labeledSignalSet

Create labeled signal set

Description

Use labeledSignalSet to store labeled signals along with label definitions. Create signal label definitions using signalLabelDefinition.

Creation

Syntax

lss = labeledSignalSet

lss = labeledSignalSet(src)

lss = labeledSignalSet(src,lbldefs)

lss = labeledSignalSet(src,lbldefs,'MemberNames',mnames)

lss = labeledSignalSet(src,lbldefs,PropertyName=Value)

Description

lss = labeledSignalSet creates an empty labeled signal set. Use addMembers to add signals to the set. Use addLabelDefinitions to add label definitions to the set.

lss = labeledSignalSet(src) creates a labeled signal set for the input data source src. Use addLabelDefinitions to add label definitions to the set.

lss = labeledSignalSet(src,lbldefs) creates a labeled signal set for the input data source src using the signal label definitions lbldefs. Use signalLabelDefinition to create signal label definitions.

lss = labeledSignalSet(src,lbldefs,'MemberNames',mnames) creates a labeled signal set for the input data source src and specifies names for the members of the set. Use setMemberNames to modify the member names. lbldefs is optional.

lss = labeledSignalSet(src,lbldefs,PropertyName=Value) sets Properties using name-value arguments. You can specify multiple name-value arguments. Enclose each property name in quotes. lbldefs is optional.

Input Arguments

src — Input data source

matrix | cell array | timetable | signalDatastore object | audioDatastore object

Input data source, specified as a matrix, a cell array, a timetable, a signalDatastore object, or an audioDatastore (Audio Toolbox) object. src implicitly specifies the number of members of the set, the number of signals in each member, and the data in each signal.

Example: {randn(10,3),randn(17,9)} has two members. The first member contains three 10-sample signals. The second member contains nine 17-sample signals.

Example: {{randn(10,1)},{randn(17,1),randn(27,1)}} has two members. The first member contains one 10-sample signal. The second member contains a 17-sample signal and a 27-sample signal.

Example: {{timetable(seconds(1:10)',randn(10,3)),timetable(seconds(1:7)',randn(7,2))},{timetable(seconds(1:3)',randn(3,1))}} has two members. The first member contains three signals sampled at 1 Hz for 10 seconds and two signals sampled at 1 Hz for 7 seconds. The second member contains one signal sampled at 1 Hz for 3 seconds.

Example: signalDatastore Object Pointing to Files

Specify the path to a set of sample sound signals included as MAT files with MATLAB®. Each file contains a signal variable and a sample rate. List the names of the files.

folder = fullfile(matlabroot,"toolbox","matlab","audiovideo");
lst = dir(append(folder,"/*.mat"));
nms = {lst(:).name}'

nms = 7×1 cell
    {'chirp.mat'   }
    {'gong.mat'    }
    {'handel.mat'  }
    {'laughter.mat'}
    {'mtlb.mat'    }
    {'splat.mat'   }
    {'train.mat'   }

Create a signal datastore that points to the specified folder. Set the sample rate variable name to Fs, which is common to all files. Generate a subset of the datastore that excludes the file mtlb.mat, which differs from the other files in that the signal variable is not called y.

sds = signalDatastore(folder,"SampleRateVariableName","Fs");
sdss = subset(sds,~strcmp(nms,"mtlb.mat"));

Use the subset datastore as the source for a labeledSignalSet object.

lss = labeledSignalSet(sdss)

lss = 
  labeledSignalSet with properties:

             Source: [1×1 signalDatastore]
         NumMembers: 6
    TimeInformation: "inherent"
             Labels: [6×0 table]
        Description: ""

 Use labelDefinitionsHierarchy to see a list of labels and sublabels.
 Use setLabelValue to add data to the set.

lbldefs — Label definitions

vector of signalLabelDefinition objects

Label definitions, specified as a vector of signalLabelDefinition objects.

mnames — Member names

character vector | string scalar | cell array of character vectors | string array

Member names, specified as a character vector, a string scalar, a cell array of character vectors, or a string array.

Example: labeledSignalSet({randn(100,1) randn(10,1)},'MemberNames',{'llama' 'alpaca'}) specifies a set of random signals with two members, 'llama' and 'alpaca'.

Properties

Description — Labeled signal set description

character vector | string scalar

Labeled signal set description, specified as a character vector or string scalar.

