For many years, I’ve used awk and gnuplot to generate plots but as I switch over to using Python for my day to day programming, I thought there might be a more Pythonic way to do plots. Google pointed me to Matplotlib and after kicking the tires, I’ve decided to retire gnuplot from my programming toolkit and replace it with matplotlib.
Matplotlib is a 2D plotting library that produces a wide variety of high quality figures suitable for interactive applications or for inserting into publications. A quick tour of the gallery shows the wide range of plots that are possible with Matplotlib. The syntax for creating plots is simple and familiar to anyone who’s used matlab for plotting.
With this new tool in my programming pocket, I thought I’d update my click plotter program to use matplotlib. (The click plotter is a program that generates plots showing how a drummer’s BPM varies over the course of a song. By looking at the generated plots it is easy to see if the beat for a song is being generated by a man or a machine. Read more about it in ‘In Search of the Click Track‘).
The matplotlib code to generate the plots is straightforward:
def plot_click_track(filename): track = track_api.Track(filename) tempo = float(track.tempo['value']) beats = track.beats times = [ dict['start'] for dict in beats ] bpms = get_bpms(times)plt.title('Click Plot for ' + os.path.basename(filename)) plt.ylabel('Beats Per Minute') plt.xlabel('Time') plt.plot(times, get_filtered_bpms(bpms), label='Filtered') plt.plot(times, bpms, color=('0.8'), label='raw') plt.ylim(tempo * .9, tempo * 1.1) plt.axhline(tempo, color=('0.7'), label="Tempo") plt.show()
The complete source code for click_plot is in the example directory of the pyechonest module.
Here are a few examples plots generated by click_plot:
To create your own click plots grab the example from the SVN repository, make sure you have pyechonest and matplotlib installed, get an Echo Nest API key and start generating click plots with the command:% click_plot.py /path/to/music.mp3