def _convert_transcription(self, in_file, out_file):
lines = open(in_file).read().splitlines()
output = []
for line in lines:
fileid, word = line.split("|")
phone = text.word2phone(word)
content = "<s> {} </s> ({})".format(phone, fileid)
output.append(content)
content = "\n".join(output)
open(out_file, "w").write(content)
def _make_cleaned_text(self):
in_file = "{}/train/text".format(self.corpus_folder)
out_file = "{}/etc/text".format(self.tmp_folder)
lines = open(in_file).read().splitlines()[:N]
output = []
for line in lines:
fileid, word = line.split("|")
phone = text.word2phone(word)
content = "<s> {} </s>".format(phone, fileid)
output.append(content)
content = "\n".join(output)
open(out_file, "w").write(content)
def _make_dictionary(self):
lines = open("{}/train/text".format(
self.corpus_folder)).read().splitlines()[:N]
phones = []
for line in lines:
fileid, word = line.split("|")
p = text.word2phone(word).split()
phones += p
phones = sorted(set(phones))
# create .dic files
lines = []
phone_units = []
for p in phones:
units = list(p)
phone_units += units
units = " ".join(units)
line = "{:20s}{}".format(p, units)
lines.append(line)
open("{}/etc/tmp.dic".format(self.tmp_folder),
"w").write("\n".join(lines))
phone_units = sorted(set(phone_units))
phone_units.append("SIL")
open("{}/etc/tmp.phone".format(self.tmp_folder),
"w").write("\n".join(phone_units))