Example: 'Description','Sleep test patients by sex and age'

Data Types: char | string

SampleRate — Sample rate values

positive scalar | vector

This property is read-only after object creation.

Sample rate values, specified as a positive scalar or a vector. This property is valid only when the data source does not contain inherent time information.

• Set SampleRate to a positive numeric scalar to specify the same sample rate for all signals in the labeled set.

• Set SampleRate to a vector to specify that each member of the labeled set has signals sampled at the same rate, but the sample rates differ from member to member. The vector must have a number of elements equal to the number of members of the set. If a member of a set has signals with different sample rates, then specify the sample rates using timetables.

Example: 'SampleRate',[1e2 1e3] specifies that the signals in the first member of a set are sampled at a rate of 100 Hz and the signals in the second member are sampled at 1 kHz.

SampleTime — Sample time values

positive scalar | vector | duration scalar | duration vector

This property is read-only after object creation.

Sample time values, specified as a positive scalar, a vector, a duration scalar, or a duration vector. This property is valid only when the data source does not contain inherent time information.

• Set SampleTime to a numeric or duration scalar to specify the same sample time for all signals in the labeled set.

• Set SampleTime to a numeric or duration vector to specify that each member of the labeled set has signals with the same time interval between samples, but the intervals differ from member to member. The vector must have a number of elements equal to the number of members of the set. If a member of a set has signals with different sample times, then specify the sample times using timetables.

Example: 'SampleTime',seconds([1e-2 1e-3]) specifies that the signals in the first member of a set have 0.01 second between samples, and the signals in the second member have 1 millisecond between samples.

TimeValues — Time values

vector | duration vector | matrix | cell array

This property is read-only after object creation.

Time values, specified as a vector, a duration vector, a matrix, or a cell array. This property is valid only when the data source does not contain inherent time information. Time values must be unique and increasing.

• Set TimeValues to a numeric or duration vector to specify the same time values for all signals in the labeled set. The vector must have the same length as all the signals in the set.

• Set TimeValues to a numeric or duration matrix or cell array to specify that each member of the labeled set has signals with the same time values, but the time values differ from member to member.

  • If TimeValues is a matrix, then it must have a number of columns equal to the number of members of the set. All signals in the set must have a length equal to the number of rows of the matrix.

  • If TimeValues is a cell array, then it must contain a number of vectors equal to the number of members of the set. All signals in a member must have a length equal to the number of elements of the corresponding vector in the cell array.

If a member of a set has signals with different time values, then specify the time values using timetables.

Example: 'TimeValues',[1:1000;0:1/500:2-1/500]' specifies that the signals in the first member of a set are sampled 1 Hz for 1000 seconds. The signals in the second member are sampled at 500 Hz for 2 seconds.

Example: 'TimeValues',seconds([1:1000;0:1/500:2-1/500]') specifies that the signals in the first member of a set are sampled 1 Hz for 1000 seconds. The signals in the second member are sampled at 500 Hz for 2 seconds.

Example: 'TimeValues',{1:1000,0:1/500:2-1/500} specifies that the signals in the first member of a set are sampled 1 Hz for 1000 seconds. The signals in the second member are sampled at 500 Hz for 2 seconds.

Example: 'TimeValues',{seconds(1:1000),seconds(0:1/500:2-1/500)} specifies that the signals in the first member of a set are sampled 1 Hz for 1000 seconds. The signals in the second member are sampled at 500 Hz for 2 seconds.

NumMembers — Number of members in set

Read-only: positive integer

This property is read-only.

Number of members in set, returned as a positive integer.

Labels — Labels table

Read-only: table

This property is read-only.

Labels table, returned as a MATLAB^® table. Each variable of Labels corresponds to a label defined for the set. Each row of Labels corresponds to a member of the data source. The row names of Labels are the member names.

Data Types: table

TimeInformation — Time information of source

'none' | 'sampleRate' | 'sampleTime' | 'timeValues' | 'inherent'

Time information of source, specified as one of the following:

• 'none' — The signals in the source have no time information.

• 'sampleRate' — The signals in the source are sampled at a specified rate.

• 'sampleTime' — The signals in the source have a specified time interval between samples.

• 'timeValues — The signals in the source have a time value corresponding to each sample.

• 'inherent' — The signals in the source contain inherent time information. MATLAB timetables are an example of such signals.

Data Types: char | string

Source — Data source of labeled signal set

matrix | cell array | timetable

This property is read-only after object creation.

