GuitarEvaluator computes onset and pitch estimation on monophonic guitar audio recordings. Onset detection is based on Madmom's Convolutional Neural Networcks algorithm; pitch estimation is based on Essentia's Melodia extractor. Evaluation can be done by executing the tests as explained below. Tests generate a log file with statistical results regarding the performance of the onset and pitch detectors. Usage::
import essentia.standard as es
from guitareval.guitar_evaluator import GuitarEvaluator
path = 'some_path/audio.wav'
sr = 22050
audio = es.MonoLoader(filename=path, sampleRate=sr)()
evaluator = GuitarEvaluator(sr)
onsets, pitch = evaluator.compute(audio)
First you need to install Essentia MIR library as explained in http://essentia.upf.edu/documentation/installing.html Then in the following commands from GuitarEvaluator root folder:
-
easy_install pip (only if you don't have pip installed)
-
make init
-
make install
-
pip install nose (if you don't have installed nose)
-
make test