def __init__(self, init_value=0, drift_target_weights=None,
change_rate_weights=None, move_freq_weights=None):
"""
Args:
init_value (float):
drift_target_weights (list): a list of 2-tuple weights
change_rate_weights (list): a list of 2-tuple weights
move_freq_weights (list): a list of 2-tuple weights
"""
# TODO: improve docs
# Set up amplitude
self._raw_value = None
self.value = init_value
if drift_target_weights is None:
self.drift_target_weights = [
(-1, 1), (0.02, 6), (0.2, 1), (0.3, 0)]
else:
self.drift_target_weights = drift_target_weights
self.drift_target = rand.weighted_rand(self.drift_target_weights)
if change_rate_weights is None:
self.change_rate_weights = [(0.00001, 100), (0.001, 5), (0.01, 1)]
else:
self.change_rate_weights = change_rate_weights
self.change_rate = rand.weighted_rand(self.change_rate_weights)
if move_freq_weights is None:
self.move_freq_weights = [(0.000001, 10), (0.01, 1)]
else:
self.move_freq_weights = move_freq_weights
self.move_freq = rand.weighted_rand(self.move_freq_weights)
def re_roll_behaviors(self):
"""
Re-randomly-generate behavior attributes.
Returns: None
"""
self.move_freq = rand.weighted_rand(self.move_freq_weights)
self.drift_target = rand.weighted_rand(self.drift_target_weights)
self.change_rate = rand.weighted_rand(self.change_rate_weights)
def get(self):
"""
Get an ``int`` value in the probability space of the object.
Returns: int
"""
return rand.weighted_rand(self.weights, round_result=True)
def apply_noise(self, noise_weights=None, uniform_amount=0.1):
"""
Add noise to every link in the network.
Can use either a ``uniform_amount`` or a ``noise_weight`` weight
profile. If ``noise_weight`` is set, ``uniform_amount`` will be
ignored.
Args:
noise_weights (Optional[(amount, weight)]): a list of weights
describing the noise to be added to each link
uniform_amount (float): the maximum amount of uniform noise
to be applied if ``noise_weights`` is not set
Returns: None
"""
# Main node loop
for node in self.node_list:
for link in node.link_list:
if noise_weights is not None:
noise_amount = round(weighted_rand(noise_weights), 3)
else:
noise_amount = round(random.uniform(
0, link.weight * uniform_amount), 3)
link.weight += noise_amount
def get(self):
"""
Get an ``int`` value in the probability space of the object.
Returns: int
"""
return rand.weighted_rand(self.weights, round_result=True)
def get(self):
"""
Get a ``float`` value in the probability space of the object.
Returns:
float: A value between the lowest and highest outcomes
in ``self.weights``
"""
return rand.weighted_rand(self.weights, round_result=False)
def get(self):
"""
Get a ``float`` value in the probability space of the object.
Returns:
float: A value between the lowest and highest outcomes
in ``self.weights``
"""
return rand.weighted_rand(self.weights, round_result=False)
def build_chunk(oscillators):
"""
Build an audio chunk and progress the oscillator states.
Args:
oscillators (list): A list of oscillator.Oscillator objects
to build chunks from
Returns:
str: a string of audio sample bytes ready to be written to a wave file
"""
step_random_processes(oscillators)
subchunks = []
for osc in oscillators:
osc.amplitude.step_amp()
osc_chunk = osc.get_samples(config.CHUNK_SIZE)
if osc_chunk is not None:
subchunks.append(osc_chunk)
if len(subchunks):
new_chunk = sum(subchunks)
else:
new_chunk = numpy.zeros(config.CHUNK_SIZE)
# If we exceed the maximum amplitude, handle it gracefully
chunk_amplitude = amplitude.find_amplitude(new_chunk)
if chunk_amplitude > config.MAX_AMPLITUDE:
# Normalize the amplitude chunk to mitigate immediate clipping
new_chunk = amplitude.normalize_amplitude(new_chunk,
config.MAX_AMPLITUDE)
# Pick some of the offending oscillators (and some random others)
# and lower their drift targets
avg_amp = (sum(osc.amplitude.value for osc in oscillators) /
len(oscillators))
for osc in oscillators:
if (osc.amplitude.value > avg_amp and rand.prob_bool(0.1) or
rand.prob_bool(0.01)):
osc.amplitude.drift_target = rand.weighted_rand(
[(-5, 1), (0, 10)])
osc.amplitude.change_rate = rand.weighted_rand(
osc.amplitude.change_rate_weights)
return new_chunk.astype(config.SAMPLE_DATA_TYPE).tostring()
def build_chunk(oscillators):
"""
Build an audio chunk and progress the oscillator states.