Data source of labeled signal set, specified as a matrix, a timetable, a cell array, or an audio datastore.

• If Source is a numeric matrix, then the labeled signal set has one member that contains a number of signals equal to the number of matrix columns.

  Example: labeledSignalSet(randn(10,3)) has one member that contains three 10-sample signals.

• If Source is a cell array of matrices, then the labeled signal set has a number of members equal to the number of matrices in the cell array. Each member contains a number of signals equal to the number of columns of the corresponding matrix.

  Example: labeledSignalSet({randn(10,3),randn(17,9)}) has two members. The first member contains three 10-sample signals. The second member contains nine 17-sample signals.

• If Source is a cell array, and each element of the cell array is a cell array of numeric vectors, then the labeled signal set has a number of members equal to the number of cell array elements. Each signal within a member can have any length.

  Example: labeledSignalSet({{randn(10,1)},{randn(17,1),randn(27,1)}}) has two members. The first member contains one 10-sample signal. The second member contains a 17-sample signal and a 27-sample signal.

• If Source is a timetable with variables containing numeric values, then the labeled signal set has one member that contains a number of signals equal to the number of variables. The time values of the timetable must be of type duration, unique, and increasing.

  Example: labeledSignalSet(timetable(seconds(1:10)',randn(10,3))) has one member that contains three signals sampled at 1 Hz for 10 seconds.

• If Source is a cell array of timetables, and each timetable has an arbitrary number of variables with numeric values, then the labeled signal set has a number of members equal to the number of timetables. Each member contains a number of signals equal to the number of variables in the corresponding timetable.

  Example: labeledSignalSet({timetable(seconds(1:10)',randn(10,3)),timetable(seconds(1:5)',randn(5,13))}) has two members. The first member contains three signals sampled at 1 Hz for 10 seconds. The second member contains 13 signals sampled at 1 Hz for 5 seconds.

• If Source is a cell array, and each element of the cell array is a cell array of timetables, then the labeled signal set has a number of members equal to the number of cell array elements. Each member can have any number of timetables, and each timetable within a member can have any number of variables.

  Example: labeledSignalSet({{timetable(seconds(1:10)',randn(10,3)),timetable(seconds(1:7)',randn(7,2))},{timetable(seconds(1:3)',randn(3,1))}}) has two members. The first member contains three signals sampled at 1 Hz for 10 seconds and two signals sampled at 1 Hz for 7 seconds. The second member contains one signal sampled at 1 Hz for 3 seconds.

• If the input data source, src, is an audio datastore, then the labeled signal set has a number of members equal to the number of files to which the datastore points. The Source property contains a cell array of character vectors with the file names. Each member contains all the signals returned by the read of the corresponding datastore file.

Object Functions

addLabelDefinitions	Add label definitions to labeled signal set
addMembers	Add members to labeled signal set
countLabelValues	Count label values
createDatastores	Create datastores pointing to signal and label data
createFeatureData	Create feature table or matrix and response vectors
editLabelDefinition	Edit label definition properties
getAlternateFileSystemRoots	Get alternate file system roots when data source of labeled signal set is a datastore
getLabelDefinitions	Get label definitions in labeled signal set
getLabeledSignal	Get labeled signals from labeled signal set
getLabelIndices	Get label indices pointing to label definitions in labeled signal set
getLabelNames	Get label names in labeled signal set
getLabelValues	Get label values from labeled signal set
getMemberNames	Get member names in labeled signal set
getSignal	Get signals from labeled signal set
head	Get top rows of labels table
labelDefinitionsHierarchy	Get hierarchical list of label and sublabel names
labelDefinitionsSummary	Get summary table of signal label definitions
merge	Merge two or more labeled signal sets
removeLabelDefinition	Remove label definition from labeled signal set
removeMembers	Remove members from labeled signal set
removePointValue	Remove row from point label
removeRegionValue	Remove row from ROI label or time-frequency ROI label
resetLabelValues	Reset labels to default values
setAlternateFileSystemRoots	Set alternate file system roots when data source of labeled signal set is a datastore
setLabelValue	Set label value in labeled signal set
setMemberNames	Set member names in labeled signal set
subset	Get new labeled signal set with subset of members

Examples

Label Definitions for Whale Songs

Consider a set of whale sound recordings. The recorded whale sounds consist of trills and moans. Trills sound like series of clicks. Moans are low-frequency cries similar to the sound made by a ship's horn. You want to look at each signal and label it to identify the whale type, the trill regions, and the moan regions. For each trill region, you also want to label the signal peaks higher than a certain threshold.

