Variable Sound Decomposition Masks for Non-Stationary Noise
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional sound decomposition techniques are inadequate in addressing non-stationary noises, such as a police siren or a dog barking, as they often fail to remove sufficient noise from sound data, leaving it 'noisy' and less than ideal.
Innovation Solution
The implementation of variable sound decomposition masking techniques, where a user-defined threshold is used to generate a mask that assigns portions of sound data to respective sources, incorporating a training model and sound separation model to improve noise removal, particularly for non-stationary noises.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional sound decomposition techniques are used, then stationary noises can be handled well, but non-stationary noises cannot be removed sufficiently
Solution Approach 1:
The patent applies dynamics by making the mask variable rather than fixed. The mask dynamically adjusts based on the characteristics of non-stationary noises detected in the sound data, allowing the decomposition process to adapt to changing noise patterns over time and frequency, thereby improving effectiveness against non-stationary noises while maintaining performance on stationary noises
2Adaptability or versatility
If a fixed mask is used in sound decomposition, then the process is simple, but it cannot adapt to variable noise characteristics
Solution Approach 1:
The patent applies preliminary action by pre-processing the sound data to identify noise characteristics and pre-computing mask values based on these characteristics. This preliminary analysis enables the mask to be generated in advance with appropriate adaptability to the specific noise present, reducing the need for complex real-time adjustments while maintaining high adaptability
3Reliability
If aggressive noise removal is applied, then noise is reduced more effectively, but desired sound components may be compromised
Solution Approach 1:
The patent applies local quality by creating a mask with different values for different time-frequency bins based on the specific characteristics of non-stationary noises detected in each region. This allows aggressive noise removal in regions where noise is prominent while preserving desired sound components in regions where they are dominant, thereby maintaining overall sound integrity while achieving effective noise removal
Data Source
AI summary
Variable sound decomposition masking techniques are described. In one or more implementations, a mask is generated that incorporates a user input as part of the mask, the user input is usable at least in part to define a threshold that is variable based on the user input and configured for use in performing a sound decomposition process. The sound decomposition process is performed using the mask to assign portions of sound data to respective ones of a plurality of sources of the sound data.