Args:
oscillators (list): A list of oscillator.Oscillator objects
to build chunks from
Returns:
str: a string of audio sample bytes ready to be written to a wave file
"""
step_random_processes(oscillators)
subchunks = []
for osc in oscillators:
osc.amplitude.step_amp()
osc_chunk = osc.get_samples(config.CHUNK_SIZE)
if osc_chunk is not None:
subchunks.append(osc_chunk)
if len(subchunks):
new_chunk = sum(subchunks)
else:
new_chunk = numpy.zeros(config.CHUNK_SIZE)
# If we exceed the maximum amplitude, handle it gracefully
chunk_amplitude = amplitude.find_amplitude(new_chunk)
if chunk_amplitude > config.MAX_AMPLITUDE:
# Normalize the amplitude chunk to mitigate immediate clipping
new_chunk = amplitude.normalize_amplitude(new_chunk,
config.MAX_AMPLITUDE)
# Pick some of the offending oscillators (and some random others)
# and lower their drift targets
avg_amp = (sum(osc.amplitude.value
for osc in oscillators) / len(oscillators))
for osc in oscillators:
if (osc.amplitude.value > avg_amp and rand.prob_bool(0.1)
or rand.prob_bool(0.01)):
osc.amplitude.drift_target = rand.weighted_rand([(-5, 1),
(0, 10)])
osc.amplitude.change_rate = rand.weighted_rand(
osc.amplitude.change_rate_weights)
return new_chunk.astype(config.SAMPLE_DATA_TYPE).tostring()
def __init__(self,
init_value,
drift_target_weights=None,
change_rate_weights=None):
"""
Args:
init_value (float): The initial amplitude level
drift_target_weights (list): a list of 2-tuple weights
change_rate_weights (list): a list of 2-tuple weights
"""
# Set up amplitude
self._raw_value = None
self.value = init_value
if drift_target_weights is None:
self.drift_target_weights = [(-1, 1), (0.02, 6), (0.2, 1),
(0.3, 0)]
else:
self.drift_target_weights = drift_target_weights
self.drift_target = rand.weighted_rand(self.drift_target_weights)
if change_rate_weights is None:
self.change_rate_weights = [(0.00001, 100), (0.001, 5), (0.01, 1)]
else:
self.change_rate_weights = change_rate_weights
self.change_rate = rand.weighted_rand(self.change_rate_weights)
def apply_noise(self, noise_weights=None, uniform_amount=0.1):
"""
Add noise to every link in the network.
Can use either a ``uniform_amount`` or a ``noise_weight`` weight
profile. If ``noise_weight`` is set, ``uniform_amount`` will be
ignored.