Signal Label Definitions

Define an attribute label to store whale types. The possible categories are blue whale, humpback whale, and white whale.

dWhaleType = signalLabelDefinition("WhaleType", ...
   LabelType="attribute", ...
   LabelDataType="categorical", ...
   Categories=["blue" "humpback" "white"], ...
   Description="Whale type"); 

Define a region-of-interest (ROI) label to capture moan regions. Define another ROI label to capture trill regions.

dMoans = signalLabelDefinition("MoanRegions", ...
   LabelType="roi", ...
   LabelDataType="logical", ...
   Description="Regions where moans occur");

dTrills = signalLabelDefinition("TrillRegions", ...
   LabelType="roi", ...
   LabelDataType="logical", ...
   Description="Regions where trills occur");        

Finally, define a point label to capture the trill peaks. Set this label as a sublabel of the dTrills definition.

dTrillPeaks = signalLabelDefinition("TrillPeaks", ...
   LabelType="point", ...
   LabelDataType="numeric", ...
   Description="Trill peaks");

dTrills.Sublabels = dTrillPeaks;

Labeled Signal Set

Create a labeledSignalSet with the whale signals and the label definitions. Add label values to identify the whale type, the moan and trill regions, and the peaks of the trills.

load labelwhalesignals
lbldefs = [dWhaleType dMoans dTrills];

lss = labeledSignalSet({whale1 whale2},lbldefs, ...
    MemberNames=["Whale1" "Whale2"], ...
   SampleRate=Fs,Description="Characterize whale song regions");     

Visualize the label hierarchy and label properties using labelDefinitionsHierarchy and labelDefinitionsSummary.

labelDefinitionsHierarchy(lss)

ans = 
    'WhaleType
       Sublabels: []
     MoanRegions
       Sublabels: []
     TrillRegions
       Sublabels: TrillPeaks
     '

labelDefinitionsSummary(lss)

ans=3×9 table
      LabelName        LabelType     LabelDataType     Categories     ValidationFunction    DefaultValue             Sublabels             Tag            Description         
    ______________    ___________    _____________    ____________    __________________    ____________    ___________________________    ___    ____________________________

    "WhaleType"       "attribute"    "categorical"    {3×1 string}       {["N/A"   ]}       {0×0 double}    {0×0 double               }    ""     "Whale type"                
    "MoanRegions"     "roi"          "logical"        {["N/A"   ]}       {0×0 double}       {0×0 double}    {0×0 double               }    ""     "Regions where moans occur" 
    "TrillRegions"    "roi"          "logical"        {["N/A"   ]}       {0×0 double}       {0×0 double}    {1×1 signalLabelDefinition}    ""     "Regions where trills occur"

The signals in the loaded data correspond to songs of two blue whales. Set the "WhaleType" values for both signals.

setLabelValue(lss,1,"WhaleType","blue");
setLabelValue(lss,2,"WhaleType","blue");

Visualize the Labels property. The table has the newly added "WhaleType" values for both signals.

lss.Labels      

ans=2×3 table
              WhaleType    MoanRegions    TrillRegions
              _________    ___________    ____________

    Whale1      blue       {0×2 table}    {0×3 table} 
    Whale2      blue       {0×2 table}    {0×3 table} 

Visualize Region Labels

Visualize the whale songs to identify the trill and moan regions.

subplot(2,1,1)
plot((0:length(whale1)-1)/Fs,whale1)
ylabel("Whale 1")

subplot(2,1,2)
plot((0:length(whale2)-1)/Fs,whale2)
ylabel("Whale 2")

Moan regions are sustained low-frequency wails.

• whale1 has moans centered at about 7 seconds, 12 seconds, and 17 seconds.

• whale2 has moans centered at about 3 seconds, 7 seconds, and 16 seconds.

Add the moan regions to the labeled set. Specify the ROI limits in seconds and the label values.

moanRegionsWhale1 = [6.1 7.7; 11.4 13.1; 16.5 18.1];
mrsz1 = [size(moanRegionsWhale1,1) 1];
setLabelValue(lss,1,"MoanRegions",moanRegionsWhale1,true(mrsz1));

moanRegionsWhale2 = [2.5 3.5; 5.8 8; 15.4 16.7];
mrsz2 = [size(moanRegionsWhale2,1) 1];
setLabelValue(lss,2,"MoanRegions",moanRegionsWhale2,true(mrsz2));

Trill regions have distinct bursts of sound punctuated by silence.

