I was wondering whether this feature has been built-in via specgram which I believe isn't the case. If not, then the best possible way to implement this using specgram or via matplotlib. If some transformation of the specgram axis can enable this feature.
For example with specgram plots the below won't work(any audio signal can be passed as first argument of specgram):
fig = figure(11)
ax = fig.add_subplot(111)
ax.specgram(tidig.train_symbol_list[0].audio,Fs=12000)
ax.set_yscale('log')
You can change the scaling of axes using the Axes.set_xscale and Axes.set_yscale functions, which both accept either linear, log or symlog as input. So to change a plot to have a log-scaled x axis you would do something like:
import matplotlib.pyplot as plt
ha = plt.subplot(111)
# Plot your spectrogram here...
ha.set_xscale('log')
Edit This seems to be a known issue. There are the commands available to do this but not in any particularly convenient way (it should just be a flag to the specgram function or set_xscale and set_yscale should work). However, there is a way to do this:
Instead of using matplotlib.pyplot.specgram use matplotlib.mlab.specgram. This computes the spectogram but does not draw it. You can then use matplotlib.pyplot.pcolor or a similar function to plot the spectogram. So try something like:
import matplotlib.pyplot as plt
import matplotlib.mlab as mlab
# Set up your data here...
Pxx, freq, t = mlab.specgram(x) # Other options, such as NFFT, may be passed
# to `specgram` here.
ha = plt.subplot(111)
ha.pcolor(t, freq, Pxx)
ha.set_xscale('log')
