zerocrossrate(___) with no output arguments plots
rate along the y-axis and the corresponding
window number along the x-axis. If the window length is equal to the
full signal length, then the function plots the length of the window along the
x-axis and the crossing rate in the middle of the window.
Count Zero Crossings in Signal
Consider a vector of ones with alternating signs. Plot the data.
x = [1 -1 1 -1 1 -1 1 -1 1 -1]; plot(x)
Compute the zero-crossing rate of
r = zerocrossrate(x)
r = 0.9500
Use the third output argument to find the locations where the crossings occur. Plot
x and the zero-crossing locations. The function returns an index at the next sample after a crossing, not necessarily the exact crossing location. The first sample is marked as a crossing point because the function considers the initial state of
x to be zero by default.
[~,~,indices] = zerocrossrate(x); plot(x) hold on plot(x(indices),'*') hold off
Compute the zero-crossing rate of
x using the comparison method. The rate differs from the value computed using the difference method.
rC = zerocrossrate(x,Method="comparison")
rC = 0.9000
Compute the zero-crossing rate of
x again using the difference method and specify zero as positive. The rate is equal to the value computed using the comparison method.
rZ = zerocrossrate(x,ZeroPositive=1)
rZ = 0.9000
Now specify the initial state of
1. The rate is equal to the previous result.
rI = zerocrossrate(x,InitialState=1)
rI = 0.9000
Count Level Crossings in Temperature Data
Load a set of temperature readings in Celsius taken every hour at Logan Airport in Boston for the entire month of January, 2011. Create a
timetable and use
retime to aggregate the data into daily means.
load bostemp t = hours(1:24*31)'; TT = timetable(t,tempC); rTT = retime(TT,'daily','mean');
Count the number of days the temperature crosses the monthly average. Plot the data and include a horizontal line at the monthly average temperature to visualize where the crossings occur.
avg = mean(TT.tempC)
avg = -1.3007
[~,count] = zerocrossrate(rTT,Level=avg)
count = 9
plot(hours(rTT.t/24),rTT.tempC) yline(avg) xlabel('Time elapsed since January 1, 2011 (days)') ylabel('Average daily temperature (\circC)') axis tight
Identify Voiced and Unvoiced Speech Using Zero Crossings
Speech can be characterized as being voiced or unvoiced. Voiced speech, such as vowel sounds, occurs when the vocal cords vibrate. In unvoiced speech, such as most consonant sounds, the vocal chords do not vibrate. You can use zero crossings to classify the voiced and unvoiced regions in an audio signal.
Load an audio signal into the MATLAB® workspace. The voice says, "Oak is strong, and also gives shade".
[y,fs] = audioread("oak.m4a"); % To hear, type soundsc(y,fs)
The signal is sampled at 44.1 kHz. Calculate the zero-crossing rate for 10 ms windows using the comparison method.
win = fs*0.01; rate = zerocrossrate(y,WindowLength=win,Method="comparison");
rate to visualize the crossing rate for each segment. Voiced speech is expected to have a low crossing rate, while unvoiced speech is expected to have a high crossing rate.
Use a threshold of
0.1 to differentiate between voiced and unvoiced segments. Create a
signalMask object that has two categories ("Unvoiced" and "Voiced") and plot the regions of interest (ROIs). Compare the regions of voiced and unvoiced speech to the location of each spoken word.
IBM® Watson Speech to Text API and Audio Toolbox™ software can be used to extract words from an audio file. Load T
ranscription.mat into the workspace. The labeled signal set contains the audio signal, ROI limits, and labels for each spoken word. For details, see Label Spoken Words in Audio Signals. Display the spoken words on the plot.
h = 0.1; idu = find(rate > h); idu(1:2) = ; vi = [(idu-1) idu]*win; m = sigroi2binmask(vi,length(y)); mask = signalMask([m ~m],Categories=["Unvoiced" "Voiced"],SampleRate=fs); plotsigroi(mask,y) load Transcription ln = getLabelNames(transcribedAudio); v = getLabelValues(transcribedAudio,1,ln); v.Value = categorical(v.Value,v.Value); RL = v.ROILimits; VL = v.Value; hold on text(mean(RL,2),-0.7*ones(size(VL)),VL,HorizontalAlignment="center", ... FontSize=11,FontWeight="bold") hold off
Zero-Crossing Rate of Streaming Data
Load an audio file containing 15 seconds of acoustic guitar music. The sample rate is 44.1 kHz. To play the music, uncomment the last line of code.