Args:
noise_weights (list): a list of weight tuples
of form ``(float, float)`` corresponding to
``(amount, weight)`` describing the noise to be
added to each link in the graph
uniform_amount (float): the maximum amount of uniform noise
to be applied if ``noise_weights`` is not set
Returns: None
Example:
>>> from blur.markov.node import Node
>>> node_1 = Node('One')
>>> node_2 = Node('Two')
>>> node_1.add_link(node_1, 3)
>>> node_1.add_link(node_2, 5)
>>> node_2.add_link(node_1, 1)
>>> graph = Graph([node_1, node_2])
>>> for link in graph.node_list[0].link_list:
... print('{} {}'.format(link.target.value, link.weight))
One 3
Two 5
>>> graph.apply_noise()
>>> for link in graph.node_list[0].link_list:
... print('{} {}'.format(
... link.target.value, link.weight)) # doctest: +SKIP
One 3.154
Two 5.321
"""
# Main node loop
for node in self.node_list:
for link in node.link_list:
if noise_weights is not None:
noise_amount = round(weighted_rand(noise_weights), 3)
else:
noise_amount = round(
random.uniform(0, link.weight * uniform_amount), 3)
link.weight += noise_amount
def apply_noise(self, noise_weights=None, uniform_amount=0.1):
"""
Add noise to every link in the network.
Can use either a ``uniform_amount`` or a ``noise_weight`` weight
profile. If ``noise_weight`` is set, ``uniform_amount`` will be
ignored.
Args:
noise_weights (list): a list of weight tuples
of form ``(float, float)`` corresponding to
``(amount, weight)`` describing the noise to be
added to each link in the graph
uniform_amount (float): the maximum amount of uniform noise
to be applied if ``noise_weights`` is not set
Returns: None
Example:
>>> from blur.markov.node import Node
>>> node_1 = Node('One')
>>> node_2 = Node('Two')
>>> node_1.add_link(node_1, 3)
>>> node_1.add_link(node_2, 5)
>>> node_2.add_link(node_1, 1)
>>> graph = Graph([node_1, node_2])
>>> for link in graph.node_list[0].link_list:
... print('{} {}'.format(link.target.value, link.weight))
One 3
Two 5
>>> graph.apply_noise()
>>> for link in graph.node_list[0].link_list:
... print('{} {}'.format(
... link.target.value, link.weight)) # doctest: +SKIP
One 3.154
Two 5.321
"""
# Main node loop
for node in self.node_list:
for link in node.link_list:
if noise_weights is not None:
noise_amount = round(weighted_rand(noise_weights), 3)
else:
noise_amount = round(random.uniform(
0, link.weight * uniform_amount), 3)
link.weight += noise_amount
def step_random_processes(oscillators):
"""
Args:
oscillators (list): A list of oscillator.Oscillator objects
to operate on
Returns: None
"""
if not rand.prob_bool(0.01):
return
amp_bias_weights = [(0.001, 1), (0.1, 100), (0.15, 40), (1, 0)]
# Find out how many oscillators should move
num_moves = iching.get_hexagram('NAIVE') % len(oscillators)
for i in range(num_moves):
pair = [gram % len(oscillators)
for gram in iching.get_hexagram('THREE COIN')]
amplitudes = [(gram / 64) * rand.weighted_rand(amp_bias_weights)
for gram in iching.get_hexagram('THREE COIN')]
oscillators[pair[0]].amplitude.drift_target = amplitudes[0]
oscillators[pair[1]].amplitude.drift_target = amplitudes[1]
def test_normal_distribution(self):
"""
Test the accuracy of ``rand.normal_distribution()``.
Use the curve it creates to generate a large number of samples,
and then calculate the real variance and mean of the resulting
sample group and compare the two within a comfortable margin.
"""
MEAN = -12
VARIANCE = 2.5
STANDARD_DEVIATION = math.sqrt(VARIANCE)
SAMPLE_COUNT = 1000
curve = rand.normal_distribution(MEAN, VARIANCE, weight_count=30)
samples = [rand.weighted_rand(curve) for i in range(SAMPLE_COUNT)]
samples_mean = sum(samples) / len(samples)
samples_variance = (sum(
(s - samples_mean)**2 for s in samples) / len(samples))
mean_diff = abs(MEAN - samples_mean)
variance_diff = abs(VARIANCE - samples_variance)
self.assertLess(mean_diff, abs(MEAN / 4))
self.assertLess(variance_diff, abs(VARIANCE / 4))
def test_weighted_rand_with_arbitrary_curve(self):
"""
Test ``rand.weighted_rand()``.
