Digital Signature Generation for Media Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing media monitoring technologies face challenges in accurately identifying media information, such as audio streams, due to inefficiencies in generating and matching digital signatures, particularly in television and radio audience metering applications, where differences in data rates between monitored and reference signatures complicate matching processes.
Innovation Solution
The method involves generating digital signatures by analyzing audio spectrum attributes, computing decision metrics for signal bands, and assigning signature bits based on these metrics, which can be derived from spectral representations or transforms like the wavelet transform, allowing for the comparison of monitored and reference signatures to identify media information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If digital signatures are generated by analyzing audio spectrum attributes and computing decision metrics for signal bands, then media information identification accuracy is improved, but device complexity increases
Solution Approach 1:
The audio spectrum is divided into multiple signal bands, with decision metrics computed for each band independently. This segmentation allows the system to capture detailed spectral characteristics improving identification accuracy while organizing complexity into manageable modular components that can be processed systematically
Solution Approach 2:
Digital signatures serve as an intermediary representation that captures essential audio spectrum attributes without requiring direct comparison of full audio streams. The signature generation process acts as a mediator that transforms complex audio data into compact, comparable forms, reducing the computational burden while maintaining identification accuracy
2Measurement precision
If monitored and reference signatures are compared to identify media information, then media tracking accuracy is improved, but data processing time increases
Solution Approach 1:
Instead of comparing entire audio streams, the system creates compact digital signature copies that represent the essential characteristics of the audio content. These signature copies can be rapidly compared and matched, maintaining high identification accuracy while dramatically reducing processing time and computational resources required
Solution Approach 2:
Digital signatures are pre-computed from audio spectrum attributes and stored for later comparison. This preliminary processing allows the system to have ready-made reference signatures available for rapid matching against monitored signatures, reducing real-time processing time while maintaining accurate media tracking
3Stability of the object's composition
If signature bits are assigned based on decision metrics from spectral representations, then signature robustness is improved, but manufacturing precision requirements increase
Solution Approach 1:
The system transforms audio spectrum attributes into decision metrics through defined mathematical relationships, creating signature bits that are robust to variations in audio encoding and transmission. By changing parameters from raw spectral values to normalized decision metrics, the system achieves greater stability while managing precision requirements through systematic transformation
Data Source
AI summary
Methods and apparatus for characterizing media are described. A disclosed example apparatus includes a transformer, a decision metric processor, a signature determiner, and a processor to implement the transformer, the decision metric processor, and/or the signature determiner. The example transformer is to convert at least a portion of a block of audio into a frequency domain representation including a plurality of frequency components. The example decision metric processor is to: define a band of the frequency components; determine a difference in energy between a first convolution of a first complex vector with a first group of frequency bins in the band and a second convolution of a second complex vector with a second group of frequency bins in the band; and determine a decision metric for the band based on the difference. The example signature determiner is to determine a signature based on a value of the decision metric.