Combined Magnitude Phase Spectrogram Signal Processing

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional signal processing techniques using magnitude spectrograms suffer from limitations due to fixed resolution in both frequency and time domains, leading to the loss of phase information and reduced performance in signal analysis.

Innovation Solution

The generation of combined spectrograms that incorporate both magnitude and phase data, achieved through time-windowed transforms, phase correction, and phase differencing, allowing for enhanced signal analysis by providing multi-tiered frequency and time resolutions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If conventional discrete Fourier transform (STFT) is used to generate magnitude spectrogram, then the spectrogram can be easily visualized and interpreted, but phase information is lost and resolution is fixed in both frequency and time domains

Engineering Contradiction:
Improvevisual interpretationVSAvoidphase information
Core Design Contradiction:
Ease of operationVSLoss of information

Solution Approach 1:

The patent segments the spectral information into two distinct components: magnitude spectrogram and phase spectrogram. By separating these components, the system can process and preserve phase information independently while maintaining the visual interpretability of magnitude information. The phase spectrogram is generated separately and then combined with the magnitude spectrogram to reconstruct the full signal representation.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent adds a new dimension to the traditional spectrogram by incorporating phase information as a separate channel or layer. Instead of discarding phase data, the system creates a multi-dimensional representation where both magnitude and phase coexist, enabling neural networks to utilize phase information for improved signal analysis while maintaining compatibility with standard visualization techniques.

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

2Measurement precision

If resolution in frequency domain is increased, then frequency analysis precision is improved, but time domain resolution is reduced

Engineering Contradiction:
Improvefrequency resolutionVSAvoidtime resolution
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent employs dynamic time-windowing strategies where the window size and overlap parameters can be adjusted based on the specific analysis requirements. By making the transform parameters adaptive rather than fixed, the system can optimize the balance between frequency and time resolution for different signal characteristics and analysis goals, allowing high frequency resolution when needed while maintaining acceptable time resolution.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system allows dynamic adjustment of key parameters including window size, hop length, and transform length. These parameter changes enable the spectrogram generation to adapt to different resolution requirements, permitting high frequency resolution for stationary signals while using shorter windows for transient analysis, thus overcoming the fixed resolution limitation of conventional STFT.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS12104955B2Device to process sample using a time-windowed transform function to generate spectral data and to use combined magnitude and phase spectrograms
Publication Date: 2024.10.01 THE BOEING CO
  • US12104955B2 patent drawing
  • US12104955B2 patent drawing
  • US12104955B2 patent drawing

AI summary

A method of signal processing includes receiving samples of a signal and processing the samples using a time-windowed transform function to generate spectral data corresponding to each time window. The method includes generating first spectrogram data based on magnitudes of the spectral data and generating second spectrogram data based on phase differences of the spectral data. The method further includes combining the first spectrogram data and the second spectrogram data to generate a combined spectrogram and processing the combined spectrogram to generate output.