• whale1 has a trill centered at about 2 seconds.

• whale2 has a trill centered at about 12 seconds.

Add the trill regions to the labeled set.

trillRegionWhale1 = [1.4 3.1];
trsz1 = [size(trillRegionWhale1,1) 1];
setLabelValue(lss,1,"TrillRegions",trillRegionWhale1,true(trsz1));

trillRegionWhale2 = [11.1 13];
trsz2 = [size(trillRegionWhale1,1) 1];
setLabelValue(lss,2,"TrillRegions",trillRegionWhale2,true(trsz2));

Create a signalMask object for each whale song and use it to visualize and label the different regions. For better visualization, change the label values from logical to categorical.

mr1 = getLabelValues(lss,1,"MoanRegions");
mr1.Value = categorical(repmat("moan",mrsz1));
tr1 = getLabelValues(lss,1,"TrillRegions");
tr1.Value = categorical(repmat("trill",trsz1));

msk1 = signalMask([mr1;tr1],"SampleRate",Fs);

subplot(2,1,1)
plotsigroi(msk1,whale1)
ylabel("Whale 1")
hold on

mr2 = getLabelValues(lss,2,"MoanRegions");
mr2.Value = categorical(repmat("moan",mrsz2));
tr2 = getLabelValues(lss,2,"TrillRegions");
tr2.Value = categorical(repmat("trill",trsz2));

msk2 = signalMask([mr2;tr2],"SampleRate",Fs);

subplot(2,1,2)
plotsigroi(msk2,whale2)
ylabel("Whale 2")
hold on

Visualize Point Labels

Label three peaks for each trill region. For point labels, you specify the point locations and the label values. In this example, the point locations are in seconds.

peakLocsWhale1 = [1.553 1.626 1.7];
peakValsWhale1 = [0.211 0.254 0.211];

setLabelValue(lss,1,["TrillRegions" "TrillPeaks"], ...
   peakLocsWhale1,peakValsWhale1,LabelRowIndex=1);

subplot(2,1,1)
plot(peakLocsWhale1,peakValsWhale1,"v")
hold off

peakLocsWhale2 = [11.214 11.288 11.437];
peakValsWhale2 = [0.119 0.14 0.15];

setLabelValue(lss,2,["TrillRegions" "TrillPeaks"], ...
   peakLocsWhale2,peakValsWhale2,LabelRowIndex=1);

subplot(2,1,2)
plot(peakLocsWhale2,peakValsWhale2,"v")
hold off

Explore Label Values

Explore the label values using getLabelValues.

getLabelValues(lss)

ans=2×3 table
              WhaleType    MoanRegions    TrillRegions
              _________    ___________    ____________

    Whale1      blue       {3×2 table}    {1×3 table} 
    Whale2      blue       {3×2 table}    {1×3 table} 

Retrieve the moan regions for the first member of the labeled set.

getLabelValues(lss,1,"MoanRegions")

ans=3×2 table
     ROILimits      Value
    ____________    _____

     6.1     7.7    {[1]}
    11.4    13.1    {[1]}
    16.5    18.1    {[1]}

Use a second output argument to list the sublabels of a label.

[value,valueWithSublabel] = getLabelValues(lss,1,"TrillRegions")

value=1×2 table
    ROILimits     Value
    __________    _____

    1.4    3.1    {[1]}

valueWithSublabel=1×3 table
    ROILimits     Value     Sublabels 
    __________    _____    ___________

                           TrillPeaks 
                           ___________
                                      
    1.4    3.1    {[1]}    {3×2 table}

To retrieve the values in a sublabel, express the label name as a two-element array.

getLabelValues(lss,1,["TrillRegions","TrillPeaks"])

ans=3×2 table
    Location      Value   
    ________    __________

     1.553      {[0.2110]}
     1.626      {[0.2540]}
       1.7      {[0.2110]}

Find the value of the third trill peak corresponding to the second member of the set.

