| jMIRUtilities |
|---|
Overview
jMIRUtilities is a set of tools for performing miscellaneous tasks relating to preparing musical data for processing by the jMIR components. The current release version can perform the following tasks:
- Batch audio song identification: Audio fingerprinting is used to identify all of the MP3 files in a specified directory and its sub-directories. A tab-delimited text file is output to a specified location, with one line per audio file, and where the file path, song title and artist name are specified for each file. Since this is intended for use in experimental evaluations, any embedded metadata tags are ignored, and only the audio signal itself is used to identify tracks. jMIRUtilities uses the jSongMiner API to perform the identifications.
- Manual instance labelling: A simple GUI is provided for quickly batch labelling sets of audio, MIDI or other files and generating a corresponding ACE XML Instance Label file.
- Automatic instance labelling: An ACE XML Instance Label file may be automatically generated based on a tab-delimited text file containing labelled instances. This is convenient for model classifications that are stored in a spreadsheet table, for example, which may easily be exported to such a tab-delimited text files.
- Extract labels from iTunes files: An iTunes XML file may be parsed and a delimited text file generated that lists the File Path, Recording Title, Artist Name, Composer Name, Album Title and/or Genre Names for each recording in the file, as requested by the user. This can be a useful pre-processing step for generating ACE XML Instance Label files from iTunes XML files via jMIRUtilities’ automatic instance labelling functionality.
- Modify instance identifiers: Text files containing mappings between old instance identifiers and new instance identifiers may be used to automatically update the instance identifiers found in an ACE XML Feature Values file.
- Feature merging: ACE XML Feature Value files that contain different features for the same instances may be merged into a single ACE XML Feature Value file. ACE XML Feature Description files that correspond to these Feature Value files may also be merged. This might be used, for example, to combine feature values extracted by jAudio with feature values extracted by jWebMiner for the same pieces of music.
- Split instances by feature types: An ACE XML Feature Values file can be automatically broken into one new ACE XML Features Values file for each feature that is contained in the original file. Each new file only contains feature values for its associated feature, and only contains instances that have the given feature value available.
- Convert ACE XML feature value files: To either Weka ARFF or CSV files.
Related Publications and Presentations
McKay, C. 2010. Automatic music classification with jMIR. Ph.D. Thesis. McGill University, Canada.
Questions and Comments