Find a large number of points from a randomly built weight distribution
and comparing the distribution against the expectation using a crude
histogram model.
"""
MIN_X = -1000
MIN_Y = 0
MAX_X = 1000
MAX_Y = 1000
curve = [(random.randint(MIN_X, MAX_X), random.randint(MIN_Y, MAX_Y))
for i in range(30)]
# Attach points to domain bounds at MIN_Y
curve.append((MIN_X, MIN_Y))
curve.append((MAX_X, MIN_Y))
# Sort the points in the curve as this is a
# requirement of _linear_interp()
curve.sort(key=lambda p: p[0])
BIN_WIDTH = 1
bins = {b: 0 for b in range(MIN_X, MAX_X, BIN_WIDTH)}
TEST_COUNT = 1000
for i in range(TEST_COUNT):
point = rand.weighted_rand(curve, round_result=False)
# Match the found point to the closest bin to the left
bins[int(math.floor(point / BIN_WIDTH) * BIN_WIDTH)] += 1
# Make sure the binning is working as expected
self.assertEqual(sum(values for i, values in bins.items()),
TEST_COUNT,
msg='This test itself is broken! '
'Not all rolled points were matched to a bin.')
sum_probability = 0
for bin_x in bins.keys():
sum_probability += rand._linear_interp(curve, bin_x)
for bin_x, count in bins.items():
bin_probability = rand._linear_interp(curve, bin_x)
expected_count = (bin_probability / sum_probability) * TEST_COUNT
self.assertLess(abs(count - expected_count), TEST_COUNT / 10)
def get(self):
"""
Render the poem as an HTML string.
Returns:
str: the body of the poem in HTML
"""
if rand.prob_bool(self.mutable_chance):
# Render text from a markov graph derived from the source text
word_list = []
word_count = rand.weighted_rand(
self.word_count_weights, round_result=True)
word_graph = Graph.from_file(self.filepath, self.distance_weights)
for i in range(word_count):
word = word_graph.pick().get_value()
word_list.append(word)
else:
# Otherwise, copy source contents literally
source_file = open(self.filepath, 'r')
word_list = source_file.read().split()
# Combine words, process markups, and return HTML
return self.render_markups(word_list)
def test_weighted_rand_with_arbitrary_curve(self):
"""
Test ``rand.weighted_rand()``.
Find a large number of points from a randomly built weight distribution
and comparing the distribution against the expectation using a crude
histogram model.
"""
MIN_X = -1000
MIN_Y = 0
MAX_X = 1000
MAX_Y = 1000
curve = [(random.randint(MIN_X, MAX_X), random.randint(MIN_Y, MAX_Y))
for i in range(30)]
# Attach points to domain bounds at MIN_Y
curve.append((MIN_X, MIN_Y))
curve.append((MAX_X, MIN_Y))
# Sort the points in the curve as this is a
# requirement of _linear_interp()
curve.sort(key=lambda p: p[0])
BIN_WIDTH = 1
bins = {b: 0 for b in range(MIN_X, MAX_X, BIN_WIDTH)}
TEST_COUNT = 1000
for i in range(TEST_COUNT):
point = rand.weighted_rand(curve, round_result=False)
# Match the found point to the closest bin to the left
bins[int(math.floor(point / BIN_WIDTH) * BIN_WIDTH)] += 1
# Make sure the binning is working as expected
self.assertEqual(sum(values for i, values in bins.items()), TEST_COUNT,
msg='This test itself is broken! '
'Not all rolled points were matched to a bin.')
