Spatial Audio Object Extraction via Subband Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for converting spatial audio formats like Ambisonics and Higher Order Ambisonics to object-based formats, such as Dolby's Atmos, often apply a broadband approach, leading to decreased discreteness in audio object directions and increased smearing, as they fail to capture panning behavior across frequency subbands.

Innovation Solution

A subband-based approach is employed to determine audio object signals by analyzing the input audio signal in frequency subbands, calculating mixing gains for each object location, and applying spatial mapping functions to generate output signals, which are then summed to produce clear and precise audio object channels with improved directionality.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If a broadband approach is used for format conversion from spatial format to object-based format, then the conversion process is simpler, but the discreteness of audio object directions decreases and smearing increases

Engineering Contradiction:
Improveconversion process complexityVSAvoidaudio object direction discreteness
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

The patent divides the broadband audio signal into multiple frequency subbands and processes each subband separately to determine object locations. This segmentation allows precise direction estimation in each frequency range while maintaining manageable processing complexity through systematic handling of subbands.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent transitions from a single broadband dimension to multiple frequency subband dimensions, enabling independent analysis of panning behavior across different frequency ranges. This dimensional expansion captures frequency-dependent spatial characteristics that broadband processing misses.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Productivity

If a broadband approach is used for format conversion, then processing is more efficient, but panning behavior across frequency subbands is not captured

Engineering Contradiction:
Improveprocessing efficiencyVSAvoidpanning behavior information
Core Design Contradiction:
ProductivityVSLoss of information

Solution Approach 1:

The audio signal is segmented into frequency subbands, allowing panning behavior to be captured in each subband while maintaining efficient processing through parallel or sequential subband handling. This preserves frequency-dependent spatial information that broadband processing would average out.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the processing parameter from broadband to subband frequency resolution, enabling capture of frequency-varying panning behavior. Each subband can have its own object location estimates, preserving the frequency-dependent spatial characteristics of the audio signal.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3622509B1Processing of a multi-channel spatial audio format input signal
Publication Date: 2021.03.24 DOLBY LABORATORIES LICENSING CORP
  • EP3622509B1 patent drawingFigure 1
  • EP3622509B1 patent drawingFigure 2
  • EP3622509B1 patent drawingFigure 3

AI summary

Apparatus, computer readable media and methods for processing a multi-channel, spatial audio format input signal. For example, one such method comprises determining object location metadata based on the received spatial audio format input signal; and extracting object audio signals based on the received spatial audio format input signal, wherein the extracting object audio signals based on the received spatial audio format input signal includes determining object audio signals and residual audio signals.