Audio Information Structuring for Compact High-Fidelity Compression

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current audio representation technologies fail to provide a compact, yet informative, format for digital audio, as existing methods either lose informativeness at low bit rates or require significant computational resources, making them unsuitable for efficient storage and transmission of complex musical and natural sounds.

Innovation Solution

The use of highly-organized structures and parameters as audio signal metrics, known as primotrons, which create complex, compact, information-rich structures through successive growth and manipulation, allowing for reduced data file size while maintaining information quality, utilizing techniques like phaseless coding, entropy coding, and affine transforms to achieve bandwidth and bit rate reduction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Loss of information

If PCM format is used for audio representation, then information completeness is improved, but data file size increases significantly

Engineering Contradiction:
Improveinformation completenessVSAvoiddata file size
Core Design Contradiction:
Loss of informationVSQuantity of substance

Solution Approach 1:

The patent extracts and removes redundant information from audio signals by analyzing natural sound redundancy patterns. It identifies and eliminates repetitive temporal, spectral, and psychoacoustic elements that do not contribute to perceptual quality, thereby reducing data file size while preserving essential information content.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent transforms audio representation from time-domain samples to frequency-domain parameters through spectral analysis. It converts raw audio data into perceptually-relevant parameters such as spectral envelopes, harmonic structures, and temporal evolution characteristics, achieving compact representation with preserved information value.

Inventive Principle:
Principle #35Parameter changes

2Quantity of substance

If psychoacoustic models are used for data reduction, then data file size decreases, but information completeness is lost

Engineering Contradiction:
Improvedata file sizeVSAvoidinformation completeness
Core Design Contradiction:
Quantity of substanceVSLoss of information

Solution Approach 1:

The patent applies different compression strategies to different portions of the audio spectrum based on local perceptual importance. It preserves high-resolution representation in critical frequency regions while applying aggressive compression in less important regions, optimizing the balance between file size and information retention at localized spectral positions.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent employs dynamic adjustment of compression parameters based on temporal and spectral context. It adapts the level of detail preserved in different time-frequency regions according to local signal characteristics, such as transient content, harmonic structure, and masking conditions, thereby maintaining information completeness where needed while reducing data elsewhere.

Inventive Principle:
Principle #15Dynamics

3Adaptability or versatility

If structured audio representations are used, then data organization is improved, but computational resources required increase

Engineering Contradiction:
Improvedata organizationVSAvoidcomputational resources
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent divides audio signals into segmented time-frequency units and processes each segment independently using standardized analysis procedures. This segmentation allows for efficient parallel processing and reduces computational complexity by breaking down the overall transformation into manageable, repetitive operations that can be optimized and cached.

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP2745211B8Multi-structural, multi-level information formalization and structuring method, and associated apparatus
Publication Date: 2019.07.17 GENERAL HARMONICS INTERNATIONAL INC

AI summary

Systems and methods are provided for structuring information, including analyzing an original digital information file (DIF) to determine an information quantity (IQ) and an information value (IV). An initial manipulation process is applied to the original DIF to form a first resulting DIF, and a subsequent manipulation process applied to the first resulting DIF to form a second resulting DIF, wherein each manipulation process removes at least one element of the processed DIF and/or represents a combination of elements with a representative element and a first indicia of an interrelationship between the representative element and one or more elements in the combination, to reduce the IQ of the processed DIF, while retaining the IV thereof within a threshold. Manipulation processes are successively applied to the previously resulting DIF until successive applications do not achieve a threshold reduction in IQ. The last resulting DIF has a primary structure with a reduced IQ and an IV within the threshold of the original IV.