getLabelValues(lss,2,["TrillRegions" "TrillPeaks"], ...
    LabelRowIndex=1,SublabelRowIndex=3)

ans=1×2 table
    Location      Value   
    ________    __________

     11.437     {[0.1500]}

Count Label Values and Create Datastores

Specify the path to a set of audio signals included as MAT files with MATLAB®. Each file contains a signal variable and a sample rate. List the names of the files.

folder = fullfile(matlabroot,"toolbox","matlab","audiovideo");
lst = dir(append(folder,"/*.mat"));
nms = {lst(:).name}'

nms = 7×1 cell
    {'chirp.mat'   }
    {'gong.mat'    }
    {'handel.mat'  }
    {'laughter.mat'}
    {'mtlb.mat'    }
    {'splat.mat'   }
    {'train.mat'   }

Create a signal datastore that points to the specified folder. Set the sample rate variable name to Fs, which is common to all files. Generate a subset of the datastore that excludes the file mtlb.mat. Use the subset datastore as the source for a labeledSignalSet object.

sds = signalDatastore(folder,SampleRateVariableName="Fs");
sds = subset(sds,~strcmp(nms,"mtlb.mat"));
lss = labeledSignalSet(sds);

Create three label definitions to label the signals:

• Define a logical attribute label that is true for signals that contain human voices.

• Define a numeric point label that marks the location and amplitude of the maximum of each signal.

• Define a categorical region-of-interest (ROI) label to pick out nonoverlapping, uniform-length random regions of each signal.

Add the signal label definitions to the labeled signal set.

vc = signalLabelDefinition("Voice",LabelType="attribute", ...
    LabelDataType="logical",DefaultValue=false);
mx = signalLabelDefinition("Maximum",LabelType="point", ...
    LabelDataType="numeric");
rs = signalLabelDefinition("RanROI",LabelType="ROI", ...
    LabelDataType="categorical",Categories=["ROI" "other"]);
addLabelDefinitions(lss,[vc mx rs])

Label the signals:

• Label 'handel.mat' and 'laughter.mat' as having human voices.

• Use the islocalmax function to find the maximum of each signal. Label its location and value.

• Use the randROI function to generate as many regions of length N/10 samples as can fit in a signal of length N given a minimum separation of N/6 samples between regions. Label their locations and assign them to the ROI category.

When labeling points and regions, convert sample values to time values. Subtract 1 to account for MATLAB array indexing and divide by the sample rate.

kj = 1;
while hasdata(sds)
    
    [sig,info] = read(sds);
    fs = info.SampleRate;

    [~,fn] = fileparts(info.FileName);
    if fn=="handel" || fn=="laughter"
        setLabelValue(lss,kj,"Voice",true)
    end
    
    xm = find(islocalmax(sig,MaxNumExtrema=1));
    setLabelValue(lss,kj,"Maximum",(xm-1)/fs,sig(xm))

    N = length(sig);
    rois = randROI(N,round(N/10),round(N/6));
    setLabelValue(lss,kj,"RanROI",(rois-1)/fs, ...
        repelem("ROI",size(rois,1)))

    kj = kj+1;
    
end

Verify that only two signals contain voices.

countLabelValues(lss,"Voice")

ans=2×3 table
    Voice    Count    Percent
    _____    _____    _______

    false      4      66.667 
    true       2      33.333 

Verify that two signals have a maximum amplitude of 1.

countLabelValues(lss,"Maximum")

ans=5×4 table
           Maximum            Count    Percent    MemberCount
    ______________________    _____    _______    ___________

    0.80000000000000004441      1      16.667          1     
    0.89113331915798421612      1      16.667          1     
    0.94730769230769229505      1      16.667          1     
    1                           2      33.333          2     
    1.0575668990330560071       1      16.667          1     

Verify that each signal has four nonoverlapping random regions of interest.

countLabelValues(lss,"RanROI")

ans=2×4 table
    RanROI    Count    Percent    MemberCount
    ______    _____    _______    ___________

    ROI        24        100           6     
    other       0          0           0     

Create two datastores with the data in the labeled signal set:

• The signalDatastore object sd contains the signal data.

• The arrayDatastore object ld contains the labeling information. Specify that you want to include the information corresponding to all the labels you created.

[sd,ld] = createDatastores(lss,["Voice" "RanROI" "Maximum"]);

Use the information in the datastores to plot the signals and display their labels.

• Use a signalMask object to highlight the regions of interest in blue.

• Plot yellow lines to mark the locations of the maxima.

