signalFrequencyFeatureExtractor

Streamline signal frequency feature extraction

Since R2021b

Description

Use signalFrequencyFeatureExtractor to extract frequency-domain features from a signal. You can use the extracted features to train a machine learning model or a deep learning network.

Creation

Syntax

sFE = signalFrequencyFeatureExtractor

sFE = signalFrequencyFeatureExtractor(Name=Value)

Description

sFE = signalFrequencyFeatureExtractor creates a signalFrequencyFeatureExtractor object with default property values.

sFE = signalFrequencyFeatureExtractor(Name=Value) specifies nondefault property values of the signalFrequencyFeatureExtractor object. For example, signalFrequencyFeatureExtractor(FrameSize=30,FrameOverlapLength=6) divides the signal into overlapping 30-sample frames and extracts features from each frame.

Properties

Main Properties

FrameSize — Number of samples per frame

positive integer

Number of samples per frame, specified as a positive integer. The object divides the signal into frames of the specified length and extracts features for each frame. If you do not specify FrameSize, or if you specify FrameSize as empty, the object extracts features for the whole signal.

Data Types: single | double

FrameRate — Number of samples between start of frames

positive integer

Number of samples between the start of frames, specified as a positive integer. The frame rate determines the distance in samples between the starting points of frames. If you specify FrameRate, then you must also specify FrameSize. If you do not specify FrameRate or FrameOverlapLength, then the object assumes FrameRate to be equal to FrameSize. You cannot specify FrameRate and FrameOverlapLength simultaneously.

Data Types: single | double

FrameOverlapLength — Number of overlapping samples between consecutive frames

positive integer

Number of overlapping samples between consecutive frames, specified as a positive integer. FrameOverlapLength must be less than or equal to the frame size. If you specify FrameOverlapLength, then you must also specify FrameSize. You cannot specify FrameOverlapLength and FrameRate simultaneously.

Data Types: single | double

SampleRate — Sample rate

[] (default) | positive scalar

Input sample rate, specified as a positive scalar in hertz.

If you do not specify SampleRate, the extract function of the object assumes the signal sampling rate as 2π Hz.

Data Types: single | double

FeatureFormat — Format of generated signal features

"matrix" (default) | "table"

Format of the signal features generated by the extract function, specified as one of these:

• "matrix" — Columns correspond to feature values.

• "table" — Each table variable corresponds to a feature value.

Note

You can generate features for multiple signals at once by specifying a datastore object input in the extract function. In this case, extract returns a cell array where each member corresponds to a feature matrix or table from a signal member of the datastore. The format of the generated features in each member follows the format specified in FeatureFormat.

Data Types: char | string

IncompleteFrameRule — Rule to handle incomplete frames

"drop" (default) | "zeropad"

Rule to handle incomplete frames, specified as one of these:

• "drop" — Drop the incomplete frame and do not use it to compute features.

• "zeropad" — Zero-pad the incomplete frame and use it to compute features.

This rule applies when the current frame size is less than the specified FrameSize property.

Data Types: char | string

ScalarizationMethod — Methods to convert feature vectors to scalar values

frequencyScalarFeatureOptions object

Since R2024b

Methods to convert feature vectors to scalar values, specified as a frequencyScalarFeatureOptions object.

You can specify methods to extract scalar values from the Features to Extract. Specify scalarization methods for the feature extractor object by using the ScalarizationMethod name-value argument or the setScalarizationMethods function.

• If you specify ScalarizationMethod, the signalFrequencyFeatureExtractor object returns the corresponding scalar values for each feature vector using the scalarization method.

  To convert a feature vector to scalar feature values:

  • You must enable the feature for extraction by setting the feature name in the signalFrequencyFeatureExtractor object to true.

  • You must specify the desired scalarization methods for each feature name using a cell array of character vectors or a string array and store the information in a frequencyScalarFeatureOptions object.

  After that, the extract function:

  • Extracts the vectors corresponding to each enabled feature.

  • Takes the list of scalarization methods compiled by the object and for each method computes the corresponding scalar value.

  • Concatenates the vector features and the scalar features.

• If you do not specify ScalarizationMethod, the signalFrequencyFeatureExtractor object does not perform any scalarization.

For more information about scalarization methods, see Scalarization Methods for Domain-Specific Signal Features.

