Generate a false colour index spectrogram to nicely visualise long duration soundscape recordings - based on Towsey et. al. 2014
calculate_index_spectrograms.py - if you run this it calculates the indices for the false colour index spectrogram, and also plot the resulting false-colour spectrogram. This file also contains the main functions which can be used in your own Python scripts if you import the file as a module. When using the file as a script, you can optionally change the input/output paths and you can also disable re-calculation of the statistics: run python calculate_index_spectrograms.py -h for information about the command-line options.
plot_index_distributions.py - used to look at the distribution of values for each of the three indices
Code tested using Python 3.7 on a Windows laptop, but should work on most other platforms.
24 hours recorded from a tropical rainforest in Sabah, Borneo. Dawn and dusk choruses are clearly visible with different patterns of calling during the day and night time
Input can be a single long file which will be analysed in 1-minute chunks, OR a folder containing a series of WAV files (we assume they are 1 minute long).
Long-duration recordings in single large WAV files can be handled by this script. On our laptop it takes about 30min to analyse 12 hours (i.e. 24x speed) -- the time taken will scale linearly with the length of the input. Note that it's NOT any faster to split the audio into chunks first.
Numpy ndarrays storing individual index spectrograms are stored in ./output_spectrograms/ folder.
- Sarab Sethi (Imperial College London)
- Dan Stowell (Queen Mary University of London)