def play():
# in example 14 we showed how to nest patterns in ways that are easier
# to conceptualize. For instance, these patterns play in order, and we
# call that a movement. However, it's often tedious to specify things
# that replay again and again. To do make this easy, we have a new
# selector to introduce - Repeatedly!
output = Performance(bpm=240, stop_seconds=100)
scale_choices = Endlessly([scale("D minor")])
source = ScaleSource(scales=scale_choices)
pattern1 = Roman(symbols=Repeatedly("1 2 3 4".split(), cycles=2))
pattern2 = Roman(symbols="4 3 2 1".split())
movement1 = Ordered(sources=[pattern1, pattern2])
pattern3 = Roman(symbols=Repeatedly("1 4 1 4 1 5 1 5".split(), cycles=2))
pattern4 = Roman(symbols="I IV V I".split())
movement2 = Ordered(sources=[pattern3, pattern4])
suite = Ordered(sources=[movement1, movement2], channel=1)
source.send_to(suite)
# for this example, we won't do anything crazy with transpositions or anything.
suite.send_to(output)
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# examples/06_subdivide showed how to chop up the incoming quarter note signal
# to create faster signals. However, sometimes in a given bar we want a mix of note durations.
# how do we do that?
# For this example, lets play whole note chords underneath a quarter note melody.
# implementation caveat:
# Because the "beat" signal from the conductor comes in every quarter note, we have to take
# care and rest in the right places of the input signal to avoid funness. This may get easier
# later.
# this is pretty close to what we did in 09_harmony.py
output = Performance(bpm=120, stop_seconds=15)
# this performance will actually complete so stop_seconds is mostly ignorable.
scale_choices = Endlessly([scale("D major")])
source = ScaleSource(scales=scale_choices)
# the underlying chords - note the rests
chords = Roman(symbols=Endlessly(
"I - - - I - - - IV - - - V - - - ".split()),
channel=1)
# every beat is a quarter note, but on the first beat in a cycle of 4
# play a WHOLE note
chords_len = Duration(lengths=Endlessly([1, 0, 0, 0]))
source.chain([chords, chords_len, output])
# the melody
# TODO: Roman should figure out if input is a string here and auto-split.
pattern2 = Roman("1 2 3 1 2 2 1 3 3 1 3 2".split(), channel=2)
pattern3 = Roman("1 3 3 1 4 4 1 3 3 1 5 -".split(), channel=2)
pattern4 = Roman("1 5 5 1 4 4 1 3 3 1 2 2".split(), channel=2)
pattern5 = Roman("1 2 4 6 4 2 1 6 4 2 1 -".split(), channel=2)
melody_sequence = [pattern2, pattern3, pattern4, pattern5]
melody = Ordered(sources=Endlessly(melody_sequence))
# we could just do: source.chain(source, ordered) here
# but it would mean later we would have less flexibility
# to make the patterns feel different.
source.chain([melody])
for pattern in melody_sequence:
source.chain([pattern, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# Let's show how to play a sequence of one length but omit every Nth note
output = Performance(bpm=120, stop_seconds=15)
source = ScaleSource(scales=scale("D minor"))
melody = Roman(symbols=Endlessly("1 2 3 4 5 6 7".split()), channel=1)
silence = Permit(when=Endlessly([1, 1, 0]))