Features to Extract

MeanFrequency — Option to extract mean frequency

false (default) | true

Option to extract the mean frequency of the power spectrum, specified as true or false. If you specify MeanFrequency as true, the signalFrequencyFeatureExtractor object extracts the mean frequency of the power spectrum and appends it to the features returned by the extract function.

Data Types: logical

MedianFrequency — Option to extract median frequency

false (default) | true

Option to extract the median frequency of the power spectrum, specified as true or false. If you specify MedianFrequency as true, the signalFrequencyFeatureExtractor object extracts the median frequency of the power spectrum and appends it to the features returned by the extract function.

Data Types: logical

BandPower — Option to extract average band power

false (default) | true

Option to extract the average band power, specified as true or false. If you specify BandPower as true, the signalFrequencyFeatureExtractor object extracts the band power and appends it to the features returned by the extract function.

Data Types: logical

OccupiedBandwidth — Option to extract occupied bandwidth

false (default) | true

Option to extract the 99% occupied bandwidth, specified as true or false. If you specify OccupiedBandwidth as true, the signalFrequencyFeatureExtractor object extracts the 99% occupied bandwidth and appends it to the features returned by the extract function.

To set parameters of the occupied bandwidth extraction, use setExtractorParameters.

setExtractorParameters(sFE,"OccupiedBandwidth",Name=Value)

Settable parameters for the occupied bandwidth extraction are:

• Percentage — Power percentage, specified as a positive integer between 0 and 100.

Data Types: logical

PowerBandwidth — Option to extract half-power bandwidth

false (default) | true

Option to extract the 3 dB (half-power) bandwidth, specified as true or false. If you specify PowerBanwidth as true, the signalFrequencyFeatureExtractor object extracts the 3 dB bandwidth value and appends it to the features returned by the extract function.

To set parameters of the half-power bandwidth extraction, use setExtractorParameters.

setExtractorParameters(sFE,"PowerBandwidth",Name=Value)

Settable parameters for the half-power bandwidth extraction are:

• RelativeAmplitude — Relative amplitude, specified as an integer.

Data Types: logical

WelchPSD — Option to extract power spectral density estimate

false (default) | true

Option to extract the power spectral density (PSD) estimate, specified as true or false. If you specify WelchPSD as true, the signalFrequencyFeatureExtractor object extracts the PSD estimate using Welch's method and appends it to the features returned by the extract function.

To set parameters of the Welch's PSD estimate, use setExtractorParameters.

setExtractorParameters(sFE,"WelchPSD",Name=Value)

Settable parameters for the Welch's PSD estimate extraction are:

• FFTLength — Number of DFT points, specified as a positive integer.

• FrequencyVector — Frequencies at which the PSD is estimated, specified as a vector with at least two elements. You can specify FrequencyVector only when FFTLength is not specified.

• OverlapLength — Number of overlapping samples, specified as a positive integer.

• Window — Window, specified as a scalar or vector.

Data Types: logical

PeakAmplitude — Option to extract peak amplitude

false (default) | true

Option to extract the peak spectral amplitudes, specified as true or false. If you specify PeakAmplitude as true, the signalFrequencyFeatureExtractor object extracts the peak amplitudes of the computed Welch PSD estimate and appends them to the features returned by the extract function.

To set parameters of the peak amplitude extraction, use setExtractorParameters.

setExtractorParameters(sFE,"PeakAmplitude",Name=Value)

Settable parameters for the peak amplitude extraction are:

• PeakType — Type of peak, specified as "minima" or "maxima".

• MaxNumExtrema — Maximum number of peaks, specified as a positive integer scalar.

• MinProminence — Minimum prominence, specified as a positive scalar. The object returns only peaks whose prominence is at least the value specified.

• MinSeparation — Minimum separation between peaks, specified as a positive scalar.

• FlatSelection — Flat region indicator, specified as one of these:

  • "center" — Indicate only the center element of a flat region as the peak.

  • "first" — Indicate only the first element of a flat region as the peak.

  • "last" — Indicate only the last element of a flat region as the peak.

  • "all" — Indicate all elements of a flat region as the peak.

Data Types: logical

PeakLocation — Option to extract peak location

false (default) | true

Option to extract the spectral peak locations, specified as true or false. If you specify PeakLocation as true, the signalFrequencyFeatureExtractor object extracts the peak locations of the computed Welch PSD estimate and appends them to the features returned by the extract function.

To set parameters of the peak location extraction, use setExtractorParameters.

setExtractorParameters(sFE,"PeakLocation",Name=Value)

Settable parameters for the peak location extraction are:

• PeakType — Type of peak, specified as "minima" or "maxima".

