def __init__(self, path='datasets/Nottingham',
source='http://www-etud.iro.umontreal.ca/~boulanni/Nottingham.zip',
train_filter='.*train.*',
valid_filter='.*valid.*',
test_filter='.*test.*', ):
super(Nottingham, self).__init__(path=path, source=source,
train_filter=train_filter,
valid_filter=valid_filter,
test_filter=test_filter)
# grab the datasets from midireading the files
train_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(config.floatX)
for f in find_files(self.path, train_filter)
]
valid_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(config.floatX)
for f in find_files(self.path, valid_filter)
]
test_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(config.floatX)
for f in find_files(self.path, test_filter)
]
self.train_inputs = numpy.concatenate(train_datasets)
self.train_targets = None
self.valid_inputs = numpy.concatenate(valid_datasets)
self.valid_targets = None
self.test_inputs = numpy.concatenate(test_datasets)
self.test_targets = None
def __init__(self, path='datasets/MuseData',
source='http://www-etud.iro.umontreal.ca/~boulanni/MuseData.zip',
train_filter='.*train.*',
valid_filter='.*valid.*',
test_filter='.*test.*', ):
super(MuseData, self).__init__(path=path, source=source,
train_filter=train_filter,
valid_filter=valid_filter,
test_filter=test_filter)
# grab the datasets from midireading the files
train_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(theano.config.floatX)
for f in find_files(self.path, train_filter)
]
valid_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(theano.config.floatX)
for f in find_files(self.path, valid_filter)
]
test_datasets = [
midiread(f, r=(21, 109), dt=0.3).piano_roll.astype(theano.config.floatX)
for f in find_files(self.path, test_filter)
]
self.train_inputs = numpy.concatenate(train_datasets)
self.train_targets = None
self.valid_inputs = numpy.concatenate(valid_datasets)
self.valid_targets = None
self.test_inputs = numpy.concatenate(test_datasets)
self.test_targets = None
def __iter__(self):
for fname in files.find_files(self.path, self.filter):
if self.preprocess is not None and callable(self.preprocess):
fname = self.preprocess(fname)
fnames = raise_to_list(fname)
for name in fnames:
yield name
def __iter__(self):
for fname in find_files(self.path, self.filter):
if self.preprocess is not None and callable(self.preprocess):
fname = self.preprocess(fname)
fnames = raise_to_list(fname)
for name in fnames:
yield name
def __iter__(self):
for fname in files.find_files(self.path, self.filter):
try:
with Image.open(fname) as im:
data = numpy.array(im)
if self.preprocess is not None and callable(self.preprocess):
data = self.preprocess(data)
data = raise_to_list(data)
for d in data:
yield d
except Exception as err:
log.exception(err.__str__())
def __iter__(self):
idx = 0
for fname in files.find_files(self.path, self.filter):
with open(fname, 'r') as f:
for line in f:
if self.preprocess is not None:
line = self.preprocess(line)
line = raise_to_list(line)
for token in line:
if idx >= self.n_future:
yield token
else:
idx += 1
def __iter__(self):
idx = 0
for fname in files.find_files(self.path, self.filter):
with open(fname, 'r') as f:
for line in f:
if self.preprocess is not None:
line = self.preprocess(line)
line = raise_to_list(line)
for token in line:
if idx >= self.n_future:
yield token
else:
idx += 1
def __iter__(self):
for fname in find_files(self.path, self.filter):
try:
with Image.open(fname) as im:
data = numpy.array(im)
if self.preprocess is not None and callable(
self.preprocess):
data = self.preprocess(data)
data = raise_to_list(data)
for d in data:
yield d
except Exception as err:
_log.exception(err.__str__())
def __iter__(self):
idx = 0
for fname in files.find_files(self.path, self.filter):
try:
with open(fname, 'r') as f:
for line in f:
if self.preprocess is not None and callable(self.preprocess):
line = self.preprocess(line)
line = raise_to_list(line)
for token in line:
if idx >= self.n_future:
yield token
else:
idx += 1
except Exception as err:
log.exception(err.__str__())
def __iter__(self):
idx = 0
for fname in find_files(self.path, self.filter):
try:
with open(fname, 'r') as f:
for line in f:
if self.preprocess is not None and callable(
self.preprocess):
line = self.preprocess(line)
line = raise_to_list(line)
for token in line:
if idx >= self.n_future:
yield token
else:
idx += 1
except Exception as err:
_log.exception(err.__str__())
def __init__(
self,
path=DEFAULT_TEDLIUM_DATASET_PATH,
window_duration = 0.01,
skip_count = 1,
max_speeches = None,
):
"""Initialize the Dataset with a given storage for TEDLIUM
path -- target path for the TED LIUM data storage
window_duration -- duration of the audio window in seconds
skip_count -- step size across the segments in the repo
used to do a very small subset of the dataset
when doing testing iterations. This allows you
to test an "epoch" across a small subset of the
40GB data-file
"""
self.window_size = 2**int(math.ceil(math.log(int(window_duration * 16000),2)))
source_filename = path + '.tar.gz'
if not os.path.exists(path):
if os.path.exists(source_filename):
# Note: this could, in theory overwrite anything on disk, as the Python
# tarfile module doesn't prevent writing outside the root directory
# (according to its docs).
file_ops.untar(source_filename, destination_dir=os.path.dirname(path))
if not os.path.exists(path):
raise RuntimeError(
"You need to download the TEDLIUM corpus (v2) from %(url)s and save it to %(path)s"%{
'url': LIUM_BASE + TEDLIUM_DOWNLOAD_URL,
'path': source_filename,
}
)
path = os.path.realpath(path)
log.info("Searching for speeches")
self.train_speeches = [
tedlium.Speech( sph, window_size=self.window_size )
for sph in file_ops.find_files(
path, '.*[/]train[/]sph[/].*[.]sph',
)
]
if max_speeches:
self.train_speeches = self.train_speeches[:max_speeches]
self.test_speeches = [
tedlium.Speech( sph, window_size=self.window_size )
for sph in file_ops.find_files(
path, '.*[/]test[/]sph[/].*[.]sph',
)
]
if max_speeches:
self.test_speeches = self.test_speeches[:max_speeches]
self.valid_speeches = [
tedlium.Speech( sph, window_size=self.window_size )
for sph in file_ops.find_files(
path, '.*[/]dev[/]sph[/].*[.]sph',
)
]
if max_speeches:
self.valid_speeches = self.valid_speeches[:max_speeches]
log.info(
"Creating speech segments (utterance records using 1/%s of the utterances)",
skip_count,
)
train_inputs,train_targets = inputs_and_targets( self.train_speeches )
valid_inputs,valid_targets = inputs_and_targets( self.valid_speeches )
test_inputs,test_targets = inputs_and_targets( self.test_speeches )
log.info("Initializing the OpenDeep dataset")
super(TEDLIUMDataset,self).__init__(
train_inputs=train_inputs,train_targets=train_targets,
valid_inputs=valid_inputs,valid_targets=valid_targets,
test_inputs=test_inputs,test_targets=test_targets,
)