# so we're playing a pattern of 1 2 3 4 5 6 7 of scale degree notes in a giant loop
# the first time through what really happens is:
# 1 2 <REST> 4 5 <REST> 7
# and the second time:
# 1 <REST> 3 4 <REST> 6 7
# and the third time:
# <REST> 2 3 <REST> 5 6 <REST>
# all of this is because the pattern and the 'silence' pattern are of unequal lengths
# this is not ALWAYS relevant in a composition, but if you want to do something
# polyrhytmic and unorthodox (not just with 'Silence') patterns of unequal length
# can be interesting.
# HOMEWORK: experiment with changing the melody patterns and the silence patterns
# HOMEWORK: change endlessly to randomly, and see what happens
source.chain([melody, silence, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def _build_players(self):
for (instrument_name, pattern_list) in self.patterns.items():
instrument = self._factory.instruments[instrument_name]
channel = instrument.channel
notation = instrument.notation
sources = []
print("PATTERN LIST= %s" % pattern_list)
if isinstance(pattern_list, str):
pattern_list = [ pattern_list ]
for pattern_name in pattern_list:
pattern = self._factory.patterns.get(pattern_name, None)
if pattern is None:
raise Exception("pattern not found: %s" % pattern_name)
real_pattern = None
if notation == 'roman':
real_pattern = Roman(symbols=pattern)
print("RP=%s" % pattern)
elif notation == 'literal':
real_pattern = Literal(symbols=pattern)
else:
raise Exception("unknown notation type for instrument: %s" % instrument_name)
sources.append(real_pattern)
self._players[instrument_name] = Ordered(sources=sources)
def play():
# now modifying our previous example, instead of playing the same two
# scales in a loop, let's play a given set of notes in each scale.
# We'll use roman notation to play the 1st, 4th, and 5th note in the scale
# followed by the 1st, 4th, and 5th major chord
# finally, we'll play the 2nd, 3rd, and 6th minor chord.
output = Performance(bpm=120, stop_seconds=10)
# this is just as before, playing one scale then the other in a loop of 7
# notes each
scale1 = scale("c6 major")
scale2 = scale("c6 minor")
scale_choices = [ dict(scale=scale1, beats=7), dict(scale=scale2, beats=7) ]
source = ScaleSource(scales=Endlessly(scale_choices))
# the scale follower will play the first 7 notes in each scale, whatever the current
# scale is. Note that a scale change that doesn't quite line up with the length
# of the roman pattern rolling over might sound a bit weird. That's ok.
roman = Roman(symbols=Endlessly("1 4 5 I IV V ii iii vi".split()), channel=1)
source.chain([roman, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# in example 13 we showed the use of Ordered to trigger
# one pattern after another. As programs grow, it might be nice
# to group them conceptually in a larger composition.
# This is a demo that shows how Ordered can nest to not work only
# with basic patterns, but also larger movements.
# The idea is this: Generators stack!
# Like before, we're setting stop seconds to 100 seconds but this composition
# will complete due to the sources being exhausted first, which is the point.
output = Performance(bpm=240, stop_seconds=100)
for scale_name in ["C major", "D minor"]:
scale_choices = Endlessly([scale(scale_name)])
source = ScaleSource(scales=scale_choices)
pattern1 = Roman(symbols="1 2 3 4 5 6 7".split())
pattern2 = Roman(symbols="i ii iii iv v v vii".split())
movement1 = Ordered(sources=[pattern1, pattern2])
pattern3 = Roman(symbols="7 6 5 4 3 2 1".split())
pattern4 = Roman(symbols="vii vi v iv iii ii i".split())
movement2 = Ordered(sources=[pattern3, pattern4])
movement2_transposer = Transpose(octaves=Endlessly([-2]))
movement2.send_to(movement2_transposer)
suite = Ordered(sources=[movement1, movement2], channel=1)
source.send_to(suite)
output.listens_to([movement1, movement2_transposer])
# BONUS TIP: we technically don't have to have just one conductor invocation, if it
# keeps it simple.
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# here's an example that might *START* to approximate a song.
# here we have two instruments playing on two different tracks.
output = Performance(bpm=60, stop_seconds=10)
# both instruments will use the same scale at the same time, but there is a scale
# change as we cycle between scales every 24 beats
scale1 = scale("c5 major_pentatonic")
scale2 = scale("e5 mixolydian")
scale_choices = [
dict(scale=scale1, beats=24),
dict(scale=scale2, beats=24)
]
source = ScaleSource(scales=Endlessly(scale_choices))
# the first instrument plays a series of chords, transposed down an octave
# from the original scale carrier signal.
roman1 = Roman(symbols=Endlessly("I IV V IV III:dim ii".split()),
channel=1)
transpose1 = Transpose(octaves=-1)
source.chain([roman1, transpose1, output])
# the second instrument plays a series of notes, but is responding to sixteenth
# notes, not quarter notes, because of the subdivide.
subdivide2 = Subdivide(splits=4)
roman2 = Roman(symbols=Endlessly("1 4 3 4 4 3 2 1".split()), channel=2)
source.chain([subdivide2, roman2, output])
