaptable to different noise environments by training it with multi-noise-condition data. The resulting mapping can handle both clean and noisy sound signals effectively, achieving multi-functionality in terms of environmental adaptability while maintaining identification accuracy.
Data Source
AI summary
A mapping, which is as a subject of learning, is a learning model that uses a sound signal as an input variable and outputs a variable indicating the cause of a sound in a vehicle. A learning method includes a signal correction process that corrects a sound signal by superimposing a noise signal on the sound signal, and an update process that updates the mapping through machine learning in which the sound signal corrected in the signal correction process serves as training data and a cause of a sound paired with the sound signal serves as teaching data.