Audio Matching via Chromaprint Segmentation and Pruning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio matching technologies face inefficiencies in identifying duplicate audio tracks due to resource-intensive comparisons of large audio fingerprints, particularly when dealing with slight duration differences, time shifts, and quality variations.
Innovation Solution
A method and system that utilize chromaprints with a fast candidate selection routine, involving the generation of short and long audio fingerprints, where candidate tracks are selected based on matching chroma words and validated through bit-by-bit comparison, using a pruning index for short fingerprints and a validation index for long fingerprints, to efficiently match incoming audio tracks with indexed tracks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If comparison of larger portions of audio fingerprints is performed to identify duplicate tracks, then matching accuracy is improved, but computational resource consumption increases
Solution Approach 1:
The patent segments the audio fingerprint comparison process into two distinct stages: (1) a fast candidate selection stage that compares only short audio fingerprints (first portions) to quickly identify potential matches, and (2) a validation stage that performs comprehensive comparison of long audio fingerprints (second portions) only on the candidate tracks identified in stage 1. This segmentation allows the system to maintain high matching accuracy through thorough validation while dramatically reducing computational resource consumption by limiting extensive comparisons to only necessary cases.
Solution Approach 2:
The patent performs preliminary action by generating and storing short audio fingerprints for all tracks in the database in advance, and uses these pre-computed short fingerprints to quickly filter and identify candidate duplicate tracks before performing the more resource-intensive long fingerprint comparison. This preliminary filtering action significantly reduces the number of tracks that require full validation, thereby lowering overall computational resource consumption while maintaining accurate duplicate detection.
2Measurement precision
If comprehensive audio fingerprint comparison is performed on all indexed tracks, then duplicate identification accuracy is improved, but processing time increases
Solution Approach 1:
The patent divides the duplicate identification process into two time-efficient segments: a rapid candidate selection phase using short audio fingerprints that quickly narrows down potential duplicates, and a focused validation phase using long audio fingerprints that confirms matches only among the small subset of candidates. This segmentation ensures high duplicate identification accuracy through thorough validation while minimizing processing time by avoiding comprehensive comparison of all indexed tracks.
Solution Approach 2:
The system performs preliminary action by pre-computing and indexing short audio fingerprints for all tracks, enabling rapid initial filtering to identify candidate duplicates within a small time window. This preliminary step dramatically reduces the candidate pool size before the validation phase, ensuring that time-consuming long fingerprint comparisons are performed on only a few tracks rather than the entire database, thus maintaining high accuracy while minimizing total processing time.
3Productivity
If short audio fingerprints with chroma words are used for candidate selection, then processing speed is improved, but matching precision may be reduced
Solution Approach 1:
The patent applies segmentation by using short audio fingerprints with chroma words exclusively for the candidate selection phase where high processing speed is critical, while reserving long audio fingerprints for the validation phase where high matching precision is essential. This segmentation allows the system to leverage the speed advantage of short fingerprints with chroma words for filtering, while ensuring final match accuracy through comprehensive comparison with long fingerprints, thus resolving the trade-off between processing speed and matching precision.
Solution Approach 2:
The short audio fingerprint with chroma words acts as an intermediary that enables fast candidate selection without being the final decision-maker. It serves as a efficient filtering mechanism that identifies potential matches, but the actual matching precision is determined by the subsequent validation using long audio fingerprints. This intermediary role allows the system to benefit from the processing speed of chroma word-based comparison while maintaining high precision through the final validation step.
Data Source
AI summary
A computer-implemented method of matching of a first incoming audio track with an indexed audio track, the method executable at a server, the method comprising: selecting the indexed audio track as a candidate audio track from a plurality of indexed audio tracks; validating the candidate audio track against the first audio track.