# homework assignment:
# change subdivide2 so it's chained BEFORE roman2
# what happens and why?
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# we've focussed a lot on randomness now. So far, we've shown how to pick random values
# for notes and how to conditionally do something or not with "when=".
# What's interesting though is that, due to the way some effects plugins work, we can choose
# to intermix effects. Here, we show that there are really two patterns going on, but only one
# is going to win and actually get to play a note.
# we're relying one 'musician' object stomping on another, but only sometimes.
output = Performance(bpm=120, stop_seconds=15)
source = ScaleSource(scales=scale("Ab blues"))
pattern1 = Endlessly("1 2 3 4 5 6 7".split())
pattern2 = Randomly(
"VII:power VI:power V:power IV:power III:power II:power I:power".split(
))
cut_over = Randomly([0, 0, 0, 1, 1, 1], mode='probability')
melody1 = Roman(symbols=pattern1, channel=1)
melody2 = Roman(symbols=pattern2, channel=2, when=cut_over)
transpose = Transpose(octaves=Endlessly([-2, -1, 0, 1, 2]))
# there is no sane reason to want to do this, but what happens is that for the first three notes we are guaranteed
# to play pattern1. Then three notes of pattern 2. The last note is acutally a toss up between pattern1 and pattern2.
# since they live in the same chain, pattern1 is going to overwrite pattern2 even though they are expressed in a chain.
# we can also even choose to overwrite the MIDI channel, which we do, all while sharing a transposition cycle.
# again, this isn't realistic for most compositions, we'd likely just set up a bunch of patterns with RESTS, but
# I wanted to show all the possibilities before leaving the theme of randomness for a while. Hopefully the
# idea of "when=" is now drummed in.
source.chain([melody1, melody2, transpose, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# here we aren't introducing any new concepts, but we're arranging some
# concepts a little differently. This is somewhat of a review, as we're going
# to start exploring some more (hopefully) musical demos in our examples now.
# This one sounds just a bit Perrey and Kingsley with the right VST.
output = Performance(bpm=60, stop_seconds=10)
# let's play in the A major scale
scale_choices = Endlessly([dict(scale=scale("a4 major"))])
source = ScaleSource(scales=scale_choices)
# we're going to alternate between the 1st, 4th, and 5th notes
# of the scale, but we didn't just jump to chords because of what
# we are about to do with the arp.
roman = Roman(symbols=Endlessly("1 4 5".split()), channel=1)
# and now for a clever use of the arp, to machine gun 4 repititions of
# each chord, with rests in between. Subdivide alone couldn't do this.
# Also notice this is running with NO tranpositions.
arp = Arp(
# no transpositions
semitones=Endlessly([0, 0, 0, 0, 0, 0, 0, 0]),
# every beat gets sliced up 8 times, no exceptions
splits=Endlessly([8]),
# try uncommenting this next line:
#octaves=Endlessly([0,0,1,0,2,0,3,0]),
# play every other note on the arp
# TODO: IDEA: seems like we should also allow arp velocity!
rests=Endlessly([0, 1, 0, 1, 0, 1, 0, 1]))
# now the output here is just machine gunned notes. Turn it into major chords.
chordify = Chordify(types=Endlessly(['major']))
source.chain([roman, arp, chordify, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# Previously, in 11_randomness we introduced the default 'choice' random mode, which just endlessly
# picks something from a list.
# in 18_randomness2.py we introduced 'probability', which returns True or False depending on the chances
# given. It's used with the "when" statement, shown in both 18_randomness2.py and 19_skipover.py for doing
# some rather fun things.
# Some more types of randomness are "exhaust" and "human_random".
# just to prove this performance will terminate by reaching the END of the performance, we're setting the stop
# seconds to a crazy long amount of time. This one won't loop endlessly, because of the way 'exhaust' mode works.
output = Performance(bpm=120, stop_seconds=9999)
# exhaust takes a list and feeds it in random order. It would be the same as shuffle sorting the list, basically, and then
# popping off the first element repeatedly. The result is all items will be used, in a random order, and then the sequence
# will stop.
#
# We normally write this in a one liner, but we'll break it up in this example to make things a bit more clear
scale_choices = [
scale("Ab blues"),
scale("C major"),
scale("Eb mixolydian")
]
scale_chooser = Randomly(scale_choices, mode='exhaust')
source = ScaleSource(scales=Randomly(scale_choices, mode='exhaust'))
# human_random is inspired by the Apple iTunes problem. In a party shuffle mode, people may get suprised to learn
# that the system played three Van Halen songs in a row. Apple changed the system to switch between artists more frequently.
