In [109]:
"""
Rhythm processing model
"""
%matplotlib inline
from __future__ import division
from time import time
import sys
sys.path.append('../') # needed to run the examples from within the package folder
import numpy as np
from scipy.signal import hilbert
from scipy.io import loadmat
from pygrfnn.network import Model, make_connections, modelFromJSON
from pygrfnn.oscillator import Zparam
from pygrfnn.grfnn import GrFNN
import matplotlib.pyplot as plt
from pygrfnn.vis import plot_connections
from pygrfnn.vis import tf_detail
from pygrfnn.vis import GrFNN_RT_plot
from pyrhythm.library import get_pattern
from daspy import Signal
from daspy.processing import onset_detection_signal, super_flux_onset_signal
RT_display = False
def get_stimulus(pattern_name="iso",
tempo=120.0,
reps=6.0,
lead_silence=0.0,
sr=16000.0,
click_freq=1200.0,
with_beat=False,
beat_freq=1800.0,
accented=False,
win_len=2048,
hop_size=512,
mel_bands=64
):
p = get_pattern(pattern_name)
x, _ = p.as_signal(tempo=tempo,
reps=reps,
lead_silence=lead_silence,
sr=sr,
click_freq=click_freq,
with_beat=with_beat,
beat_freq=beat_freq,
accented=accented)
x = Signal(x, sr=sr)
tx = x.time_vector()
s = onset_detection_signal(x, win_len=win_len, hop_size=hop_size, mel_bands=mel_bands)
# fps = 120
# s = Signal(super_flux_onset_signal(x, fps=fps), sr=fps)
rms = np.sqrt(np.sum(s**2)/len(s))
s *= 0.06/rms
s = Signal(hilbert(s), sr=s.sr)
t = s.time_vector()
dt = 1/s.sr
# print "SR: ", s.sr
return s, t, dt, x, tx
In [111]:
pattern_name = "bossa"
s, t, dt, x, tx = get_stimulus(pattern_name,
tempo=120.0,
reps=2,
hop_size=128,
win_len=2048,
)
plt.figure()
plt.figure(figsize=(14,4))
plt.plot(t, np.real(s), zorder=10)
plt.plot(tx, x/30, alpha=0.3)
Out[111]:
In [113]:
def rhythm_model():
rhythm_model_definition = {
"name": "Sensory Motor Rhythm model",
"layers": [
{
"name": "sensory network",
"zparams": {
"alpha": 0.00001,
"beta1": -2.0,
"beta2": -2.0,
"delta1": 0.0,
"delta2": 0.0,
"epsilon": 1.0
},
"frequency_range": [0.375, 12.0],
"num_oscs": 320,
"stimulus_conn_type": "allfreq",
"input_channel": 0
},
# {
# "name": "motor network",
# "zparams": {
# "alpha": -0.4,
# "beta1": 1.75,
# "beta2": -1.25,
# "delta1": 0.0,
# "delta2": 0.0,
# "epsilon": 1.0
# },
# "frequency_range": [0.375, 12.0],
# "num_oscs": 321,
# "stimulus_conn_type": "active"
# }
],
"connections": [
{
"source_name": "sensory network",
"target_name": "sensory network",
"modes": [0.333333333333, 0.5, 1, 2.0, 3.0],
"amps": [1, 1, 1, 1, 1],
"strength": 1.0,
"range": 1.05,
"connection_type": "2freq",
"self_connect": False,
"weight": 0.25
},
# {
# "source_name": "sensory network",
# "target_name": "motor network",
# # "modes": [0.333333333333, 0.5, 1, 2.0, 3.0],
# # "amps": [1, 1, 1, 1, 1],
# "modes": [1.0],
# "amps": [1.0],
# "strength": 1.25,
# "range": 1.05,
# "connection_type": "allfreq", #"2freq",
# "self_connect": True,
# "weight": 0.4
# },
# {
# "source_name": "motor network",
# "target_name": "motor network",
# "modes": [0.333333333333, 0.5, 1, 2.0, 3.0],
# "amps": [1, 1, 1, 1, 1],
# "strength": 1.0,
# "range": 1.05,
# "connection_type": "2freq",
# "self_connect": False,
# "weight": 0.1
# },
# {
# "source_name": "motor network",
# "target_name": "sensory network",
# "modes": [0.333333333333, 0.5, 1, 2.0, 3.0],
# "amps": [1, 1, 1, 1, 1],
# "strength": 0.2,
# "range": 1.05,
# "connection_type": "2freq",
# "self_connect": True,
# "weight": 0.05
# }
]
}
return modelFromJSON(rhythm_model_definition)
In [114]:
gain = 0.25
model = rhythm_model()
tic = time()
model.run(gain*s, t, dt)
print "Run time: {:0.1f} seconds".format(time() - tic)
TF = model.layers()[-1].Z
f = model.layers()[-1].f
#
plt.figure()
tf_detail(TF, t, f, t_detail=t[-1])
# mean field
r = np.sum(TF, 0)
rms = np.sqrt(np.sum(r*np.conj(r))/len(r))
r *= 0.06/rms
plt.figure()
plt.plot(t, np.real(r))
plt.plot(t, np.real(s))
Out[114]:
In [ ]:
In [ ]: