def _loadpath_word(self, path):
""" Load from multiple files...
"""
log.debug(
unicode(self) + " loading transcriptions from multiple files.")
wordlevel = {}
for filename in type_files(os.listdir(path), LAB_EXT):
with codecs.open(os.path.join(path, filename),
encoding="utf-8") as infh:
text = infh.read()
#parsing by assuming words are whitespace delimited:
wordlist = text.split()
if len(wordlist) == 0:
raise Exception("File '%s' is empty..." %
(os.path.join(path, filename)))
#assuming unique basenames... (a reasonable assumption)
key = parse_path(filename)[2]
if key in wordlevel:
raise Exception("basename '%s' is not unique..." % (key))
wordlevel[key] = " ".join(wordlist)
if len(wordlevel) == 0:
raise Exception("No transcriptions found in '%s'..." % (path))
return wordlevel
def _loadpath_phone(self, path):
""" Load from multiple files...
"""
log.debug(
unicode(self) + " loading transcriptions from multiple files.")
phonelevel = {}
boundaries = {}
filenames = []
try:
for ext in Utterance.SUPPORTED_EXTS:
filenames.extend(type_files(os.listdir(path), ext))
except OSError:
raise
if len(set(filenames)) != len(filenames):
raise Exception("Non unique basenames exist....")
filenames.sort()
for filename in filenames:
key = parse_path(filename)[2]
utt = Utterance(os.path.join(path, filename))
phonelevel[key] = " ".join([entry[1] for entry in utt.entries])
b = [float_to_htk_int(entry[0]) for entry in utt.entries]
if all(b) == False:
boundaries[key] = None
else:
boundaries[key] = b
return phonelevel, boundaries
def _loadpath_word(self, path):
""" Load from multiple files...
"""
log.debug(unicode(self) + " loading transcriptions from multiple files.")
wordlevel = {}
for filename in type_files(os.listdir(path), LAB_EXT):
with codecs.open(os.path.join(path, filename), encoding="utf-8") as infh:
text = infh.read()
#parsing by assuming words are whitespace delimited:
wordlist = text.split()
if len(wordlist) == 0:
raise Exception("File '%s' is empty..."
% (os.path.join(path, filename)))
#assuming unique basenames... (a reasonable assumption)
key = parse_path(filename)[2]
if key in wordlevel:
raise Exception("basename '%s' is not unique..." % (key))
wordlevel[key] = " ".join(wordlist)
if len(wordlevel) == 0:
raise Exception("No transcriptions found in '%s'..."
% (path))
return wordlevel
def _loadpath_phone(self, path):
""" Load from multiple files...
"""
log.debug(unicode(self) + " loading transcriptions from multiple files.")
phonelevel = {}
boundaries = {}
filenames = []
try:
for ext in Utterance.SUPPORTED_EXTS:
filenames.extend(type_files(os.listdir(path), ext))
except OSError:
raise
if len(set(filenames)) != len(filenames):
raise Exception("Non unique basenames exist....")
filenames.sort()
for filename in filenames:
key = parse_path(filename)[2]
utt = Utterance(os.path.join(path, filename))
phonelevel[key] = " ".join([entry[1] for entry in utt.entries])
b = [float_to_htk_int(entry[0]) for entry in utt.entries]
if all(b) == False:
boundaries[key] = None
else:
boundaries[key] = b
return phonelevel, boundaries
def __init__(self, wavlocation, featsconflocation):
""" Initialise...
"""
if not os.path.isdir(wavlocation):
raise Exception("'%s' is not an existing directory..." % wavlocation)
log.debug(unicode(self) + " loading audio files at '%s'." % (wavlocation))
self.wavlocation = wavlocation
self.wavfilelist = type_files(os.listdir(self.wavlocation), WAVE_EXT)
self.wavfilelist.sort()
self.hcopy_parms = self._loadFeatConf(featsconflocation)
def __init__(self, wavlocation, featsconflocation):
""" Initialise...
"""
if not os.path.isdir(wavlocation):
raise Exception("'%s' is not an existing directory..." %
wavlocation)
log.debug(
unicode(self) + " loading audio files at '%s'." % (wavlocation))
self.wavlocation = wavlocation
self.wavfilelist = type_files(os.listdir(self.wavlocation), WAVE_EXT)
self.wavfilelist.sort()
self.hcopy_parms = self._loadFeatConf(featsconflocation)
