[PDF] More on the Dial Tone Example: Using Octave to Plot the Signal and

More on the Dial Tone Example: Using Octave to Plot the Signal and the Spectrum Note that the python script dial_tone.py saved the 350 Hz sinusoidal signal in the file called audio.dat. Suppose that we want to analyze the data using Octave. The floating binary format is not compatible with MATLAB or Octave. To change it to 32-bit floating format, we use the function: read_float_binary(filename, [count]) Specifically, if all of our files are in the nbfm directory, open the terminal and type the following (Octave) code to generate the time-domain plot: octave cd nbfm data = read_float_binary('audio.dat'); size(data) # just to be sure it is not empty plot(data), axis([0 2000 -.1 .1]) # plot only first 2000 values Continuing with the Octave code, we can approximate the magnitude spectrum of the data using the fft: Fs = 48000; # sampling frequency of data N = length(data) # number of data points spec = fft(data); # numerical approx. of FT df = Fs/N; # spacing between samples on freq. axis min_f = -Fs/2; # min freq. for which fft is calculated max_f = Fs/2 - df; # max freq. for which fft is calculated f = [min_f : df : max_f]; # horizontal values size(f) # should equal N y = abs(fftshift(spec)); # magnitude of shifted spectrum plot(f, y) If you plan to use Octave frequently, you may want to save this in the form of a function m-file, passing in the arguments Fs and data. The resulting plots are shown below:

The sinusoid that was saved in the audio.dat file was of amplitude .1 and frequency 350 Hz, consistent with the graphical results.