johnglover / eurasip2011

Code needed to replicate the results from my 2011 paper in the open access EURASIP Journal on Advances in Signal Processing.

Real-time Detection of Musical Onsets with Linear Prediction and Sinusoidal Modelling - EURASIP 2011

Code needed to replicate the results from my 2011 paper in the open access EURASIP Journal on Advances in Signal Processing.

The article (and reference information) can be found here: http://asp.eurasipjournals.com/content/2011/1/68

Send comments/queries to john dot c dot glover at nuim dot ie

Dependencies

• Modal (and all related dependencies)

Modal Onset Database

The most recent onset database (set of annotated samples) is available from Dropbox

Use

Make sure that the variable data_path in the main __init__.py file corresponds to the directory that your modal onset database is in, and that onsets_path corresponds to the name of the onset database. This defaults to the data folder in the package directory.

Run:

$ python analysis.py

This will create a HDF5 file containing the analysis results called analysis.hdf5 in the same directory.

After making an analysis database, run

$ python results.py

to build a results database.

Running

$ python results_by_odf.py

will create plots of the precision, recall and f-measure results in a directory called images.

Running

$ python results_by_sound_type.py

will print the average F-measure results for each "type" of sound (for each ODF separately). The current sound types are Non-Pitched Percussive, Pitched Non-Percussive, Pitched Percussive and Mixed.