• Add a red axis label to the signals that contain human voices.

tiledlayout flow

while hasdata(sd)

    [sg,nf] = read(sd);
    
    lbls = read(ld);
    
    nexttile
    
    msk = signalMask(lbls{:}.RanROI{:},SampleRate=nf.SampleRate);
    plotsigroi(msk,sg)
    colorbar off
    xlabel('')
    
    xline(lbls{:}.Maximum{:}.Location, ...
        LineWidth=2,Color="#EDB120")
    
    if lbls{:}.Voice{:}
        ylabel("VOICED",Color="#D95319")
    end

end

function roilims = randROI(N,wid,sep)

num = floor((N+sep)/(wid+sep));
hq = histcounts(randi(num+1,1,N-num*wid-(num-1)*sep),(1:num+2)-1/2);
roilims = (1 + (0:num-1)*(wid+sep) + cumsum(hq(1:num)))' + [0 wid-1];

end

Label Signal in Time-Frequency Domain

Since R2025a

Label Gaussian atoms in the time-frequency domain using a time-frequency region-of-interest (ROI) label definition and spectrogram options.

Generate Signal and Visualize Spectrogram

Generate a signal that consists of a voltage-controlled oscillator and four Gaussian atoms. The signal is sampled at 14 kHz for two seconds. Plot the spectrogram of the signal.

Fs = 14000;
t = (0:1/Fs:2)';
st = 0.01;
gaussFun = @(A,x,mu,f) exp(-(x-mu).^2/(2*st^2)).*sin(2*pi*f.*x)*A';
atomTimeCenters = [0.2 0.5 1 1.75];
atomFreqCenters = [2 6 2 5]*1000;
s = gaussFun([1 1 1 1]/10,t,atomTimeCenters,atomFreqCenters);
x = vco(chirp(t+.1,0,t(end),3).*exp(-2*(t-1).^2),[0.1 0.4]*Fs,Fs);
s = s/10+x;

bt = 0.2;
tr = 0.05;
op = 99;
pspectrum(s,Fs,"spectrogram", ...
    Leakage=bt,TimeResolution=tr,OverlapPercent=op)

The spectrogram shows four patches in time-frequency domain that correspond with the Gaussian atoms. Define the times and frequencies for all the atoms.

atomTimes = atomTimeCenters'+[-st st]*5.5;
atomFreqs = atomFreqCenters'+[-1 1]*200;

Label Signal in Time-Frequency Domain

Create a logical time-frequency ROI label definition to label the Gaussian atoms. Specify spectrogram options with leakage properties.

opts = labelSpectrogramOptions("leakage", ...
    Leakage=40*(1-bt),Overlap=op, ...
    TimeResolutionMode="specify",TimeResolution=tr);

lblDef = signalLabelDefinition("Atom", ...
    LabelDataType="logical", ...
    LabelType="roiTimeFrequency",TimeFrequencyOptions=opts);

Create a labeled signal set from the signal and time-frequency ROI label definition.

lss = labeledSignalSet(s,lblDef,SampleRate=Fs);

Label the four atoms in time-frequency domain. Set the label values to true.

setLabelValue(lss,1,"Atom",atomTimes,atomFreqs,true(1,4))

Visualize Time-Frequency Image and Label Mask

Create datastores from the labeled signal set for the time-frequency ROI label.

imSize = [512 768];
[sds,ads] = createDatastores(lss,"Atom", ...
    TimeFrequencyMapFormat="image", ...
    TimeFrequencyImageSize=imSize, ...
    TimeFrequencyLabelFormat="mask", ...
    TimeFrequencyMaskPriority=true);

Read and show the time-frequency image.

imagesc(read(sds))

Read the label mask and display it above the time-frequency image.

lbl = read(ads);
im = zeros([imSize 3]);
im(:,:,1) = lbl{1};
hold on
imagesc(im,AlphaData=0.5*lbl{1})
hold off

Version History

Introduced in R2018b

R2025a: Generate labeled signal sets with time-frequency ROI labels

The labeledSignalSet object supports ROI time-frequency label definitions.

• Specify label definitions that include "roiTimeFrequency" labels.

• Create datastores that point to spectrograms and label masks and use them to train segmentation and detection deep learning models.

• Edit time-frequency labels programmatically. Set label values, get label definitions, names, indices, and labeled signals, and remove region values.