opensoundscape 0.3.0.1

Open source, scalable acoustic classification for ecology and conservation

Opensoundscape

Installation

We recommend you run Opensoundscape into a Python 3.7 virtual environment. If you don't already have Python installed you can use the Anaconda distribution. Simply select your platform (Windows, macOS, or Linux) and download a Python 3.7 64-bit installer. You will also need to run a MongoDB server which is used to store data for Opensoundscape. The documentation for this process is available here. The documentation also explains how to start the service on various platforms.

Now, you can get Opensoundscape via PyPI. I will show the conda commands below, but we use virtualenvwrapper internally.

1. Create the environment and install Opensoundscape: conda create --name opensoundscape python=3.7 opensoundscape=0.3.0.1
2. Activate the environment: conda activate opensoundscape
3. Check if everything worked: opensoundscape -h
4. Deactivate the environment when finished: conda deactivate

Singularity Container

Currently, Singularity is only working on Linux. The developers recently showed a development version which works on macOS. You can pull our container from here.

1. Get the container: singularity pull library://barrymoo/default/opensoundscape:0.3.0.1
2. Check if Opensoundscape can run: singularity run --app opensoundscape opensoundscape_0.3.0.1.sif -h
3. Check if MongoDB can run: singularity run --app mongodb opensoundscape_0.3.0.1.sif -h

Quick Start Guide

Going to run through a quick example of running Opensoundscape. First, we need some data

• The CLO-43SD-AUDIO dataset:

cd ~/Downloads
wget https://datadryad.org/bitstream/handle/10255/dryad.111783/CLO-43SD-AUDIO.tar.gz
tar -xzf CLO-43SD-AUDIO.tar.gz
rm CLO-43SD-AUDIO.tar.gz

• Download our training & prediction split of a subset of the CLO-43SD dataset:

cd ~/Downloads/CLO-43SD-AUDIO/
wget https://raw.github.com/rhine3/opso-support/master/clo-43sd-train-small.csv
wget https://raw.github.com/rhine3/opso-support/master/clo-43sd-predict-small.csv

• Make a new directory to run Opensoundscape in, using ~/clo-43sd below

cd ~/clo-43sd
wget https://raw.github.com/rhine3/opso-support/master/opso-test-small.ini

• Edit the .ini to reflect the absolute path of your Downloads folder, e.g. with vim: vim opso-test-small.ini
• Start the MongoDB daemon in another terminal: mongod --config /usr/local/etc/mongod.conf
• Run Opensoundscape:

opensoundscape init -i opso-test-small.ini
opensoundscape spect_gen -i opso-test-small.ini > spect-gen-output-small.txt
opensoundscape model_fit -i opso-test-small.ini > model-fit-output-small.txt
opensoundscape predict -i opso-test-small.ini > predict-output-small.txt

Contributions

Contributions are highly encouraged! Our development workflow is a combination of virtualenvwrapper and poetry.

• Get poetry
• Get virtualenvwrapper
• Link poetry and virtualenvwrapper, add something like the following to your ~/.bashrc (don't forget to source it!)

# virtualenvwrapper + poetry
export WORKON_HOME=~/.cache/pypoetry/virtualenvs
source /usr/bin/virtualenvwrapper_lazy.sh

• Fork the github repository, and clone it
  • We use black pre-commit hooks for formatting
• Build the virtual environment for opensoundscape: poetry install
• Activate your opensoundscape environment: workon opensoundscape-py3.7
• Check Opensoundscape runs: opensoundscape -h
• To go back to your system's Python: deactivate
• Running the tests: poetry run pytest tests