# in CAMP, human random will play each item in a list once, before going back and starting over with a new random selection.
note_choices = "1 2 3 4 5 6 7".split()
note_chooser = Randomly(note_choices, mode='human_random')
melody = Roman(symbols=note_chooser, channel=1)
# combining these two concepts together, we are going to play 3 random scales in sequence, and then FOR EACH SCALE, play
# the scale notes in random order.
source.chain([melody, chordify, transpose, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# in example 11 we showed off some random functionality
# in example 17 we showed how to silence notes based on that
# knowledge of 'random'.
# here is a probabilistic way to use random, that combines the concepts.
output = Performance(bpm=120, stop_seconds=15)
source = ScaleSource(scales=scale("Ab blues"))
melody = Roman(symbols=Endlessly("1 2 3 4 5 6 7".split()), channel=1)
silence = Permit(when=Randomly([1, 0.5], mode='probability'))
# so we're playing a scale pattern, but every other note has a 50% chance
# of not being played.
source.chain([melody, silence, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def test_simple_band(self):
output = Performance(bpm=120, stop_seconds=10)
scale1 = scale("c6 major")
scale2 = scale("c6 minor")
source = ScaleSource(
scales=[dict(scale=scale1, beats=7),
dict(scale=scale2, beats=7)])
subdivide = Subdivide(splits=[4])
roman = Roman(symbols="1 2 3 4 I IV V iii".split(), channel=1)
follower = ScaleFollower(lengths=[7])
chordify = Chordify(types=['power'])
shift = Transpose(octaves=[-3], channel=2)
source.chain([subdivide, roman, output])
source.chain([follower, chordify, shift, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# this is sort of a variation of example 08. Except now, rather than
# playing two different patterns that might clash, we're selecting chord
# patterns for instrument 2 to harmonize with what instrument 1 is playing.
output = Performance(bpm=120, stop_seconds=10)
# both instruments will use the same scale at the same time, but there is a scale
# change as we cycle between scales every 24 beats
scale1 = scale("c5 major_pentatonic")
scale2 = scale("e5 mixolydian")
scale_choices = [ dict(scale=scale1, beats=24), dict(scale=scale2, beats=24) ]
source = ScaleSource(scales=Endlessly(scale_choices))
# the first instrument plays a series of notes
roman1 = Roman(symbols=Endlessly("1 2 4 1 2 3 1 2 4 3 3".split()), channel=1)
source.chain([roman1, output])
# the second instrument transposes that note down two octaves and plays a power
# chord. We could pass in an array of chords to vary the chord type, but this is
# probably going to sound less prone to clashing.
chordify = Chordify(types=Endlessly(["power"]), channel=2)
transpose = Transpose(octaves=Endlessly([-2]))
source.chain([roman1, chordify, transpose, output])
# note that roman 1 is part of each chain, but the channel number is overridden
# in the second set. This can be done because the event objects are copied as they
# are passed between each layer. Technically the channel can be overriden at any
# time. Ideas for future chaos, perhaps?
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# Permit, introduced in the example 18_randomness2.py allows us to
# silence an entire chain when certain conditions are met.
# However, sometimes, we may want to decide to apply an affect only
# when something is true. Because this could be useful ANYWHERE
# the same type of "when" logic can actually be attached to ANYTHING.
# here is a probabilistic way to use random, that combines the concepts.
output = Performance(bpm=120, stop_seconds=15)
source = ScaleSource(scales=scale("Ab blues"))
melody = Roman(symbols=Endlessly("1 2 3 4 5 6 7".split()), channel=1)
chordify = Chordify(types=Endlessly(['major', 'minor']),
when=Randomly([0, 0.45], mode='probability'))
transpose = Transpose(octaves=Endlessly([2, -2]),
when=Randomly([0, 0, 0.75], mode='probability'))
# the result is every other note has a 40% chance of becoming a chord, which is always alternating
# major and minor when it happens
# every THIRD note has a 75 percent chance of being transposed, which will be alternating +2/-2 octaves
# when it happens
# so now, we have easily inserted CONDITIONAL effects. The use of when=Randomly wasn't required.
# we could also have used Endlessly or Repeatedly. Though keep in mind if using Repeatedly, when
# the event chain is exhausted, that particular part of the performance will stop. And when everything stops,
# the performance is done. Because this is likely being applied to an effect chain, Repeatedly probably
# doesn't make the most sense with "when". But Endlessly? Sure!
source.chain([melody, chordify, transpose, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# so far what we've done has been mostly theoretical.
