Code underlying the BSc thesis: Efficient Auditory Coding for Bat Vocalizations
This is the pipeline I used for my Bachelor's Thesis on Efficient Auditory Coding for Bat Vocalizations. It includes four Python notebooks. The first one, 'download', includes scripts for downloading bat recordings from the ChiroVox dataset, as bulk download functionality is not available. The second one, 'preprocessing', contains analytics about the data distribution, as well as preprocessing steps, such as resampling, denoising and cropping the data into individual bat calls using a custom energy-based approach at bat call detection. Lastly, the third and fourth notebooks, namely 'analysis' and 'results' include tests about the performance of auditory kernels, their relative usage and activation patterns, as well as clustering of bat calls using low-fidelity reconstructions and sparsity experiments.