def get_trans_array(ori_phone_file, tgt_phone_file):
#transform the orders of phone to kaldi format
import utils
ori = utils.read_phone_txt(ori_phone_file)
ori.append('spn')
tgt = utils.read_phone_txt(tgt_phone_file, 0)
new_tgt = []
for x in tgt:
for s in ['<', '#']:
if x.startswith(s):
break
else:
new_tgt.append(x)
trans = []
for ix, x in enumerate(new_tgt):
for iy, y in enumerate(ori):
if x == y:
trans.append(iy)
return np.array(trans)
import utils
import sys
import argparse
parser = argparse.ArgumentParser()
parser.add_argument("phone_table_path", type=str)
parser.add_argument("pdf_table_path", type=str)
parser.add_argument("--typ",
type=str,
default='mono',
choices=['mono', 'biphone'])
args = parser.parse_args()
phones = utils.read_phone_txt(args.phone_table_path, 0)
valid_phone = []
for x in phones:
for s in ['<', '#']:
if x.startswith(s):
break
else:
valid_phone.append(x)
if args.typ == 'mono':
utils.write_phone_file(valid_phone, args.pdf_table_path, True)
else:
L = []
for i in range(len(valid_phone)):
for j in range(len(valid_phone) + 1):
if j == 0:
L.append('start' + '_' + valid_phone[i])
else:
import sys
import utils as utl
phone_map_txt = sys.argv[1]
phone_list_txt = sys.argv[2]
phone_list = utl.read_phone_txt(phone_map_txt)
phone_list.append('spn') # for oov word, which is requried for kaldi
utl.write_phone_file(phone_list, phone_list_txt)
import utils
import sys
phone_table_path = sys.argv[1]
pdf_table_path = sys.argv[2]
phones = utils.read_phone_txt(phone_table_path, 0)
L = []
for x in phones:
for s in ['<', '#']:
if x.startswith(s):
break
else:
L.append(x)
utils.write_phone_file(L, pdf_table_path, True)
sys.path.append('scripts/')
import utils
parser = argparse.ArgumentParser()
parser.add_argument("phone_list_txt", type=str)
parser.add_argument(
"--self_loop_prob",
type=float,
default=0.5,
help=
"Probabilty of staying in the same state. (1-self_loop_prob) is the probability of transition to other state"
)
args = parser.parse_args()
phone_list = utils.read_phone_txt(args.phone_list_txt, 0)
same_phone = []
trans_phone = []
for idx, phone in enumerate(phone_list):
if phone in ['sil', 'spn']:
same_phone.append(idx)
if '_' in phone:
x, y = phone.split('_')
if x == y:
same_phone.append(idx)
else:
trans_phone.append(idx)
print("<Topology>")
print("<TopologyEntry>")
import sys
sys.path.append('scripts/')
import utils
import math
#48 phones
monophones_txt = sys.argv[1]
all_phones_txt = sys.argv[2]
self_loop_prob = sys.argv[3]
trans_prob = -math.log(1 - float(self_loop_prob))
phones = utils.read_phone_txt(monophones_txt)
cd_phones = utils.read_phone_txt(all_phones_txt, 0)
print("0 1 0 0")
for idx, phone in enumerate(cd_phones):
if '_' in phone:
x, y = phone.split('_')
iy = phones.index(y)
if x == y:
print("1 {} {} {}".format(iy + 2, idx, idx))
else:
print("1 {} {} {} {:.5f}".format(iy + 2, idx, idx, trans_prob))
for idx, phone in enumerate(cd_phones):
if '_' in phone:
x, y = phone.split('_')
iy = phones.index(y)
ix = phones.index(x)
if x == y:
print("{} {} {} {}".format(ix + 2, iy + 2, idx, idx))
else:
import sys, os
sys.path.append('scripts/')
import utils
monophone_list = sys.argv[1] # not include spn
dict_dir = sys.argv[2]
phone_list = utils.read_phone_txt(monophone_list)
with open(os.path.join(dict_dir, 'extra_questions.txt'), 'w') as f:
f.write('')
with open(os.path.join(dict_dir, 'optional_silence.txt'), 'w') as f:
f.write('sil' + '\n')
with open(os.path.join(dict_dir, 'silence_phones.txt'), 'w') as f:
f.write('sil' + '\n')
f.write('spn' + '\n')
with open(os.path.join(dict_dir, 'nonsilence_phones.txt'), 'w') as f:
for phone1 in phone_list:
for phone2 in phone_list:
f.write(phone1 + '_' + phone2 + '\n')
with open(os.path.join(dict_dir, 'lexicon.txt'), 'w') as f:
for phone1 in phone_list:
for phone2 in phone_list:
f.write("{} {} {}\n".format(phone2, phone1 + '_' + phone2,
phone2 + '_' + phone2))
f.write("{} {}\n".format(phone1, phone1 + '_' + phone1))
f.write("<UNK> spn\n")