Audio Object Clustering Using Perceptual Importance Metrics

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing audio signal processing technologies face challenges in efficiently compressing object-based audio data for playback systems, particularly in limited bandwidth scenarios, as traditional clustering methods based on spatial proximity alone fail to optimize the perceived quality of complex content with many active objects.

Innovation Solution

The method involves grouping audio objects based on their perceptual importance, determined by factors like loudness and content type, and using error thresholds to minimize spatial errors, allowing for dynamic clustering and de-clustering of objects, and distributing object signals across clusters using panning techniques.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If audio objects are clustered based on spatial proximity alone, then the clustering process is simple and fast, but the perceived quality degrades in complex content with many sparsely distributed objects

Engineering Contradiction:

Improveclustering processing efficiencyVSAvoidspatial quality accuracy

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The patent changes the clustering parameters from purely spatial coordinates to include perceptual importance metrics such as loudness, content type, and signal-to-noise ratio. This allows objects to be clustered based on their perceptual significance rather than just spatial proximity, improving spatial quality accuracy for complex content while maintaining processing efficiency through defined importance thresholds

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent creates a composite clustering criterion that combines multiple parameters (spatial position, loudness, content type, SNR) into a unified perceptual importance metric. This composite approach allows the system to leverage both simple spatial clustering for efficient cases and perceptual-based clustering for complex cases, achieving high accuracy without sacrificing processing efficiency

Inventive Principle:

Principle #40Composite materials

2Quantity of substance

If the number of output clusters is restricted to reduce bandwidth, then transmission efficiency improves, but spatial quality degrades due to significant spatial errors

Engineering Contradiction:

Improvedata transmission volumeVSAvoidspatial position accuracy

Core Design Contradiction:

Quantity of substanceVSManufacturing precision

Solution Approach 1:

The patent changes the clustering objective from minimizing spatial error to minimizing perceptual error. By using perceptual importance as a weighting factor, the system can restrict the number of output clusters for bandwidth efficiency while maintaining spatial quality for perceptually important objects. Objects with low perceptual importance can be clustered more aggressively without degrading overall perceived quality

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent applies different clustering strictness levels to different objects based on their perceptual importance. Perceptually important objects (high loudness, critical content types) maintain accurate spatial positioning even when the number of clusters is restricted, while less important objects are clustered more aggressively. This local quality approach ensures bandwidth efficiency without sacrificing spatial quality where it matters most

Inventive Principle:

Principle #3Local quality

3Manufacturing precision

If all audio objects are transmitted individually, then spatial quality is maintained, but bandwidth requirements become prohibitively high for mobile and broadcast distribution

Engineering Contradiction:

Improvespatial fidelityVSAvoidbandwidth consumption

Core Design Contradiction:

Manufacturing precisionVSQuantity of substance

Solution Approach 1:

The patent merges spatially proximate and perceptually similar objects into clusters, transmitting a single set of parameters for the entire cluster rather than individual parameters for each object. This significantly reduces bandwidth consumption for mobile and broadcast distribution while maintaining spatial fidelity for perceptually important objects. The merging is performed intelligently based on perceptual importance to preserve quality

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent applies partial clustering based on perceptual importance thresholds. Not all objects are clustered to the same degree - only those below certain importance thresholds are merged. This partial action approach reduces bandwidth consumption sufficiently for practical distribution while preserving spatial fidelity for objects that require it, achieving a practical compromise rather than extreme compression

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS9805725B2Object clustering for rendering object-based audio content based on perceptual criteria

Publication Date: 2017.10.31 DOLBY LABORATORIES LICENSING CORP

AI summary

Embodiments are directed a method of rendering object-based audio comprising determining an initial spatial position of objects having object audio data and associated metadata, determining a perceptual importance of the objects, and grouping the audio objects into a number of clusters based on the determined perceptual importance of the objects, such that a spatial error caused by moving an object from an initial spatial position to a second spatial position in a cluster is minimized for objects with a relatively high perceptual importance. The perceptual importance is based at least in part by a partial loudness of an object and content semantics of the object.