Spaces:
Running
Running
Yurii Paniv
commited on
Commit
•
30f2cde
1
Parent(s):
305bf1a
Improve English handling
Browse files- tests/test_formatter.py +1 -1
- ukrainian_tts/formatter.py +5 -2
tests/test_formatter.py
CHANGED
@@ -3,7 +3,7 @@ from ukrainian_tts.formatter import preprocess_text
|
|
3 |
|
4 |
def test_formatter():
|
5 |
examples = [
|
6 |
-
("Quality of life update", "
|
7 |
("Він украв 20000000 $", "він украв двадцять мільйонів долар"),
|
8 |
(
|
9 |
"111 000 000 000 доларів державного боргу.",
|
|
|
3 |
|
4 |
def test_formatter():
|
5 |
examples = [
|
6 |
+
("Quality of life update", "кваліті оф ліфе юпдате"),
|
7 |
("Він украв 20000000 $", "він украв двадцять мільйонів долар"),
|
8 |
(
|
9 |
"111 000 000 000 доларів державного боргу.",
|
ukrainian_tts/formatter.py
CHANGED
@@ -2,7 +2,8 @@ import num2words
|
|
2 |
import re
|
3 |
|
4 |
|
5 |
-
def preprocess_text(text
|
|
|
6 |
# currencies
|
7 |
text = text.replace("$", "долар")
|
8 |
text = text.replace("₴", "гривня")
|
@@ -59,6 +60,9 @@ def preprocess_text(text, use_autostress_model=False):
|
|
59 |
text = text.replace("0", "нуль ")
|
60 |
# speak english alphabet using brute force transliteration
|
61 |
english = {
|
|
|
|
|
|
|
62 |
"a": "а",
|
63 |
"b": "б",
|
64 |
"c": "ц",
|
@@ -91,5 +95,4 @@ def preprocess_text(text, use_autostress_model=False):
|
|
91 |
text = text.replace(english_char.upper(), english[english_char].upper())
|
92 |
text = text.replace(english_char, english[english_char])
|
93 |
|
94 |
-
text = text.lower()
|
95 |
return text
|
|
|
2 |
import re
|
3 |
|
4 |
|
5 |
+
def preprocess_text(text):
|
6 |
+
text = text.lower()
|
7 |
# currencies
|
8 |
text = text.replace("$", "долар")
|
9 |
text = text.replace("₴", "гривня")
|
|
|
60 |
text = text.replace("0", "нуль ")
|
61 |
# speak english alphabet using brute force transliteration
|
62 |
english = {
|
63 |
+
"qu": "кв",
|
64 |
+
"ch": "ч",
|
65 |
+
"ph": "ф",
|
66 |
"a": "а",
|
67 |
"b": "б",
|
68 |
"c": "ц",
|
|
|
95 |
text = text.replace(english_char.upper(), english[english_char].upper())
|
96 |
text = text.replace(english_char, english[english_char])
|
97 |
|
|
|
98 |
return text
|