Noise-Based Interest Point Density Pruning for Audio Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Audio matching systems face scalability issues due to the large storage requirements and computational demands of reference fingerprints with numerous interest points, which can be exacerbated by noise and distortion in audio samples, limiting the system's ability to accurately identify audio samples.
Innovation Solution
Implementing noise-based interest point density pruning, where the number of interest points in reference fingerprints is reduced while increasing the number in audio sample fingerprints based on the noise level, using a noise detection component, interest point detection component, density component, and fingerprint component to generate subsets of interest points and fingerprints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of interest points in reference fingerprints is increased to improve matching accuracy, then the identification accuracy is improved, but the storage requirements and computational demands increase significantly
Solution Approach 1:
The patent applies local quality by differentiating the treatment of interest points based on their temporal density characteristics. Interest points are selectively pruned or retained based on their local density properties, allowing the system to maintain high matching accuracy for critical regions while reducing storage requirements for less informative regions. This selective approach enables the system to achieve optimal balance between accuracy and storage efficiency.
Solution Approach 2:
The patent utilizes parameter changes by dynamically adjusting the density threshold and pruning parameters based on the audio sample characteristics and noise level. The system modifies the number of interest points retained in fingerprints based on detected noise conditions, allowing adaptive optimization of the balance between matching accuracy and computational/storage requirements for each specific audio sample.
2Measurement precision
If the number of interest points in reference fingerprints is increased to improve matching accuracy, then the identification accuracy is improved, but the computational demands and processing time increase
Solution Approach 1:
The patent applies local quality by differentiating the treatment of interest points based on their temporal density characteristics. Interest points are selectively pruned or retained based on their local density properties, allowing the system to maintain high matching accuracy for critical regions while reducing computational demands for less informative regions.
Solution Approach 2:
The patent applies partial action by retaining only a subset of interest points rather than all detected interest points. The system uses density-based pruning to select a representative subset that provides sufficient information for accurate matching while significantly reducing the computational load compared to processing all interest points.
3Adaptability or versatility
If noise-based interest point density pruning is applied to reduce storage requirements, then the scalability is improved, but the matching accuracy may be compromised in noisy environments
Solution Approach 1:
The patent implements feedback by using noise level detection to dynamically adjust the interest point pruning strategy. The system continuously monitors the audio sample characteristics and adjusts the density threshold and pruning parameters accordingly, ensuring that matching accuracy is maintained while scalability is improved through adaptive pruning.
Solution Approach 2:
The patent utilizes parameter changes by dynamically adjusting the density threshold and pruning parameters based on the detected noise level. In noisy environments, the system modifies the number of interest points retained in fingerprints to maintain optimal balance between matching accuracy and computational/storage requirements.
Data Source
AI summary
Systems and methods for noise based interest point density pruning are disclosed herein. The systems include determining an amount of noise in an audio sample and adjusting the amount of interest points within an audio sample fingerprint based on the amount of noise. Samples containing high amounts of noise correspondingly generate fingerprints with more interest points. The disclosed systems and methods allow reference fingerprints to be reduced in size while increasing the size of sample fingerprints. The benefits in scalability do not compromise the accuracy of an audio matching system using noise based interest point density pruning.