sum_probability = 0
for bin_x in bins.keys():
sum_probability += rand._linear_interp(curve, bin_x)
for bin_x, count in bins.items():
bin_probability = rand._linear_interp(curve, bin_x)
expected_count = (bin_probability / sum_probability) * TEST_COUNT
self.assertLess(abs(count - expected_count), TEST_COUNT / 10)
def step_random_processes(oscillators):
"""
Args:
oscillators (list): A list of oscillator.Oscillator objects
to operate on
Returns: None
"""
if not rand.prob_bool(0.01):
return
amp_bias_weights = [(0.001, 1), (0.1, 100), (0.15, 40), (1, 0)]
# Find out how many oscillators should move
num_moves = iching.get_hexagram('NAIVE') % len(oscillators)
for i in range(num_moves):
pair = [
gram % len(oscillators)
for gram in iching.get_hexagram('THREE COIN')
]
amplitudes = [(gram / 64) * rand.weighted_rand(amp_bias_weights)
for gram in iching.get_hexagram('THREE COIN')]
oscillators[pair[0]].amplitude.drift_target = amplitudes[0]
oscillators[pair[1]].amplitude.drift_target = amplitudes[1]
def test_normal_distribution(self):
"""
Test the accuracy of ``rand.normal_distribution()``.
Use the curve it creates to generate a large number of samples,
and then calculate the real variance and mean of the resulting
sample group and compare the two within a comfortable margin.
"""
MEAN = -12
VARIANCE = 2.5
STANDARD_DEVIATION = math.sqrt(VARIANCE)
SAMPLE_COUNT = 600
curve = rand.normal_distribution(MEAN, VARIANCE)
samples = [rand.weighted_rand(curve) for i in range(SAMPLE_COUNT)]
samples_mean = sum(samples) / len(samples)
samples_variance = (
sum((s - samples_mean) ** 2 for s in samples) /
len(samples)
)
mean_diff = abs(MEAN - samples_mean)
variance_diff = abs(VARIANCE - samples_variance)
self.assertLess(mean_diff, abs(MEAN / 5))
self.assertLess(variance_diff, abs(VARIANCE / 5))
from pprint import pformat
from random import randint
from blur import rand
# import basic (manually entered) data on the poems
from __poems_basic_preconfig import poems
if __name__ == '__main__':
for poem in poems:
# Decide the likelihood that a poem will be markov-ed on view
# Manually entered data had mutable chances of 0, 0.5, and 1
# For 3 basic categories of preference
# (All but poem eight have a chance to be mutable)
if poem['mutable_chance'] == 0 and poem['name'] != 'eight':
poem['mutable_chance'] = rand.weighted_rand([(0, 100), (0.03, 10),
(0.15, 0)])
elif poem['mutable_chance'] == 0.5:
poem['mutable_chance'] = rand.weighted_rand(
rand.normal_distribution(0.5, 0.8, 0, 1))
else:
poem['mutable_chance'] = rand.weighted_rand([(0.85, 0), (0.9, 10),
(1, 100)])
poem['position_weight'] = rand.weighted_rand(
rand.normal_distribution(poem['position_weight'], 3))
# Build distance weights for markov graph derivation
keys = [
rand.weighted_rand(rand.normal_distribution(4, 30), True)
for i in range(randint(10, 20))
]
distance_weights = dict([
(key,
detune_base_pitches_weights = [(frequency_map[10], 50),
(frequency_map[0], 1),
(frequency_map[2], 30),
(frequency_map[3], 40),
(frequency_map[5], 80),
(frequency_map[7], 30),
(frequency_map[9], 20)]
octave_choice_weights = [(1/8, 20),
(1/4, 15),
(1/2, 10),
(1, 5),
(2, 5),
(4, 5)]
# Find detuned pitches
pitches = [((rand.weighted_choice(detune_base_pitches_weights) + # Base pitch
rand.weighted_rand(detune_weights)) * # Detune
rand.weighted_choice(octave_choice_weights)) # Set Octave
for i in range(50)]
amp_multiplier_weights = [(0.05, 10), (0.2, 2), (0.7, 1)]
for pitch in pitches:
osc_list.append(
oscillator.Oscillator(
pitch,
amplitude.AmplitudeHandler(
init_value=0,
drift_target_weights=[
(-2, 30), (0.02, 8), (0.05, 2), (0.1, 0.1), (0.3, 0)],
change_rate_weights=[
(0.00001, 12000),
(0.0001, 100),
def test_weighted_rand_and_choice_with_one_weight_equivalent(self):
weight_list = [('The Only Weight', 2)]
self.assertEqual(rand.weighted_rand(weight_list),
rand.weighted_choice(weight_list))
def test_weighted_rand_and_choice_with_one_weight_equivalent(self):
weight_list = [('The Only Weight', 2)]
self.assertEqual(rand.weighted_rand(weight_list),
rand.weighted_choice(weight_list))
def test_weighted_rand_with_one_weight_returns_it(self):
weight_list = [('The Only Weight', 2)]
expected_result = weight_list[0][0]
self.assertEqual(rand.weighted_rand(weight_list), expected_result)
def render_markups(self, word_list):
"""
Render a list of words and markups to html with automatic line breaks.