• MaxNumExtrema — Maximum number of peaks, specified as a positive integer scalar.

• MinProminence — Minimum prominence, specified as a positive scalar. The setExtractorParameters function returns only peaks whose prominence is at least the value specified.

• MinSeparation — Minimum separation between peaks, specified as a positive scalar.

• FlatSelection — Flat region indicator, specified as one of these:

  • "center" — Indicate only the center element of a flat region as the peak.

  • "first" — Indicate only the first element of a flat region as the peak.

  • "last" — Indicate only the last element of a flat region as the peak.

  • "all" — Indicate all elements of a flat region as the peak.

Data Types: logical

Note

To compute frequency features, signalFrequencyFeatureExtractor first estimates the PSD of the input time-domain signal using Welch's method. The object uses the computed Welch PSD and corresponding frequency vector to compute the specified features. You can configure the computed Welch PSD estimate using the setExtractorParameters function.

Object Functions

extract	Extract time-domain, frequency-domain, or time-frequency-domain features
generateMATLABFunction	Create MATLAB function compatible with C/C++ code generation
getExtractorParameters	Get current parameter values of feature extractor object
setExtractorParameters	Set nondefault parameter values for feature extractor object

Examples

Extract Frequency-Domain Features From Signal

Generate 1024 samples of a chirp sampled at 1024 kHz. The chirp has an initial frequency of 50 kHz and reaches 100 kHz at the end of the sampling. Add white Gaussian noise such that the signal-to-noise ratio is 40 dB. Plot the power spectral density (PSD) and annotate the mean frequency and occupied bandwidth of the signal.

nSamp = 1024;
Fs = 1024e3;
SNR = 40;

t = (0:nSamp-1)'/Fs;

x0 = chirp(t,50e3,nSamp/Fs,100e3);
x = x0+randn(size(x0))*std(x0)/db2mag(SNR);

[pxx,f]= pwelch(x,[],[],[],Fs);
obw(pxx,f);
m = meanfreq(pxx,f);
xline(m/1000,"--", ...
    "Mean Frequency Estimate: "+m/1000+" kHz")

Create a signalFrequencyFeatureExtractor object to extract the mean frequency, 99% occupied bandwidth, and 3 dB bandwidth of the signal.

sFE = signalFrequencyFeatureExtractor(SampleRate=Fs, ...
    MeanFrequency=true,OccupiedBandwidth=true,PowerBandwidth=true)

sFE = 
  signalFrequencyFeatureExtractor with properties:

   Properties
              FrameSize: []
              FrameRate: []
             SampleRate: 1024000
    IncompleteFrameRule: "drop"
          FeatureFormat: "matrix"

   Enabled Features
     MeanFrequency, OccupiedBandwidth, PowerBandwidth

   Disabled Features
     MedianFrequency, BandPower, WelchPSD, PeakAmplitude, PeakLocation


   

Call the extract function to extract the specified features.

[features,info] = extract(sFE,x)

features = 1×3
104 ×

    7.2252    4.3783    3.7773

info = struct with fields:
        MeanFrequency: 1
    OccupiedBandwidth: 2
       PowerBandwidth: 3

To view the extracted features in a table, modify the FeatureFormat property of the object.

sFE.FeatureFormat = "table";
featureTable = extract(sFE,x)

featureTable=1×5 table
    FrameStartTime    FrameEndTime    MeanFrequency    OccupiedBandwidth    PowerBandwidth
    ______________    ____________    _____________    _________________    ______________

          1               1024            72252              43783              37773     

You can use the getExtractorParameters function to view parameters used to compute a specified feature. The occupied bandwidth measures the bandwidth containing 99% of the total power for the input signal by default. Use the setExtractorParameters function to change the percentage to 95% and extract the specified features again.

params = getExtractorParameters(sFE,"OccupiedBandwidth")

params = struct with fields:
    Percentage: []

params.Percentage = 95;
setExtractorParameters(sFE,OccupiedBandwidth=params)
featuresTable2 = extract(sFE,x)

featuresTable2=1×5 table
    FrameStartTime    FrameEndTime    MeanFrequency    OccupiedBandwidth    PowerBandwidth
    ______________    ____________    _____________    _________________    ______________

          1               1024            72252              39840              37773     

More About

Scalarization Methods for Domain-Specific Signal Features

To set the scalarization methods for features in time domain, frequency domain, or time-frequency domain, select the domain-specific feature extractor objects and scalarization method specification. Refer to the following table for the list of domain-specific features from which you can extract scalar features.

