Audio Signal Objectification Through Neural Source Separation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Conventional audio processing techniques struggle with converting non-object-based audio content into object-based audio content, which is laborious and expensive, and often result in audio signals being decoded to a channel-based format, limiting the ability to manipulate individual audio sources.

Innovation Solution

Utilizing machine learning models, particularly deep neural networks, trained through supervised learning on short audio snippets, to extract and separately manipulate individual audio sources in real-time or near real-time, enabling dynamic control of audio sources such as spatial position and volume.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional audio processing techniques are used to convert non-object-based audio content into object-based audio content, then the conversion process becomes laborious and expensive, but the ability to manipulate individual audio sources is improved

Engineering Contradiction:

Improveability to manipulate individual audio sourcesVSAvoidconversion process complexity

Core Design Contradiction:

Ease of operationVSDevice complexity

Solution Approach 1:

The patent replaces manual, mechanical audio conversion processes with an automated machine learning system. The neural network automatically separates audio sources and enables individual manipulation without requiring expert intervention in complex conversion processes, thus resolving the contradiction between ease of operation and device complexity

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs self-service by automatically analyzing and separating audio sources using the trained neural network model. The audio processing system handles the complex conversion task autonomously without requiring external expert intervention, making object-based manipulation accessible while reducing operational complexity

Inventive Principle:

Principle #25Self-service

2Adaptability or versatility

If non-object-based audio content is converted into object-based audio content using traditional methods, then individual audio sources can be manipulated independently, but the process is laborious and expensive requiring expert involvement

Engineering Contradiction:

Improveindependent manipulation of audio sourcesVSAvoidconversion process accessibility

Core Design Contradiction:

Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The patent substitutes manual expert processing with an automated neural network system that performs audio source separation. This automation enables independent manipulation of audio sources while eliminating the need for expert involvement, thus improving both adaptability and ease of manufacture

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system changes the fundamental parameter of audio representation from mixed channel-based signals to separated object-based sources through the neural network's automatic separation process. This parameter transformation enables versatile manipulation while making the process accessible without expert intervention

Inventive Principle:

Principle #35Parameter changes

3Reliability

If audio signals are decoded to channel-based format, then compatibility with legacy systems is maintained, but the ability to separately manipulate individual audio sources is limited

Engineering Contradiction:

Improvesystem compatibilityVSAvoidmanipulation flexibility

Core Design Contradiction:

ReliabilityVSEase of operation

Solution Approach 1:

The patent implements a dynamic system that can operate in multiple modes: maintaining channel-based compatibility for legacy systems while simultaneously providing object-based separation for enhanced manipulation. The neural network enables flexible switching between these modes, resolving the contradiction between reliability and ease of operation

Inventive Principle:

Principle #15Dynamics

Solution Approach 2:

The system achieves universality by being capable of both channel-based playback for compatibility and object-based separation for manipulation flexibility. The same neural network infrastructure supports both operational modes, allowing the system to adapt to different requirements without sacrificing either compatibility or manipulation capability

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250322835A1Objectification of audio signals

Publication Date: 2025.10.16 BANG & OLUFSEN AS

AI summary

Techniques for dynamic audio objectification are described. Embodiments include providing a first audio snippet from an audio signal to a machine learning model trained based on audio snippets labeled with an audio source and receiving, from the machine learning model, a subset of the first audio snippet that is associated with the audio source. Embodiments include, after playing the reconstituted first audio snippet, receiving a changed configuration relating to the audio source. Embodiments include providing a second audio snippet from the audio signal to the machine learning model and receiving, from the machine learning model, a subset of the second audio snippet that is associated with the audio source. Embodiments include playing a reconstituted second audio snippet based on the subset of the second audio snippet and the changed configuration, wherein an audibly perceptible parameter of the audio source is changed in the reconstituted second audio snippet.