Min/Max Filter for Audio Matching False Positives
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In large-scale audio matching systems, the computational expense of generating projections of hits for potential matches is high due to the need to filter out numerous false positive matches, which are prevalent when comparing audio samples to millions of reference samples.
Innovation Solution
A min/max filter is applied to dynamically filter out false positive matches by using a match threshold and window size parameters to assess the temporal alignment of hits between audio and reference samples, reducing the number of potential matches that require further inspection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a large number of reference samples are compared to identify audio samples, then the accuracy of matching is improved, but the computational cost increases significantly
Solution Approach 1:
The patent applies preliminary filtering actions before the main matching process. By using the min/max filter to pre-process reference samples and eliminate obvious false positives before generating hit projections, the system reduces the number of samples requiring computationally expensive operations, thereby lowering overall computational cost while maintaining matching accuracy
Solution Approach 2:
The matching process is segmented into multiple stages: initial hit generation, min/max filtering stage, and projection generation stage. This segmentation allows the system to apply different computational strategies to different subsets of data, processing only promising candidates through the most expensive operations
2Reliability
If all reference samples containing hits are processed to generate projections, then the reliability of match determination is improved, but the processing time increases
Solution Approach 1:
The min/max filter performs preliminary action by evaluating reference samples before projection generation. It uses min/max value comparisons to quickly eliminate false positives, ensuring that only samples with genuine matching potential proceed to the time-consuming projection generation step, thus maintaining reliability while reducing processing time
Solution Approach 2:
The system applies different processing quality levels to different reference samples. Samples passing the min/max filter receive full processing with projection generation, while those failing the filter receive minimal processing and are discarded, optimizing the allocation of computational resources based on local sample characteristics
Data Source
AI summary
Systems and methods are provided herein relating to audio matching. Descriptors can be generated for a received audio signal and matched with reference descriptors. Potential matching reference samples can then be filtered based on, at least in part, a number of hits, a match threshold, and a window size. As more hits are accumulated for a reference sample, the more likely the reference sample is to pass through the filter. Eliminating potential false positive matches before performing more computational demanding matching algorithms can increase efficiency within an audio matching system.