Repeating pattern sequenceΒΆ
Run the cochlea on a sequence composed of 1 repeating pattern This pattern of 50ms appears 10 times and each repetition is separated by a noise segment (i.e. a non-repeating pattern)
import os
# import matplotlib
# matplotlib.use('TkAgg')
import numpy as np
from scipy.io import wavfile
import seaborn as sns
from simplecochlea import Cochlea
import simplecochlea
sns.set_context('paper')
Load the file
root_dirpath = os.path.dirname(simplecochlea.__file__)
sample_data_dir = os.path.join(root_dirpath, 'sample_data')
fs, sequence = wavfile.read(os.path.join(sample_data_dir, 'sample_sequence_10_50ms_1.wav'))
Create the cochlea
fmin, fmax, freq_scale, n_channels = 200, 8000, 'erbscale', 100
comp_factor, comp_gain = 0.3, 1.5
tau, v_thresh, v_spike = np.linspace(0.001, 0.0004, n_channels), np.linspace(0.3, 0.17, n_channels), 0.5
cochlea = Cochlea(n_channels, fs, fmin, fmax, freq_scale, comp_factor=comp_factor, comp_gain=comp_gain,
lif_tau=tau, lif_v_thresh=v_thresh, lif_v_spike=v_spike)
Run the sequence through the cochlea
spikelist_seq, _ = cochlea.process_input(sequence)
Out:
Function : process_input - Time elapsed : 0.642064094543457
Plot the spikelist
spikelist_seq.plot()
We know the repeating pattern is repeating every 50ms, the sequence starts with a noise segment and in total, there are 20 segments (10 time the pattern and 10 interleaved noise segments). Thus we can set the pattern_id of the spikes in the output spikelist, with the set_pattern_id_from_time_limits method.
chunk_duration, n_chunks = 0.050, 20
t_start = np.arange(0, chunk_duration*n_chunks, chunk_duration)
t_end = t_start + chunk_duration
pattern_id = [1, 2] * 10
pattern_names = {1: 'Noise', 2: 'Pattern'}
spikelist_seq.set_pattern_id_from_time_limits(t_start, t_end, pattern_id, pattern_names)
Replot the spikelist to see the results :
spikelist_seq.plot()
Total running time of the script: ( 0 minutes 2.376 seconds)