Install¶
AVA requires Python3 and standard packages available in the Anaconda distribution. In addition to these packages, AVA also requires PyTorch (>=v1.1), UMAP, and affinewarp.
Installation¶
AVA can be installed by opening a shell, navigating to a suitable directory, and entering the following commands:
$ git clone https://github.com/jackgoffinet/autoencoded-vocal-analysis.git
$ cd autoencoded-vocal-analysis
$ pip install .
This will install AVA as a python package in your current environment.
Operating systems¶
AVA is built and tested with Ubuntu and OSX operating systems. It has also run on Windows, but open issues in PyTorch make for slow training.
GPU Acceleration¶
Like most neural network models, the VAE trains much faster on GPU than on CPU. Training times on GPU are about a few minutes to an hour and roughly 10x slower on CPU. If you have a GPU, make sure to install PyTorch with CUDA.