# to construct larger songs, obviously, one thing has to occur
# after another. To do this, we could just pass in GIGANTIC
# arrays to everything but that would suck!
# now, everything in CAMP is built off the idea of generators.
# this means the system can pretty well know if one thing is done
# playing and more on to the next.
# while we could just pass in huge note arrays, that kind of sucks.
# What if we want a flute to play
# it's part and then have the trombone come in? To do that,
# we need a facility for ordering - to say "this comes next"
# NOTE: we're setting stop seconds to 100 seconds but this composition
# will complete due to the sources being exhausted first, which is the point.
# we're not writing an endlessly generative piece here, but moving towards
# more explicit songwriting tools. The point of this example is to show
# we are writing a finite song in PIECES.
output = Performance(bpm=240, stop_seconds=100)
scale_choices = Endlessly([scale("a3 major")])
source = ScaleSource(scales=scale_choices)
pattern1 = Roman(symbols="1 2 3 4".split())
pattern2 = Roman(symbols="I I I I".split())
pattern2_transpose = Transpose(octaves=Endlessly([2]))
pattern2.send_to(pattern2_transpose)
# in this example, we're going to play pattern1 until it is exhausted
# then pattern2, and when that's done, we're done.
ordered = Ordered(sources=[pattern1, pattern2], channel=1)
# HOMEWORK ASSIGNMENT:
# uncomment this line to randomly pick a pattern and play it until
# it is exhausted, but keep doing that endlessly. Patterns might
# repeat, see notes about advanced usages of Randomly in examples/11_randomness.py
#
# ordered = Ordered(sources=Randomly([pattern1,pattern2]), channel=1)
# HOMEWORK ASSIGNMENT:
#
# uncomment this next line to play pattern1, pattern2 in an endless sequence
# ordered = Ordered(Endlessly([pattern1,pattern2]))
# TODO: IDEA: it might be nice to have a "beats" flag on ordered where it can stop
# producing after a certain number of beats
# ordered = Ordered(Randomly([pattern1,pattern2,pattern3]), beats=24)
# this would allow for patterns of patterns, and better nesting.
# TODO: IDEA: anything that can take a "beats" should really take a bars=, it's just easier
# to think about
# HOMEWORK: using a pattern in Ordered that uses Endlessly will cause the pattern to NOT switch.
# Try this out and understand why pattern 2 will never play.
# pattern1 = Roman(symbols=Endlessly("3 6 3 6 5 1 2".split()))
# ordered = Ordered([pattern1,pattern2,pattern3])
# TODO: IDEA, make the above homework example actually work, with something like:
# pattern1 = Roman(symbols=Endlessly("3 6 3 6 5 1 2".split()))
# ordered = Ordered([ dict(pattern=pattern1, beats=8), dict(pattern=pattern2, beats=8) ])
# NOTE: in simple cases we could connect ordered to the output, but we'd miss the transposition,
# and I wanted to show off a more complete example. This also underscores the idea that source.chain
# is just syntactic sugar and you don't have to use it.
source.send_to(ordered)
pattern1.send_to(output)
pattern2_transpose.send_to(output)
conductor = Conductor(signal=[source], performance=output)
conductor.start()
def play():
