Audio Content Selection for Subjective Preference Judgement

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for subjective preference judgments in audio processing are hindered by measurement noise due to inconsistencies in user responses and the need for large datasets, with manual content selection being time-consuming and laborious, and are challenged by the issue of content licensing when moving from laboratory settings to consumer devices.

Innovation Solution

A system that defines a user-specific media corpus based on consumption history or preferred genres, generates candidate media segments, derives signal characteristics, and processes them with different algorithms to reduce noise, allowing users to provide preference judgments on processed signals, with machine learning to optimize the selection process.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If random selections of content from a corpus of audio are used to capture average preference rating, then generic application to all content types is achieved, but a large number of user responses are required to identify consistent preference pattern

Engineering Contradiction:

Improvegeneric application to all content typesVSAvoidnumber of user responses

Core Design Contradiction:

Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent changes the parameters of content selection by moving from random selection to selection based on signal statistics (spectral content, temporal dynamics, signal level). This allows the system to identify consistent preference patterns with fewer user responses while maintaining adaptability to different content types through the use of multiple content segments with varying characteristics

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The system performs preliminary analysis of audio content to identify segments with specific signal statistics before presenting them to users. This preliminary action of selecting content based on measured characteristics reduces the number of responses needed, as each carefully selected segment provides more informative data about user preferences

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If a variety of content is presented to users, then coverage of different content types is improved, but measurement noise from individual responses is exacerbated

Engineering Contradiction:

Improvecoverage of different content typesVSAvoidmeasurement noise in preference ratings

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent segments audio content into multiple distinct segments with different signal statistics (different spectral content, temporal dynamics, signal levels). By presenting multiple segments rather than a variety of different audio pieces, the system maintains coverage of different content characteristics while reducing measurement noise, as each segment provides a controlled comparison basis

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The system ensures homogeneity in the presentation format and comparison structure across all content segments. Each segment is processed with the same signal processing strategies and presented in a consistent manner, which reduces measurement noise while still allowing assessment across different content types through the varied signal statistics of the segments

Inventive Principle:

Principle #33Homogeneity

3Manufacturing precision

If manual assessment of audio segment candidates is performed, then content selection criteria can be evaluated, but the process is time consuming and laborious

Engineering Contradiction:

Improvecontent selection criteria evaluationVSAvoidtime for content selection process

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical assessment with automated computer-based measurement and selection of audio segments. Signal processing algorithms automatically evaluate content candidates based on signal statistics (spectral centroid, spectral flux, zero-crossing rate, etc.), eliminating the time-consuming manual process while maintaining precise evaluation of selection criteria

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-assessment of audio content by automatically measuring signal statistics and selecting appropriate segments without human intervention. The content selection process serves itself through automated analysis, reducing both time and labor requirements while maintaining objective evaluation of selection criteria

Inventive Principle:

Principle #25Self-service

Data Source

PatentEP3944100A1Method of selecting a suitable content for subjective preference judgement

Publication Date: 2022.01.26 MIMI HEARING TECHNOLOGIES GMBH

Figure 1a
Figure 1b
Figure 2

AI summary

The present disclosure relates to a method for processing media content. The method comprises receiving a definition of a user specific media corpus, the media corpus comprising a plurality of media files; ranking candidate media segments generated from the media corpus according to one or more signal characteristics derived from the candidate media segments; processing the media signals of a subset of the ranked candidate media segments with different signal processings; presenting the differently processed media signals to a user for a user preference judgement; and determining a preferred signal processing based on user responses to the presentations.