This method performs several processing steps preparing the poem text
for HTML delivery. It:
* Converts `---` to stochastic length dashes in the form of empty
`<span class="variable-length-dash"></span>` tags
* Converts `|||` to stochastic height line breaks in the form of
empty `<span class="variable-height-break"></span>` tags
* Spontaneously inserts horizontal blank space between words in the
form of empty `<span class="horizontal-blank-space"></span>` tags
* Calculates the position of line breaks and renders them as divs
in the form `<div class="poem-line"> ... </div>`
Line breaks are triggered after every word which exceeds
`LINE_LENGTH`. This character limit ignores HTML tags,
allowing lines containing spans (variable-length-dash or
horizontal-blank-space) to intentionally visually exceed the apparent
right edge of the poem.
Args:
word_list (list[str]): The list of words (as well as punctuation
marks and markups) to render.
Returns:
str: The contents of `word_list` rendered as HTML
"""
working_text = word_list[:] # Copy of word_list to avoid side-effects
lines = [] # List of lines in the generated poem
current_line = [] # List of words in the current line
visible_char_count = 0 # Number of visible chars in current line
for word in working_text:
if word == '---':
# Render triple dashes to variable length visible dashes
# (in the form of inline-block spans)
dash_length = rand.weighted_rand(self.dash_length_weights)
word = variable_length_dash(dash_length)
elif word == '|||':
# Render triple pipes as variable height breaks
# (in the form of fixed-height spans)
y_gap = rand.weighted_rand(
self.y_gap_height_weights)
word = variable_height_break(y_gap)
else:
# Otherwise, the word will be rendered literally as visible
# text, so count it toward the visible character count used
# in placing line breaks
visible_char_count += len(word)
# Roll to insert x-axis gaps
if rand.prob_bool(self.x_gap_freq):
x_gap = rand.weighted_rand(self.x_gap_length_weights)
# Sometimes place space before word, sometimes after (50/50)
word = horizontal_blank_space(x_gap) + word
# Break lines when LINE_LENGTH is exceeded
if visible_char_count > LINE_LENGTH:
visible_char_count = 0
lines.append(''.join(current_line))
current_line = []
# Handle spaces appropriately for punctuation marks
if word in PUNCTUATIONS:
current_line.append(word)
else:
current_line.append(' ' + word)
# Attach final line
if current_line:
lines.append(''.join(current_line))
return (''.join((surround_with_tag(line, 'div', 'class="poem-line"')
for line in lines)))
def __init__(self,
filename,
immutable_id,
title='',
mutable_chance=None,
position_weight=None,
distance_weights=None,
word_count_weights=None,
gap_before_weights=None,
left_pad_weights=None,
x_gap_freq_weights=None,
x_gap_length_weights=None,
y_gap_height_weights=None,
dash_length_weights=None,
):
"""
Args:
filename (str): Name of text file containing the poem source
located in `SOURCE_DIR`
title (str): Title of the poem
immutable_id (int): The immutable unique ID of this poem.