Feature domain	Feature extractor object	Scalarization method specification	Features that support scalarization
Time	signalTimeFeatureExtractor	timeScalarFeatureOptions object	PeakValue
Frequency	signalFrequencyFeatureExtractor	frequencyScalarFeatureOptions object	PeakAmplitude WelchPSD
Time-frequency	signalTimeFrequencyFeatureExtractor	timeFrequencyScalarFeatureOptions object	All time-frequency features

For a given feature vector v with N elements, the scalarization method options convert v to a scalar s as follows.

All Signal Domains

• "Mean" — Mean, defined as the average value of v.

  $$s=\overline{v}=\frac{1}{N}{\displaystyle \sum _{i=1}^N{v}_i}$$

• "StandardDeviation" — Standard deviation of the elements of v, normalized by N-1.

  $$s=\sqrt{\frac{1}{N-1}{\displaystyle \sum _{i=1}^N|v_i-\overline{v}{|}^2}}$$

• "PeakValue" — Peak value, defined as the maximum absolute value of v.

  $$s=v_p=\underset{i}{\max }\left|v_i\right|$$

• "Kurtosis" — Kurtosis, defined as the ratio between the fourth moment of v and the squared second moment of v.

  $$s=\frac{\frac{1}{N}{\displaystyle \sum _{i=1}^N{(v_i-\overline{v})}^4}}{{\left[\frac{1}{N}{\displaystyle \sum _{i=1}^N{(v_i-\overline{v})}^2}\right]}^2}$$

• "Skewness" — Skewness, defined as the ratio between the third moment of v and the second moment of v raised to the power of 1.5.

  $$s=\frac{\frac{1}{N}{\displaystyle \sum _{i=1}^N{(v_i-\overline{v})}^3}}{{\left[\frac{1}{N}{\displaystyle \sum _{i=1}^N{(v_i-\overline{v})}^2}\right]}^{3/2}}$$

Frequency and Time-Frequency Signal Domains

• "ClearanceFactor" — Clearance factor, defined as the ratio between the peak value of v and the squared mean of the square roots of the absolute values of v.

  $$s=\frac{v_p}{{\left(\frac{1}{N}{\displaystyle \sum _{i=1}^N\sqrt{\left|v_i\right|}}\right)}^2}$$

• "CrestFactor" — Crest factor, defined as the ratio between the peak value of v and the root-mean-square value of v.

  $$s=\frac{v_p}{\sqrt{\frac{1}{N}{\displaystyle \sum _{i=1}^N{v}_i{}^2}}}$$

• "Energy" — Energy, defined as the sum of the squared values of v.

  $$s={\displaystyle \sum _{i=1}^N{v}_i{}^2}$$

• "Entropy" — Entropy, defined as the sum of plog₂p values, where p is the vector of normalized squared values of v with respect to their sum.

  $$s={\displaystyle \sum _{i=1}^N{p}_i{\log }_2{p}_i},$$

  where

  $$p=\frac{v^2}{{\displaystyle \sum _{i=1}^N{v}_i{}^2}}.$$

  Note

  The scalarization method "Entropy" is not supported for the WaveletEntropy nor the SpectralEntropy features.

• "ImpulseFactor" — Impulse factor, defined as the ratio between the peak value of v and the average absolute value of v.

  $$s=\frac{v_p}{\frac{1}{N}{\displaystyle \sum _{i=1}^N\left|v_i\right|}}$$

Extended Capabilities

C/C++ Code Generation

Generate C and C++ code using MATLAB® Coder™.

Usage notes and limitations:

• You cannot generate code directly from signalFrequencyFeatureExtractor. You can generate C/C++ code from the function returned by generateMATLABFunction.

GPU Arrays

Accelerate code by running on a graphics processing unit (GPU) using Parallel Computing Toolbox™.

This function fully supports GPU arrays. For more information, see Run MATLAB Functions on a GPU (Parallel Computing Toolbox).

Version History

Introduced in R2021b

R2024b: Specify scalarization methods for frequency-domain signal features

The signalFrequencyFeatureExtractor function supports specifying methods to extract scalar features in frequency domain.

R2023a: Use gpuArray inputs

The signalFrequencyFeatureExtractor object supports gpuArray inputs. You must have Parallel Computing Toolbox™ to use this functionality.