if __name__ == "__main__":
H5_dico = read_data_header()
Males = H5PY_RW()
Males.open_h5_dataset('test.h5py', subset=males_keys(H5_dico))
Males.set_chunk(config.chunk_size)
Males.shuffle()
print 'Male voices loaded: ', Males.length(), ' items'
Females = H5PY_RW()
Females.open_h5_dataset('test.h5py', subset=females_keys(H5_dico))
Females.set_chunk(config.chunk_size)
Females.shuffle()
print 'Female voices loaded: ', Females.length(), ' items'
Mixer = Mixer([Males, Females])
das_model = DAS(S=len(Mixer.get_labels()), T=config.chunk_size)
das_model.init()
for i in range(100):
print 'Step #', i
X, Y, Ind = Mixer.get_batch(64)
x_mixture = []
for x in X:
_, x_recons = istft_(x.T)
x_mixture.append(x_recons)
X = X[:, :, :128]
import numpy as np
import tensorflow as tf
import config
import os
H5_dic = read_metadata()
chunk_size = 512*40
males = H5PY_RW('test_raw.h5py', subset = males_keys(H5_dic))
fem = H5PY_RW('test_raw.h5py', subset = females_keys(H5_dic))
print 'Data with', len(H5_dic), 'male and female speakers'
print males.length(), 'elements'
print fem.length(), 'elements'
mixed_data = Mixer([males, fem], chunk_size= chunk_size, with_mask=False, with_inputs=True)
####
#### PREVIOUS MODEL CONFIG
####
N = 256
max_pool = 256
batch_size = 32
learning_rate = 0.01
config_model = {}
config_model["type"] = "DPCL_train_front"
config_model["batch_size"] = batch_size
file = 'test_raw_16k.h5py'
H5_dic = read_metadata()
chunk_size = 512 * 10
males = H5PY_RW(file,
subset=males_keys(H5_dic)).set_chunk(chunk_size).shuffle()
fem = H5PY_RW(file,
subset=females_keys(H5_dic)).set_chunk(chunk_size).shuffle()
print 'Data with', len(H5_dic), 'male and female speakers'
# Mixing the dataset
from data.dataset import Mixer
mixed_data = Mixer([males, fem], with_mask=False, with_inputs=True)
# Training set selection
mixed_data.select_split(0)
# Model pretrained loading
N = 256
max_pool = 128
batch_size = 8
learning_rate = 0.001
config_model = {}
config_model["type"] = "pretraining"
config_model["batch_size"] = batch_size
import config
import os
H5_dic = read_metadata()
chunk_size = 512 * 40
males = H5PY_RW('test_raw.h5py', subset=males_keys(H5_dic))
fem = H5PY_RW('test_raw.h5py', subset=females_keys(H5_dic))
print 'Data with', len(H5_dic), 'male and female speakers'
print males.length(), 'elements'
print fem.length(), 'elements'
mixed_data = Mixer([males, fem],
chunk_size=chunk_size,
with_mask=False,
with_inputs=True,
shuffling=True)
####
#### PREVIOUS MODEL CONFIG
####
N = 256
max_pool = 256
batch_size = 16
learning_rate = 0.01
config_model = {}
config_model["type"] = "pretraining"
H5_dico = read_data_header()
males = H5PY_RW()
males.open_h5_dataset('test_raw.h5py', subset=males_keys(H5_dico))
males.set_chunk(5 * 4 * 512)
males.shuffle()
print 'Male voices loaded: ', males.length(), ' items'
fem = H5PY_RW()
fem.open_h5_dataset('test_raw.h5py', subset=females_keys(H5_dico))
fem.set_chunk(5 * 4 * 512)
fem.shuffle()
print 'Female voices loaded: ', fem.length(), ' items'
Mixer = Mixer([males, fem], with_mask=False, with_inputs=True)
adapt_model = Adapt.load('jolly-firefly-9628',
pretraining=False,
separator=DPCL)
# adapt_model.init()
print 'Model DAS created'
testVar = raw_input("Model loaded : Press Enter")
cost_valid_min = 1e10
Mixer.select_split(0)
learning_rate = 0.01
for i in range(config.max_iterations):
X_in, X_mix, Ind = Mixer.get_batch(1)