Unlike `title`, this should not be subject to change by
random processes.
mutable_chance (float): 0-1 probability to be mutable
distance_weights (dict): Dict of distance weights to be
used in rand.from_file()
position_weight (list[tuple]): Weight for position in the
book order. Higher values relative to the values in the other
poems indicates a stronger likelihood to appear near the
beginning
word_count_weights (list[tuple]): List of weight tuples for
how many words will appear in the poem if the poem is mutable
gap_before_weights (list[tuple]): List of weight tuples for
how much space should appear before the poem, in em's.
left_pad_weights (list[tuple]): List of weight tuples for
how far the poem should be padded on the left,
in device-width %.
x_gap_freq_weights (list[tuple]): List of weight tuples for how
frequently x-axis gaps should be inserted between words in
the rendered poem. On initialization, this value is used
to calculate `self.x_gap_freq`, which is the 0-1 probability
for an x-axis gap to be inserted between any two given words.
x_gap_length_weights (list[tuple): List of weight tuples for the
length of inserted x-axis gaps, in em's.
y_gap_height_weights (list[tuple): List of weight tuples for how
tall inserted y-axis gaps should be, in em's.
dash_length_weights (list[tuple]): List of weight tuples for
the length of dashes triggered by `---` marks in the
source text.
"""
self.immutable_id = immutable_id
self.title = title
self.filepath = os.path.join(SOURCE_DIR, filename)
self.mutable_chance = (mutable_chance
if mutable_chance
else _default_mutable_chance)
self.distance_weights = (distance_weights
if distance_weights
else _default_distance_weights)
self.position_weight = (position_weight
if position_weight
else _default_position_weight)
self.word_count_weights = (word_count_weights
if word_count_weights
else _default_word_count_weights)
self.x_gap_length_weights = (x_gap_length_weights
if x_gap_length_weights
else _default_x_gap_length_weights)
self.y_gap_height_weights = (y_gap_height_weights
if y_gap_height_weights
else _default_y_gap_height_weights)
self.dash_length_weights = (dash_length_weights
if dash_length_weights
else _default_dash_length_weights)
# Some args are used to calculate attributes on init
self.gap_before = rand.weighted_rand(
gap_before_weights if gap_before_weights
else _default_gap_before_weights)
self.left_pad = rand.weighted_rand(
left_pad_weights if left_pad_weights
else _default_left_pad_weights)
self.x_gap_freq = rand.weighted_rand(
x_gap_freq_weights if x_gap_freq_weights
else _default_x_gap_freq_weights)
(0.3, 1), (0.4, 0)
])))
# Initialize softer oscillators slightly out of tune with consonant pitches
detune_weights = rand.normal_distribution(0, 20)
detune_base_pitches_weights = [(frequency_map[10], 50), (frequency_map[0], 1),
(frequency_map[2], 30), (frequency_map[3], 40),
(frequency_map[5], 80), (frequency_map[7], 30),
(frequency_map[9], 20)]
octave_choice_weights = [(1 / 8, 20), (1 / 4, 15), (1 / 2, 10), (1, 5), (2, 5),
(4, 5)]
# Find detuned pitches
pitches = [
((
rand.weighted_choice(detune_base_pitches_weights) + # Base pitch
rand.weighted_rand(detune_weights)) * # Detune
rand.weighted_choice(octave_choice_weights)) # Set Octave
for i in range(50)
]
amp_multiplier_weights = [(0.05, 10), (0.2, 2), (0.7, 1)]
for pitch in pitches:
osc_list.append(
oscillator.Oscillator(
pitch,
amplitude.AmplitudeHandler(
init_value=0,
drift_target_weights=[(-2, 30), (0.02, 8), (0.05, 2),
(0.1, 0.1), (0.3, 0)],
change_rate_weights=[(0.00001, 12000), (0.0001, 100),
(0.001, 10)],
def test_weighted_rand_with_one_weight_returns_it(self):
weight_list = [('The Only Weight', 2)]
expected_result = weight_list[0][0]
self.assertEqual(rand.weighted_rand(weight_list), expected_result)