# in example 15 we show how to drift between different patterns.
# so far, we've mostly been showing patterns inside a scale, because
# that sounds good for melodic instruments. Not everything lives
# inside a scale though - accidentals etc. Transpose and the arp
# can move by semitones to overcome that.
# however, when dealing with drum kits, the scale is pretty much never
# relevant
# Usually these things trigger with something like C3 for a kick drum, D4
# for a Tom, and so on -- but it varies by drumkit.
# Having a transpose or scale change in there would
# mess it all up.
# Here's an example of switching through some drum patterns, and it also
# fires multiple types of drum hits on a single channel.
# when setting this example up, put any synth you want on MIDI channel 2,
# but on MIDI channel 1 put some kind of drumkit. You may need to change
# the MIDI note names to make it sound right
K = "F4"
H = "A4"
C = "Bb4"
output = Performance(bpm=120, stop_seconds=15)
melody_trigger = ScaleSource(scales=scale("D minor"))
melody = Roman(symbols=Endlessly("1 - - - - 2 - - - -".split()), channel=1)
drum_trigger = Subdivide(splits=4,
channel=2) # gimme 16th notes for the drum tracks
# kick drum on the quarter notes
kicks = Literal(symbols=Endlessly([
K, None, None, None, K, None, None, None, K, None, None, None, K, None,
None, None
]))
# hihat on the 8th notes every other bar
# but wait one bar before starting that up.
hihat = Ordered(sources=Endlessly([
Literal(symbols="- - - - - - - - - - - - - - - -".split()),
Literal(symbols=Endlessly([
H, None, H, None, H, None, H, None, H, None, H, None, H, None, H,
None
]))
]))
# cymbals on the half notes
cymbals = Literal(symbols=Endlessly([
C, None, None, None, None, None, None, C, None, None, None, None, None,
None
]))
melody_trigger.send_to(melody)
drum_trigger.send_to([kicks, hihat, cymbals])
output.listens_to([melody, kicks, hihat, cymbals])
conductor = Conductor(signal=[melody_trigger, drum_trigger],
performance=output)
conductor.start()
def play():
# in example 10 we showed you how powerful the arpeggiator can get.
# well, that's great for some degrees of chaos, but what if we want
# to let the computer make MORE decisions?
output = Performance(bpm=120, stop_seconds=10)
# randomness is one way to do that.
# in example 03 we showed how to use Endlessly to loop over a set of input.
# that used SELECTORS, but Endlessly is just a LINEAR selector that loops
# when it gets to the end of the input.
# randomness can also be accessed, using the Randomly selector, which
# chooses an item from a list.
# each choice is then repeated for a given number of times.
# here lets pick some random scales, but whatever we pick, hold that choice
# for 7 beats, which we covered in example 01.
random_scale_choices = Randomly([
# each of these items are scale specifications. You've seen these in the very first examples.
dict(scale=scale("c6 major"), beats=7),
dict(scale=scale("Db4 minor_pentatonic"), beats=7),
dict(scale=scale("Eb3 mixolydian"), beats=7)
])
# note that the random selector could repeat a selection as currently
# implemented and/or specified. See ideas below.
# note if we want a random choice to be made EVERY beat we can simplify
# our usage of randomly like so:
random_note_choices = Randomly("1 2 3 4 5 6 7".split())
# what does that do? It randomly plays one of 7 notes in the current scale.
# that's syntactically clean, but super basic. Let's override that with something
# showing more options available to Randomly.
# we're playing CHORDS now, and if we choose the "I" chord we are going to play
# it twice in the current scale (which might change after we play the first, so this is
# a minor lie). All within the current scale, of course.
random_note_choices = Randomly([
dict(value="I", hold=2),
dict(value="IV", hold=1),
dict(value="V", hold=1),
dict(value="i", hold=1),
dict(value="ii", hold=1),
dict(value="iv:dim7", hold=1),
])
# HOMEWORK: add some more scales to the random_scale_choices above and take out some
# notes from random_note_choices. Mix scales with notes!
# FUTURE IDEA: also support a notion of frequency controlling what will be drawn.
# frequencies should add up to 1. This would still be compatible with "hold" above.
#random_note_choices = Randomly([
# dict(value="I", frequency=0.4),
# dict(value="IV", frequency=0.2),
# dict(value="V", frequency=0.2),
# dict(value="i", frequency=0.1),
# dict(value="ii", frequency=0.05),
# dict(value="iv:dim7", frequency=0.05),
#])
# for more examples, see the demos labelled randomness2.py, randomness3.py, etc.
# we want to cover some other things first, so I'm not going to go completelyinto
# randomness now.
source = ScaleSource(scales=random_scale_choices)
roman = Roman(symbols=random_note_choices, channel=1)
source.chain([roman, output])
conductor = Conductor(signal=[source], performance=output)
conductor.start()