def test_1d(self):
bn = False
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
params['net'] = dict()
params['net']['shape'] = [16, 1] # Shape of the image
params['net']['generator'] = dict()
params['net']['generator']['latent_dim'] = 100
params['net']['generator']['full'] = [2 * 8]
params['net']['generator']['nfilter'] = [2, 32, 32, 1]
params['net']['generator']['batch_norm'] = [bn, bn, bn]
params['net']['generator']['shape'] = [[5], [5], [5], [5]]
params['net']['generator']['stride'] = [1, 2, 1, 1]
params['net']['generator']['data_size'] = 1
params['net']['discriminator'] = dict()
params['net']['discriminator']['full'] = [32]
params['net']['discriminator']['nfilter'] = [16, 32, 32, 32]
params['net']['discriminator']['batch_norm'] = [bn, bn, bn, bn]
params['net']['discriminator']['shape'] = [[5], [5], [5], [3]]
params['net']['discriminator']['stride'] = [2, 2, 2, 1]
params['net']['discriminator']['data_size'] = 1
X = np.random.rand(101, 16)
dataset = Dataset(X)
wgan = GANsystem(WGAN, params)
wgan.train(dataset)
img = wgan.generate(2)
assert (len(img) == 2)
assert (img.shape[1:] == (16, 1))
img = wgan.generate(500)
assert (len(img) == 500)
def test_generate(self):
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
params['net'] = dict()
params['net']['generator'] = dict()
params['net']['generator']['latent_dim'] = 5
X = np.random.rand(101, 16, 16)
dataset = Dataset(X)
wgan = GANsystem(WGAN, params)
wgan.train(dataset)
img = wgan.generate(2)
img = wgan.generate(500)
assert (len(img) == 500)
z = np.random.randn(133, 5)
img1 = wgan.generate(z=z)
img2 = np.zeros(shape=img1.shape)
for i in range(133):
img2[i] = wgan.generate(z=np.reshape(z[i], [1, 5]))
np.testing.assert_almost_equal(img1, img2, decimal=6)
def test_paulinasystem(self):
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
X = np.random.rand(101, 16, 16)
dataset = Dataset(X)
wgan = PaulinaGANsystem(WGAN, params)
wgan.train(dataset)
img = wgan.generate(2)
img = wgan.generate(500)
assert (len(img) == 500)
def test_patchupscalegan1dborder(self):
bn = False
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
params['net'] = dict()
params['net']['shape'] = [8, 2]
params['net']['generator'] = dict()
params['net']['generator']['latent_dim'] = 8
params['net']['generator']['full'] = [16]
params['net']['generator']['nfilter'] = [8, 32, 1]
params['net']['generator']['batch_norm'] = [bn, bn]
params['net']['generator']['shape'] = [[3], [3], [3]]
params['net']['generator']['stride'] = [1, 1, 1]
params['net']['generator']['data_size'] = 1
params['net']['generator']['borders'] = dict()
params['net']['generator']['borders']['width_full'] = None
params['net']['generator']['borders']['nfilter'] = [4, 1]
params['net']['generator']['borders']['batch_norm'] = [bn, bn]
params['net']['generator']['borders']['shape'] = [[5], [3]]
params['net']['generator']['borders']['stride'] = [2, 2]
params['net']['generator']['borders']['data_size'] = 1
params['net']['generator']['borders']['width_full'] = 2
params['net']['discriminator'] = dict()
params['net']['discriminator']['full'] = [32]
params['net']['discriminator']['nfilter'] = [16, 32]
params['net']['discriminator']['batch_norm'] = [bn, bn]
params['net']['discriminator']['shape'] = [[5], [3]]
params['net']['discriminator']['stride'] = [2, 2]
params['net']['discriminator']['data_size'] = 1
params['net']['upsampling'] = 2
X = np.random.rand(101, 8, 2)
dataset = Dataset(X)
wgan = GANsystem(UpscalePatchWGANBorders, params)
wgan.train(dataset)
borders = np.random.rand(500, 4, 1)
X_smooth = np.random.rand(500, 4, 1)
img = wgan.generate(N=2, X_smooth=X_smooth[:2], borders=borders[:2])
assert (len(img) == 2)
assert (img.shape[1:] == (4, 1))
img = wgan.generate(N=500,
X_smooth=X_smooth[:500],
borders=borders[:500])
assert (len(img) == 500)
def test_patchupscalegan3d(self):
bn = False
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
params['net'] = dict()
params['net']['shape'] = [8, 8, 8, 8]
params['net']['generator'] = dict()
params['net']['generator']['latent_dim'] = 8 * 8 * 8
params['net']['generator']['full'] = []
params['net']['generator']['nfilter'] = [8, 32, 1]
