Synthesized Engine Sound Correction via Neural Network Masking

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional synthesized engine sounds do not effectively adapt to changing surrounding noise environments, leading to interference from external noises and an inadequate driver experience, particularly in electric vehicles where noise recognition is crucial for pedestrian safety.

Innovation Solution

An apparatus and method utilizing an artificial neural network (ANN) to learn masking level information from the surrounding noise environment, allowing for real-time correction of the synthesized engine sound to minimize noise interference and enhance the driver's experience.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional synthesized engine sound is output without considering surrounding noise environment, then the system is simple and computation is fast, but the synthesized sound is interrupted by external noise and driver experience is degraded

Engineering Contradiction:

Improvedriver experienceVSAvoidsystem complexity

Core Design Contradiction:

Ease of operationVSDevice complexity

Solution Approach 1:

The system dynamically adjusts the synthesized engine sound based on real-time surrounding noise environment detection. The noise environment is continuously monitored and the synthesized sound parameters are adaptively modified to maintain driver experience across varying conditions.

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system implements feedback by detecting the surrounding noise environment and using this information to correct the synthesized engine sound. The detected noise characteristics feed back into the synthesis process to ensure the output sound remains effective despite external interference.

Inventive Principle:

Principle #23Feedback

2Reliability

If synthesized engine sound is corrected based on masking level information from ANN, then the synthesized sound is protected from noise interference, but computation complexity increases

Engineering Contradiction:

Improvesynthesized sound qualityVSAvoidcomputation complexity

Core Design Contradiction:

ReliabilityVSDevice complexity

Solution Approach 1:

The artificial neural network is pre-trained offline to learn the relationship between surrounding noise environments and appropriate masking levels. This preliminary action transfers the computational burden from runtime to training time, enabling fast inference during actual operation while maintaining high reliability.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The system replaces complex real-time noise analysis and sound correction computations with a pre-trained neural network model. The ANN substitutes for traditional signal processing methods, providing efficient and reliable correction based on learned patterns rather than computationally intensive real-time calculations.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS10997962B2Apparatus and method for synthesizing engine sound

Publication Date: 2021.05.04 LG ELECTRONICS INC

AI summary

A method for synthesizing an engine sound includes outputting a first synthesized engine sound, obtaining a change in a first surrounding noise environment, learning an artificial neural network to obtain first masking level information corresponding to the obtained change in the first surrounding noise environment, generating a second synthesized engine sound by correcting the first synthesized engine sound based on the obtained first masking level information, and outputting the generated second synthesized engine sound.