Wind Noise Detection via Spectral Distribution Differences

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing wind noise detection methods in digital signal processing struggle to accurately differentiate wind noise from non-wind sounds due to phase differences and spectral level variations across microphones, leading to false detections, especially at higher frequencies and with increased microphone spacing.

Innovation Solution

A method and device that process digitized microphone signals by calculating the difference between the distributions of signal sample magnitudes from multiple microphones, ignoring phase differences and focusing on the unique impact of wind noise on signal distributions, with a detection threshold to indicate wind noise presence, allowing for selective wind noise reduction in specific frequency sub-bands.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If microphone spacing is increased to improve wind noise detection capability, then wind noise detection sensitivity is improved, but false detection of non-wind sounds as wind noise increases

Engineering Contradiction:
Improvewind noise detection sensitivityVSAvoidfalse detection rate
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent divides the audio spectrum into multiple frequency sub-bands and processes each sub-band separately. This segmentation allows the system to apply different detection thresholds and methods to different frequency ranges, reducing false detections while maintaining wind noise detection sensitivity. The correlation calculation is performed independently for each sub-band, enabling targeted wind noise suppression without affecting the entire spectrum.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent changes the detection parameter from raw signal correlation to spectral correlation. By transforming the signals into the frequency domain and calculating correlation in the spectral domain, the system can distinguish wind noise from non-wind sounds more effectively. The spectral correlation coefficient is computed as a function of frequency, allowing the system to adapt to different operating conditions and microphone configurations.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If spectral correlation method is used to detect wind noise, then detection accuracy across frequency spectrum is improved, but computational complexity increases

Engineering Contradiction:
Improvedetection accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent applies partial action by focusing computational resources only on frequency sub-bands where wind noise is actually present. Instead of processing the entire spectrum uniformly, the system identifies and processes only the relevant sub-bands, reducing overall computational complexity while maintaining detection accuracy. The correlation calculation is performed selectively based on the detected wind noise characteristics in each sub-band.

Inventive Principle:
Principle #16Partial or excessive action

3Object-affected harmful factors

If wind noise suppression is applied across the entire audible spectrum, then wind noise reduction is maximized, but naturalness of audio signal deteriorates

Engineering Contradiction:
Improvewind noise reductionVSAvoidaudio naturalness
Core Design Contradiction:
Object-affected harmful factorsVSLoss of information

Solution Approach 1:

The patent applies local quality by suppressing wind noise only in the specific frequency sub-bands where it is detected, rather than applying uniform suppression across the entire spectrum. The system calculates spectral correlation coefficients for each sub-band and applies suppression only where the correlation indicates wind noise presence. This localized approach preserves the natural characteristics of audio signals in sub-bands unaffected by wind noise.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP3172906B1Method and apparatus for wind noise detection
Publication Date: 2019.04.03 CIRRUS LOGIC INT SEMICON LTD
  • EP3172906B1 patent drawingFigure 1~2
  • EP3172906B1 patent drawingFigure 3~4
  • EP3172906B1 patent drawingFigure 5~6

AI summary

Processing digitized microphone signal data in order to detect wind noise. A first signal and a second signal are obtained from at least one microphone. The first and second signals reflect a common acoustic input, and are either temporally distinct or spatially distinct, or both. The first signal is processed to determine a first distribution of the samples of the first signal. The second signal is processed to determine a second distribution of the samples of the second signal. A difference between the first distribution and the second distribution is calculated. If the difference exceeds a detection threshold, an indication is output that wind noise is present.