Audio Fingerprint Matching via Unordered Portion Comparison
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio fingerprint matching technologies face challenges in accurately identifying songs with noise, distortion, time-stretching, or time-squeezing, and consume significant processing resources, especially when dealing with unordered or unanticipated audio content.
Innovation Solution
The system performs unordered matching of audio fingerprints by generating spectral analysis-based fingerprints and comparing portions of known audio fingerprints against unknown audio fingerprints, using a probe to detect matches without temporal growth, and applying disorder and positional penalties to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If ordered comparison is used to match audio segments, then processing can be performed sequentially with lower memory requirements, but matching accuracy deteriorates when noise, distortion, or time-stretching is present
Solution Approach 1:
The audio segments are divided into multiple portions, and comparisons are performed between all portions of the unknown segment and all portions of the known segment simultaneously, rather than sequentially. This segmentation approach enables parallel processing while maintaining comprehensive matching accuracy across different audio conditions.
Solution Approach 2:
The matching process transitions from a one-dimensional sequential comparison (ordered) to a multi-dimensional simultaneous comparison (unordered). By comparing all portions across all segments at once, the system creates a comprehensive match score that accounts for various audio transformations, effectively adding dimensional complexity to improve reliability.
2Reliability
If unordered comparison of all portions is performed, then matching accuracy improves with noise and distortion tolerance, but processing resource consumption increases
Solution Approach 1:
The system merges the comparison results of all portion pairs into a single comprehensive match score. By combining the individual match scores from all comparisons between portions of unknown and known segments, the system achieves robust noise tolerance while consolidating processing outcomes efficiently.
Solution Approach 2:
The system creates multiple copies of the comparison logic, applying the same matching algorithm to all portions simultaneously. This parallel copying of the comparison process enables comprehensive audio matching with noise and distortion tolerance, while the structured approach allows for efficient resource utilization through systematic repetition of the matching algorithm.
3Measurement precision
If time-stretched or time-squeezed versions are matched using ordered comparison, then processing remains simple, but identification accuracy deteriorates
Solution Approach 1:
The matching system dynamically adapts to time-stretched or time-squeezed audio by performing unordered comparisons that do not assume fixed temporal relationships. The system flexibly matches portions based on spectral similarity rather than temporal position, enabling accurate identification of time-transformed audio versions through adaptive, non-sequential processing.
Data Source
AI summary
A computing system includes an interface configured to interface and communicate with an external media resource, a memory that stores operational instructions and a processing module operably coupled to the interface and to the memory, such that the processing module, when operable within the computing system based on the operational instructions, is configured to is configured to perform various operations. The computing system receives an audio fingerprint from the external medial resource, divides the audio file segment into a plurality of portions and compares the audio file segment with a plurality of stored audio segments. The computing system then determines whether a threshold number of the plurality of portions of a stored audio segment compare favorably with a threshold number of the plurality of portions of the audio file segment and based thereon determines whether the audio file segment and the stored audio segment correspond to alternate versions of a same song.