params['net']['generator']['batch_norm'] = [bn, bn]
params['net']['generator']['shape'] = [[3, 3, 3], [3, 3, 3], [3, 3, 3]]
params['net']['generator']['stride'] = [1, 1, 1]
params['net']['generator']['data_size'] = 3
params['net']['discriminator'] = dict()
params['net']['discriminator']['full'] = [32]
params['net']['discriminator']['nfilter'] = [16, 32]
params['net']['discriminator']['batch_norm'] = [bn, bn]
params['net']['discriminator']['shape'] = [[3, 3, 3], [3, 3, 3]]
params['net']['discriminator']['stride'] = [2, 2]
params['net']['discriminator']['data_size'] = 3
params['net']['upsampling'] = 2
X = np.random.rand(101, 8, 8, 8, 8)
dataset = Dataset(X)
class UgradedGAN(UpscalePatchWGAN, CosmoWGAN):
pass
wgan = GANsystem(UgradedGAN, params)
wgan.train(dataset)
borders = np.random.rand(500, 8, 8, 8, 7)
X_down = np.random.rand(500, 4, 4, 4, 1)
img = wgan.generate(N=2, X_down=X_down[:2], borders=borders[:2])
assert (len(img) == 2)
assert (img.shape[1:] == (8, 8, 8, 1))
img = wgan.generate(N=500, X_down=X_down[:500], borders=borders[:500])
assert (len(img) == 500)
def test_lapgan1d(self):
bn = False
params = dict()
params['optimization'] = dict()
params['optimization']['epoch'] = 1
params['summary_every'] = 4
params['save_every'] = 5
params['print_every'] = 3
params['net'] = dict()
params['net']['shape'] = [16, 1]
params['net']['generator'] = dict()
params['net']['generator']['latent_dim'] = 16 * 16
params['net']['generator']['full'] = []
params['net']['generator']['nfilter'] = [8, 32, 1]
params['net']['generator']['batch_norm'] = [bn, bn]
params['net']['generator']['shape'] = [[5], [5], [5]]
params['net']['generator']['stride'] = [1, 1, 1]
params['net']['generator']['data_size'] = 1
params['net']['discriminator'] = dict()
params['net']['discriminator']['full'] = [32]
params['net']['discriminator']['nfilter'] = [16, 32]
params['net']['discriminator']['batch_norm'] = [bn, bn]
params['net']['discriminator']['shape'] = [[5], [3]]
params['net']['discriminator']['stride'] = [2, 2]
params['net']['discriminator']['data_size'] = 1
params['net']['upsampling'] = 2
X = np.random.rand(101, 16)
dataset = Dataset(X)
wgan = GANsystem(LapWGAN, params)
wgan.train(dataset)
X_down = np.random.rand(500, 8, 1)
img = wgan.generate(N=2, X_down=X_down[:2])
assert (len(img) == 2)
assert (img.shape[1:] == (16, 1))
img = wgan.generate(N=500, X_down=X_down[:500])
assert (len(img) == 500)
tgrads[np.isnan(tgrads)] = 0 tgrads = np.clip(tgrads, -1, 1) fgrads[np.isnan(fgrads)] = 0 fgrads = np.clip(fgrads, -1, 1) print(np.max(tgrads[:, :256, :])) print(np.min(tgrads[:, :256, :])) print(np.mean(tgrads[:, :256, :])) print(np.max(fgrads[:, :256, :])) print(np.min(fgrads[:, :256, :])) print(np.mean(fgrads[:, :256, :])) dataset = Dataset(np.stack([preprocessed_images[:, :256], tgrads[:, :256], fgrads[:, :256]], axis=-1)) time_str = 'commands_md64_tgrads_fgrads_squared_8k' global_path = '../../saved_results' name = time_str from gantools import blocks bn = False md = 64 params_discriminator = dict() params_discriminator['stride'] = [2,2,2,2,2] params_discriminator['nfilter'] = [md, 2*md, 4*md, 8*md, 16*md] params_discriminator['shape'] = [[5, 5], [5, 5], [5, 5], [5, 5], [5, 5]]
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = raw_data['logspecs']
for i in range(2, 7):
mat_path = "../../data/test_spectrograms_and_derivs_" + str(i) + ".mat"
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = np.concatenate(
[preprocessed_images, raw_data['logspecs']])
del raw_data
print(preprocessed_images.shape)
print(np.max(preprocessed_images[:, :128, :]))
print(np.min(preprocessed_images[:, :128, :]))
print(np.mean(preprocessed_images[:, :128, :]))
dataset = Dataset(preprocessed_images[:, :128])
time_str = 'commands_md64_4k_squared'
global_path = '../../saved_results'
name = time_str
from gantools import blocks
bn = False
md = 64
params_discriminator = dict()
params_discriminator['stride'] = [2, 2, 2, 2, 2]
params_discriminator['nfilter'] = [md, 2 * md, 4 * md, 8 * md, 16 * md]
params_discriminator['shape'] = [[6, 6], [6, 6], [6, 6], [6, 6], [6, 6]]
