AVN GUI

If you would prefer to use AVN without any coding, check out our GUI (Graphical User Interface)! The AVN GUI can calculate acoustic, timing, and syntax features of song, generate multiple useful song visualizations, and calculate the similarity between two birds songs.

Installation

The GUI is currently only available on Windows. If you would be interested in collaborating on building a Mac or Linux compatible version, feel free to reach out!

To install the GUI on Windows, simply download this .zip file: version 0.2.1a, unzip it, and double click the ‘avn_gui.exe’ file. This will launch the GUI, which may take a few seconds.

from IPython.display import Image

Image('Gui_home.png', embed = True, width = 800)

Using the GUI

The GUI has multiple tabs which each perform different types of analyses. Most of these analyses require that your song files be segmented into syllables. We recommend segmenting your songs using WhisperSeg. You can then generate syllable labels using the ‘Labeling’ tab. Those labels can then be used to calculate syntax and acoustic features, and segmentation data or labeling data can be used for timing analysis and similarity scoring. The ‘Feature Set Calculations’ tab lets you calculate all AVN acoustic, syntax and timing features for a given bird with a single click. Each tab has its own ‘Info’ subtab with more information about the tab’s functionality.

If you don’t have any segmented zebra finch song of your own, feel free to use our demo data. We suggest generating syllable labels as a first step using the ‘Labeling’ tab. There, select the ‘/Segmentations/B792_wseg.csv’ file as the ‘Segmentation File’, and the main ‘B792’ folder as the ‘Song Directory’.

Some processes take longer than others in the GUI, and all take longer when working with more song data. Using the demo data (with 58 .wav files), no single processes should take more than a couple of minutes. It is typical for the GUI to freeze or enter a ‘Not Responding’ state while doing these longer processes. If you see this, please don’t close the GUI. So long as there is no error message printed in the GUI’s associated command prompt window, the GUI is still working on the calculation and will unfreeze once the calculation is complete.

The similarity scoring, in particular, is very slow if you are comparing over 5000 syllables from each bird. We recommend randomly sampling ~2000 syllables per bird to compare. We have found that this results in reliable EMD dissimilarity scores and only takes a couple minutes to compute. We plan to add this random sampling to the GUI in a future release. Thanks for your understanding.

Feedback

If you encounter any issues or errors with this GUI, please let us know. We currently consider the AVN GUI to be in ‘alpha’, meaning we are still actively adding features and you may encounter bugs or performance issues. Reporting these issues will help us create a better GUI for everyone and we really appreciate any feedback you may have about the GUI’s interface, functionality, etc. If you would like to share your feedback, please feel free to raise an issue on the AVN Github page, or send me an email at therese.koch1@gmail.com .