def text2symbols(text, dtype):
clean_char = custom_english_cleaners(text.rstrip()).rstrip()
if clean_char[-1] in ['.', ',']:
while clean_char[-1] in ['.', ',']:
clean_char = clean_char[:-1]
clean_char = clean_char + '.'
elif clean_char[-1] in ['!', '?']:
clean_char = clean_char
else:
clean_char = clean_char + '.'
if dtype == 'char':
return clean_char
clean_phone = []
for s in g2p(clean_char.lower()):
if (s in [',', '!', '.', '?', "'"]) and (clean_phone[-1] == ' '):
clean_phone.pop()
clean_phone.append(s)
elif '@' + s in symbol_to_id:
clean_phone.append('@' + s)
else:
clean_phone.append(s)
return clean_phone
def frontend(self, text):
"""Clean text and then convert to id sequence."""
text = custom_english_cleaners(text)
charseq = list(text)
idseq = []
for c in charseq:
if c.isspace():
idseq += [self.char_to_id["<space>"]]
elif c not in self.char_to_id.keys():
idseq += [self.char_to_id["<unk>"]]
else:
idseq += [self.char_to_id[c]]
idseq += [self.idim - 1] # <eos>
return torch.LongTensor(idseq).view(-1).to(self.device)
def main():
parser = argparse.ArgumentParser()
parser.add_argument("--lang_tag", type=str, default=None, nargs="?",
help="language tag (can be used for multi lingual case)")
parser.add_argument("--spk_tag", type=str,
help="speaker tag")
parser.add_argument("jsons", nargs="+", type=str,
help="*_mls.json filenames")
parser.add_argument("out", type=str,
help="output filename")
parser.add_argument("trans_type", type=str, default="phn",
choices=["char", "phn"],
help="Input transcription type")
args = parser.parse_args()
dirname = os.path.dirname(args.out)
if len(dirname) != 0 and not os.path.exists(dirname):
os.makedirs(dirname)
with codecs.open(args.out, "w", encoding="utf-8") as out:
for filename in sorted(args.jsons):
with codecs.open(filename, "r", encoding="utf-8") as f:
js = json.load(f)
for key in sorted(js.keys()):
uid = args.spk_tag + "_" + key.replace(".wav", "")
content = js[key]["clean"]
text = custom_english_cleaners(content.rstrip())
if args.trans_type == "phn":
clean_content = text.lower()
text = g2p(clean_content)
if args.lang_tag is None:
line = "%s %s \n" % (uid, text)
else:
line = "%s <%s> %s\n" % (uid, args.lang_tag, text)
out.write(line)
f_g2p("")
except ImportError:
raise ImportError("g2p_en is not installed. please run `. ./path.sh && pip install g2p_en`.")
except LookupError:
# NOTE: we need to download dict in initial running
nltk.download("punkt")
def g2p(text):
"""Convert grapheme to phoneme."""
tokens = filter(lambda s: s != " ", f_g2p(text))
return ' '.join(tokens)
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument('text', type=str, help='text to be cleaned')
parser.add_argument("trans_type", type=str, default="kana",
choices=["char", "phn"],
help="Input transcription type")
args = parser.parse_args()
with codecs.open(args.text, 'r', 'utf-8') as fid:
for line in fid.readlines():
id, _, content = line.split("|")
clean_content = custom_english_cleaners(content.rstrip())
if args.trans_type == "phn":
text = clean_content.lower()
clean_content = g2p(text)
print("%s %s" % (id, clean_content))
root_dir = '/hd0/dataset/VCTK/VCTK-Corpus/wav48'
data_dir = '/hd0/speech-aligner/preprocessed/VCTK20_engspks'
os.makedirs(data_dir, exist_ok=True)
os.makedirs(os.path.join(data_dir, 'char_seq'), exist_ok=True)
os.makedirs(os.path.join(data_dir, 'phone_seq'), exist_ok=True)
os.makedirs(os.path.join(data_dir, 'melspectrogram'), exist_ok=True)
g2p = G2p()
metadata = {}
with codecs.open(csv_file, 'r', 'utf-8') as fid:
for line in fid.readlines():
id, text, spk = line.split("|")
id = os.path.splitext(id)[0]
clean_char = custom_english_cleaners(text.rstrip())
clean_phone = []
for s in g2p(clean_char.lower()):
if '@' + s in symbol_to_id:
clean_phone.append('@' + s)
else:
clean_phone.append(s)
metadata[id] = {'char': clean_char, 'phone': clean_phone}
stft = TacotronSTFT(filter_length=1024,
hop_length=256,
win_length=1024,
n_mel_channels=80,
sampling_rate=16000,
mel_fmin=55.0,
choices=["char", "phn"],
help="Input transcription type")
parser.add_argument("--lowercase",
type=bool,
default=False,
help="Lower case the result or not")
args = parser.parse_args()
# clean every line in transcription file first
with codecs.open(args.transcription_path, 'r', 'utf-8') as fid:
for line in fid.read().splitlines():
segments = line.split(" ")
# clean contents
content = ' '.join(segments[:-1])
clean_content = custom_english_cleaners(content)
# get id by taking off the parentheses
id = segments[-1][1:-1]
if args.trans_type == "phn":
text = clean_content.lower()
clean_content = g2p(text)
if args.lowercase:
clean_content = clean_content.lower()
if args.lang_tag == "":
print("{} {}".format(id, clean_content))
else:
print("{} {}".format(