Hi Mengjia,
You can use array concatenation to introduce pauses into an audio array.
Assuming you need to have a 1s pause and you will have Fs number of samples in one second (Fs is the sampling frequency), refer the following script for an example,
load gong.mat
% Fs -> Sampling Frequency
% y -> Audio data
% A 1 second pause, in one second you will have Fs number of samples
pause = zeros(Fs,1);
% New array Y
Y = [y(1:4201);pause;y(4202:end)]; % Array concatenation
sound(Y)