Fs = 44100; y = audioread("guitartune.wav"); % sound(y,Fs)
Buffer the audio signal into overlapping frames, each with 4096 samples. Use an overlap of 512 samples.
winLength = 4096; overlap = 512; [yB,~] = buffer(y,winLength,overlap,"nodelay");
Obtain the zero-crossing rate for each frame. To account for frame overlap, specify the initial state as the previous value of the first sample in each frame.
init = winLength - overlap; state = 0; zcr = ; for i = 1:size(yB,2) zcr = [zcr;zerocrossrate(yB(:,i),InitialState=state)]; state = yB(init,i); end
Plot the audio signal and overlay the zero-crossing rate for each frame.
figure yyaxis left x = 0:1/Fs:(numel(y)-1)/Fs; plot(x',y) xlabel("Seconds") ylabel("Amplitude") yyaxis right xx = (1:size(yB,2))*((winLength-overlap)/Fs); plot(xx',zcr) ylabel("Zero-crossing rate")
Compute the zero-crossing rate of the unbuffered signal. To obtain a result equivalent to the zero-crossing rate of the buffered signal, set
512. Determine if the two results are equal.
zcr_batch = zerocrossrate(y,WindowLength=winLength,OverlapLength=overlap); isequal(zcr_batch,zcr)
ans = logical 1
x — Data
real-valued vector | real-valued matrix
Data, specified as a real-valued vector or matrix. If
x is a
matrix, the function returns the zero-crossing rate as a row vector where each value
corresponds to a column of data.
TT — Input timetable
Input timetable, specified as a
TT must contain uniformly sampled single- or double-precision
data. The RowTimes property must
vector with increasing and finite values. If
TT is a timetable with
a single variable containing a matrix, or a timetable with multiple variables each
containing a vector, then the function analyzes each channel independently.
Specify optional pairs of arguments as
the argument name and
Value is the corresponding value.
Name-value arguments must appear after other arguments, but the order of the
pairs does not matter.
Before R2021a, use commas to separate each name and value, and enclose
Name in quotes.
uses the comparison method to compute the rate at which
InitialState — Previous states
0 (default) | vector
Previous states of
x, specified as a vector whose number of
elements is equal to the number of input channels.
zerocrossrate(x,InitialState=[1 0 –1 3]) returns the
crossing rates of a four-channel input signal
Method — Method for computing zero-crossing rate
"difference" (default) |
Method for computing the zero-crossing rate, specified as
"comparison". If you do not
Method, the function uses the difference method to
compute the crossing rate.
comparison— The function marks the
indicesas true where a crossing is fully completed.
difference— The function marks the
indicesas true where abs(sign(xi)–sign(xi–1)) > 0.
zerocrossrate(x,Method="comparison") computes the
crossing rate of
x using the comparison method.
WindowLength — Window length
Window length over which to compute the crossing rate, specified as a positive integer. The default window length is the signal length.
zerocrossrate(x,WindowLength=20) returns the crossing
rates for 20-sample windows in
zerocrossrate(x,WindowLength=fs*0.05) returns the
crossing rates for 50 ms windows in
x given a sample rate
OverlapLength — Number of overlapping samples
0 (default) | positive integer
Number of overlapping samples between adjoining segments, specified as a positive integer. The overlap must be smaller than the window length.
zerocrossrate(x,OverlapLength=0) returns the crossing
rates of segments with no overlap.
returns the crossing rates of overlapping segments with five samples of
Level — Signal level
0 (default) | real scalar
Signal level for which the crossing rate is computed, specified as a real scalar.
The function subtracts the
Level value from the signal and then
finds the zero crossings. If you do not specify
function uses the default value of
0 and returns the zero-crossing
zerocrossrate(x,Level=1) returns the rate at which the
Threshold — Threshold
0 (default) | real scalar
Threshold value above and below the
Level value over which
the crossing rate is computed, specified as a real scalar. The function sets all the
values of the input in the range [–
0 and then finds the zero
zerocrossrate(x,Threshold=0.1) returns the crossing
rate with a tolerance of –0.1 to 0.1.
When you specify both
Threshold, the function first subtracts the level value from
the input and then sets to
0 the resulting input values that are
in the range [–
TransitionEdge — Transitions
"both" (default) |
Transitions to include when counting the zero crossings, specified as
"both". If you specify
"falling", the function
counts only negative-going transitions. If you specify
the function counts only positive-going transitions.
zerocrossrate(x,TransitionEdge="rising") returns the
crossing rate of
x for only positive-going
ZeroPositive — Sign convention
false (default) |
Sign convention, specified as a logical scalar. If you specify
ZeroPositive as true, the function considers
0 to be positive. If you specify
ZeroPositive as false, the function considers
1, and +
1 to have
distinct signs following the convention of the
zerocrossrate(x,ZeroPositive=1) returns the crossing
rate of the input signal
x and considers zero as
rate — Zero-crossing rate
row vector | matrix
Zero-crossing rate, returned as a row vector or a matrix. When
WindowLength is equal to the signal length,
rate is a row vector whose number of elements is equal to the
number of channels in
WindowLength is smaller than the signal length, the function
rate as a matrix where the i-th row
contains the crossing rate for the i-th window and the
j-th column corresponds to the j-th input
count — Number of crossings
Number of crossings, returned as an N-by-M matrix where N is the number of windows and M is the number of input channels. The i-th row corresponds to the crossing count for the i-th window and the j-th column corresponds to the crossing count for the j-th channel.
indices — Logical indices
Logical indices at the signal locations where crossings occur, returned as an
array where N is the number of windows and M is
the number of input channels.
Indices might not represent exact signal crossing locations. The
zerocrossrate function returns an index at the next sample
following a crossing.
C/C++ Code Generation
Generate C and C++ code using MATLAB® Coder™.
Usage notes and limitations:
Code generation does not support disabling dynamic memory allocation when the window length is specified and the input is more than one channel.
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).