You can use fft() function in matlab. Here is an example

t=0:1/8000:1;

x=2*sin(2*pi*1000*t); %Input signal

%figure, plot(t,x);

X=fft(x,8192);

t1=0:8000/8192:(8000/2)-(8000/8192); %Freq scale for Fs=8000

Mag=abs(X);

plot(t1,Mag(1:4096));

A small modification/genralisation of the code by Subhash Kumar

x = % Say x is the Vibration signal you have

N = length(x); % Assuming single channel -- no of samples

Fs = ___ % Sampling Frequency to be specified by you

t = [0:1:N-1]/Fs; % time scale for the entire duration of the signal

Mag = abs(fft(x,Fs)); % Magnitude response

plot(t,Mag) %plot of Magnitude response

Adding to the above answers of the colleagues,you have to fix the sampling frequency according to the Nyquist criterion such that fs>= 2fm, with fm the maximum frequency expected in the signal. As the sampling frequency increases the sampling also in the frequency domain increases and one gets more resolved spectral representation of the signal.

Best wishes