tgrads[np.isnan(tgrads)] = 0
tgrads = np.clip(tgrads, -1, 1)
fgrads[np.isnan(fgrads)] = 0
fgrads = np.clip(fgrads, -1, 1)
print(np.max(tgrads[:, :256, :]))
print(np.min(tgrads[:, :256, :]))
print(np.mean(tgrads[:, :256, :]))
print(np.max(fgrads[:, :256, :]))
print(np.min(fgrads[:, :256, :]))
print(np.mean(fgrads[:, :256, :]))
dataset = Dataset(
np.stack([preprocessed_images[:, :256], tgrads[:, :256], fgrads[:, :256]],
axis=-1))
time_str = 'commands_md64_tgrads_fgrads_8k'
global_path = '../../saved_results'
name = time_str
from gantools import blocks
bn = False
md = 64
params_discriminator = dict()
params_discriminator['stride'] = [2, 2, 2, 2, 2]
params_discriminator['nfilter'] = [md, 2 * md, 4 * md, 8 * md, 16 * md]
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = raw_data['logspecs']
for i in range(2, 7):
mat_path = "../../data/test_spectrograms_and_derivs_" + str(i) + ".mat"
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = np.concatenate(
[preprocessed_images, raw_data['logspecs']])
del raw_data
print(preprocessed_images.shape)
print(np.max(preprocessed_images[:, :128, :]))
print(np.min(preprocessed_images[:, :128, :]))
print(np.mean(preprocessed_images[:, :128, :]))
dataset = Dataset(preprocessed_images[:, :128])
time_str = 'commands_md32_4k'
global_path = '../../saved_results'
name = time_str
from gantools import blocks
bn = False
md = 32
params_discriminator = dict()
params_discriminator['stride'] = [2, 2, 2, 2, 2]
params_discriminator['nfilter'] = [md, 2 * md, 4 * md, 8 * md, 16 * md]
params_discriminator['shape'] = [[12, 3], [12, 3], [12, 3], [12, 3], [12, 3]]
def load_audio_dataset(shuffle=True,
scaling=1,
patch=False,
augmentation=False,
spix=None,
smooth=None,
type='nsynth'):
''' Load a Nsynth dataset object.
Arguments
---------
* shuffle: shuffle the data (default True)
* scaling : downscale the image by a factor (default 1)
* path : downscale the image by a factor (default 1)
* scaling : downscale the image by a factor (default 1)
'''
if type == 'nsynth':
sig = load_nysnth_rawdata()
sig = sig[:, :2**15]
elif type == 'piano':
sig = load_piano_rawdata()
else:
raise ValueError('Incorrect value for type')
if len(sig.shape) == 1:
sig = np.reshape(sig, [1, len(sig)])
# if augmentation and (not patch):
# raise ValueError('Augementation works only with patches.')
# 1) Transform the data
def transform(x):
x = x / (2**15)
x = (0.99 * x.T / np.max(np.abs(x), axis=1)).T
return x
sig = transform(sig)
# 2) Downsample
Nwin = 32
if scaling > 1:
# sig = blocks.downsample(sig, scaling)
sig = transformation.downsample_1d(sig, scaling, Nwin=Nwin)
if smooth is not None:
sig = sig[:, :(sig.shape[1] // smooth) * smooth]
sig_down = transformation.downsample_1d(sig, smooth, Nwin=Nwin)
sig_smooth = transformation.upsamler_1d(sig_down, smooth, Nwin=Nwin)
sig = np.concatenate(
(np.expand_dims(sig, axis=2), np.expand_dims(sig_smooth, axis=2)),
axis=2)
if patch:
slice_fn = partial(transformation.slice_1d_patch, spix=spix)
else:
if spix is not None:
slice_fn = partial(transformation.slice_1d, spix=spix)
else:
slice_fn = do_nothing
if augmentation:
transform = partial(transformation.random_shift_1d,
roll=False,
spix=spix)
else:
transform = do_nothing
# 3) Make a dataset
dataset = Dataset(sig,
shuffle=shuffle,
transform=transform,
slice_fn=slice_fn)
return dataset
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = raw_data['logspecs']
for i in range(2, 7):
mat_path = "../../data/test_spectrograms_and_derivs_" + str(i) + ".mat"
raw_data = scipy.io.loadmat(mat_path)
preprocessed_images = np.concatenate(
[preprocessed_images, raw_data['logspecs']])
del raw_data
print(preprocessed_images.shape)
print(np.max(preprocessed_images[:, :256, :]))
print(np.min(preprocessed_images[:, :256, :]))
print(np.mean(preprocessed_images[:, :256, :]))
dataset = Dataset(preprocessed_images[:, :256])
time_str = 'commands_md64_8k'
global_path = '../../saved_results'
name = time_str
from gantools import blocks
bn = False
md = 64
params_discriminator = dict()
params_discriminator['stride'] = [2, 2, 2, 2, 2]
params_discriminator['nfilter'] = [md, 2 * md, 4 * md, 8 * md, 16 * md]