Python packages for audio onset detection Evaluation

In the "src" dir the tool is organised into python packages as followed:-

• Eval : statistical tools for evaluating onsets prediction vs ground-truth
• Predict : global package to generate pool and onsets text files
  • OnsetNovelty : package grouping multiple onset novelty functions
  • Slice : package grouping multiple multidimensional "peak-picking" algorithms
• Utils : common routines and functions for dealing with files and the environment.

conf.py at the top level holds environment specific things - customise this for your setup.

Specify folders in conf.py

for already computed onset textfiles : run Eval/__init__.py

for computing : specify desired methods in OnsetNovelty's and Slice's __init__.py then run Predict's init.py

you can use last computation using pool cache (basicaly set fromFile=true in config.py)

1. Install PyDev

• http://pydev.org/manual_101_install.html
• You may need to install Oracle Java (http://www.oracle.com/technetwork/java/javase/downloads/jdk8-downloads-2133151.html)
• Then set Java -> Installed JREs -> to Java SE 8
• Restart and in Preferences you should now have PyDev appearing
• Set Interpreters > Python Interpreter to "/usr/local/bin/python" or whatever

2. Go to Eclipse Marketplace

• Install EGit

3. Pull down code

• Go to File -> Import
• Select "Projects from Git"
• Clone URI "https://github.com/GiantSteps/EvalOnsets"
• Import existing project to your workspace destination

Make sure to set the "src" directory as the PYTHONPATH to detect the packages.

• essentia
• signal
• sklearn
• scypi
• matplotlib
• modal (https://github.com/johnglover/modal)

*opencv