Audio Soft-Clipping Detection Using Mixture Model Fitting

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing digital audio signal processing techniques, such as histogram-based methods, are inefficient in detecting soft-clipping distortion, which occurs when audio signal amplitudes are compressed within the available range, leading to distortion that is not easily identifiable.

Innovation Solution

The use of a mixture model-based approach, specifically fitting a combination of a Laplacian distribution and two symmetrical Gaussian distributions to audio samples, allows for the detection and measurement of soft-clipping distortion through expectation maximization and machine learning algorithms, enabling accurate identification of soft-clipping distortion metrics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If histogram-based methods are used for soft-clipping detection, then the detection process is simple to implement, but the detection accuracy and efficiency are insufficient

Engineering Contradiction:
ImproveEase of implementationVSAvoidDetection accuracy
Core Design Contradiction:
Ease of manufactureVSMeasurement precision

Solution Approach 1:

The patent transforms the detection approach by changing the mathematical parameters from simple histogram counting to mixture model probability distributions. By fitting audio samples to a mixture of Laplacian and Gaussian distributions, the system achieves higher detection accuracy while maintaining computational feasibility through expectation maximization algorithms.

Inventive Principle:
Principle #35Parameter changes

Solution Approach 2:

The patent replaces the mechanical histogram-based counting method with a probabilistic mixture model approach. This substitution uses statistical theory and machine learning algorithms to detect soft-clipping distortion, achieving superior accuracy by modeling the underlying probability distributions of audio signals under different clipping conditions.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If mixture model-based approach is used, then detection accuracy is improved, but computational complexity increases

Engineering Contradiction:
ImproveDetection accuracyVSAvoidComputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent performs preliminary action by pre-defining the mixture model structure (combination of Laplacian and Gaussian distributions) and their parameter ranges before actual detection. The expectation maximization algorithm pre-computes optimal parameters during the fitting process, which reduces the computational burden during real-time detection and simplifies the overall system complexity.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The mixture model fitting process is self-adjusting through expectation maximization, where the algorithm automatically optimizes distribution parameters based on the input audio data characteristics. This self-service mechanism eliminates the need for manual parameter tuning and reduces system complexity by allowing the model to adapt automatically to different audio signals.

Inventive Principle:
Principle #25Self-service

3Productivity

If traditional histogram-based methods are used, then processing speed is maintained, but soft-clipping distortion cannot be accurately identified

Engineering Contradiction:
ImproveProcessing speedVSAvoidDetection reliability
Core Design Contradiction:
ProductivityVSReliability

Solution Approach 1:

The patent changes the detection parameters from discrete histogram bins to continuous probability distribution parameters. By using mixture model parameters (means, variances, mixing coefficients) that capture the essential characteristics of soft-clipped audio signals, the system achieves both high processing speed and reliable detection accuracy simultaneously.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3542363B1Mixture model based soft-clipping detection
Publication Date: 2020.02.05 GOOGLE LLC
  • EP3542363B1 patent drawingFigure 1
  • EP3542363B1 patent drawingFigure 2
  • EP3542363B1 patent drawingFigure 3

AI summary

Mixture model based soft-clipping detection includes receiving input audio samples, generating soft-clipping information indicating whether the input audio samples include soft-clipping distortion, and outputting the soft-clipping information. Generating the soft-clipping information includes fitting a mixture model to the input audio samples, wherein fitting the mixture model to the input audio samples includes generating a fitted mixture model, such that the fitted mixture model has fitted parameters, and evaluating a soft-clipping distortion metric based on the parameters of the fitted mixture model, wherein evaluating the soft-clipping distortion metric includes identifying a soft-clipping distortion value.