Audio Fingerprint Clustering for Set Top Box Viewership Tracking

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

The fragmentation of TV viewership in national markets due to diverse communication channels makes it challenging to provide accurate and efficient viewership ratings for TV programs, as existing methods struggle to aggregate and combine viewership data from different content providers effectively.

Innovation Solution

A system and method that utilize audio fingerprints generated by set top boxes to identify TV programs, where these fingerprints are grouped and compared to determine corresponding TV program information, allowing for accurate viewership tracking across multiple channels and providers, using a computer server system that partitions and matches audio fingerprints with channel metadata to provide reliable viewership data.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Quantity of substance

If audio fingerprints are collected from multiple set top boxes across different communication channels, then viewership data coverage is improved, but data aggregation and matching accuracy deteriorates due to fragmentation across satellite, cable, phone line, and Internet providers

Engineering Contradiction:
Improveviewership data coverageVSAvoidviewership rating accuracy
Core Design Contradiction:
Quantity of substanceVSMeasurement precision

Solution Approach 1:

The system segments the viewership data collection process by having each set top box independently generate audio fingerprints from its local audio signal. Each STB processes its own audio stream through fingerprint generation, creating discrete, identifiable data units that can be aggregated from multiple sources without requiring centralized processing of raw audio signals.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Audio fingerprints serve as an intermediary element between diverse communication channels and the central server. Instead of directly comparing raw audio signals from different providers (satellite, cable, Internet), the system uses standardized fingerprint representations as a common language that enables accurate matching across heterogeneous sources while preserving measurement precision.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Power

If each set top box generates its own audio fingerprints locally, then server processing load is reduced, but the complexity of fingerprint generation and transmission increases at the client side

Engineering Contradiction:
Improveserver processing capacityVSAvoidset top box processing complexity
Core Design Contradiction:
PowerVSDevice complexity

Solution Approach 1:

The computationally intensive fingerprint generation process is extracted from the central server and relocated to individual set top boxes. Each STB locally extracts fingerprint features from its audio signal, transforming the server's role from active processor to passive aggregator, thereby significantly reducing server processing load while distributing computational complexity to client devices.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

Each set top box autonomously generates its own audio fingerprints without requiring server-side processing assistance. The STBs independently perform audio signal analysis, fingerprint extraction, and transmission, making the system self-service oriented at the client level and freeing server resources for aggregation and matching operations.

Inventive Principle:
Principle #25Self-service

3Productivity

If audio fingerprints are grouped by similarity to identify TV programs, then viewership aggregation efficiency is improved, but the computational complexity of comparing and clustering fingerprints increases

Engineering Contradiction:
Improveviewership data aggregation efficiencyVSAvoidfingerprint matching complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

Audio fingerprints are pre-processed and standardized before aggregation, with each STB generating fingerprints in a consistent format that facilitates efficient matching. The preliminary fingerprint generation at source includes feature extraction and normalization, reducing the computational burden during the clustering phase and enabling scalable aggregation across large numbers of set top boxes.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2868109B1Generating a sequence of audio fingerprints at a set top box
Publication Date: 2016.08.17 GOOGLE LLC
  • EP2868109B1 patent drawingFigure 1A
  • EP2868109B1 patent drawingFigure 1B
  • EP2868109B1 patent drawingFigure 2A

AI summary

A computer-implemented method for clustering audio fingerprints from different set top boxes into groups based on their similarities is disclosed. The computer server receives audio fingerprints generated at different set top boxes and partitions them into multiple groups such that audio fingerprints within a group correspond to a respective TV program being played at the respective set top boxes. Upon receipt of an audio fingerprint from a set top box, the computer server identifies a group of audio fingerprints that are substantially similar to the audio fingerprint, determines TV program information for the audio fingerprint in accordance with the channel metadata associated with the identified group of audio fingerprints, and returns the TV program information to the requesting set top box.