The GS-API is a Python/C++ library for manipulating musical symbolic data.
The GS-API (GiantSteps API) provides Python and C++ classes and an interface for dealing with musical data. Its main features are:
- flexible input/output from/to JSON/MIDI.
- Rhythm generation, both agnostic and based on styles.
- Style and music-theoretical based harmony progression generation.
- More to come.
Installing the latests stable release can be done via pip:
pip install gsapi
Python
The python modules reside in the gsapi subfolder:
- Build and install:
cd gsapi
python setup.py build
python setup.py install
The following lines will create a Pattern p with two events: one event tagged as "Kick" starting at time 0, with a duration of 1, a Midi Note Number 64 and a Midi velocity of 127; and a second event tagged "Snare" starting at time 1, with a duration of 3, Midi Note Number 62 and Midi velocity of 51.
from gsapi import *
p = gspattern.Pattern()
p.addEvent(gspattern.Event(startTime=0, duration=1, pitch=64, velocity=127, tag="Kick"))
p.addEvent(gspattern.Event(1, 3, 62, 51, "Snare"))
Now, let us get all files in a specified folder and slice them into 16 beat patterns:
from gsapi import *
# we select a folder containing midi files:
dataset = gsdataset.Dataset(midiFolder="your/folder",midiGlob="*.mid", midiMap=gsdefs.generalMidiMap)
allPatternsSliced = []
for midiPattern in dataset.patterns:
for sliced in midiPattern.splitInEqualLengthPatterns(16):
allPatternsSliced+=[sliced]
print(allPatternsSliced)
We can use the GS-API to characterize a pattern with any given descriptor. Following the previous example, we could estimate the rhythmic density of the patterns:
from gsapi import *
density = descriptors.Density()
for pattern in dataset.patterns:
kickPattern = pattern.getPatternWithTags(tagToLookFor="kick")
densityOfKick = density.getDescriptorForPattern(kickPattern)
We can easily extract transition probabilities from corpora of music in order to generate new music in that style:
from gsapi import *
markovchain = styles.MarkovStyle(order=3, numSteps=32, loopDuration=16)
MarkovStyle.generateStyle(allPatternsSliced)
newPattern = MarkovStyle.generatePattern()
JSON and MIDI
We encourage the use of JSON to be able to work with consistent and reusable datasets, as midi files tend to have different MIDI mappings, structures, or even suspicious file format implementations. Thus we provide a flexible MIDI i/o module gsio for tagging events with respect to their pitch, channel and trackName. Note to tag mapping is reppresented by dictionaries where keys represent tags and values are rules that such tags have to validate.
- Rules are either lists or single condition that are OR'ed.
- Each condition is either a tuple with expected pitch number and channel, or an integer representing the expected pitch value.
The following snippet will return a list of Patterns with events tagged as 'Kick' if Midi pitch is 30 in any channel; 'Snare' if Midi pitch is 32 and channel is 4; and 'ClosedHihat' if MIDI pitch is 33 on whatever channel or MIDI pitch is 45 on any channel.
from gsapi import *
midiGlobPath = '/path/to/midi/*.mid'
NoteToTagsMap = {"Kick":30,
"Snare":(32,4),
"ClosedHihat":[(33,'*'),45]}
listOfGSPatterns = gsio.fromMidiCollection(midiGlobPath, NoteToTagsMap)
All submodules within the GS-API are documented and can provide basic help by typing help(moduleName).
help(gspattern.Pattern)
help(gsdataset.Dataset)
In the examples folder you can find more detailed documentation, ipython notebook tutorials and other integration examples with Pd, music21 and JUCE.