warbler

warbler is a pipeline for segmenting, clustering and visualizing Adelaide's warbler songs, which uses a dataset from Dr. David Logue of The Birdsong Lab. This pipeline uses UMAP to perform dimensionality reduction, and HDBSCAN and Fuzzy C-Means for clustering. We also developed avs for the purpose of excluding noise from our dataset.

This research is motivated by the repositories of Dr. Tim Sainburg (AVGN and vocalization segmentation) and Dr. Mara Thomas (UMAP Tutorial).

UMAP and Fuzzy C-Means Clustering

UMAP and HDBSCAN

Prerequisites

• pyenv or Python 3.11.2
• Graphviz
• ffmpeg

Windows

• https://graphviz.org/download/#windows

• https://ffmpeg.org/download.html#build-windows

Linux (Debian)

sudo apt install graphviz graphviz-dev

sudo apt install ffmpeg

Setup

pyenv

pyenv install 3.11.2

pyenv local 3.11.2

Virtual Environment

python -m venv venv

Windows

"venv/Scripts/activate"

Unix

source venv/bin/activate

Packages

pip install -U -r requirements.txt && pip install -U -r local.txt