mrfakename
commited on
Commit
β’
03572f6
1
Parent(s):
b8401e1
Enhancements
Browse files- melo/split_utils.py +1 -1
melo/split_utils.py
CHANGED
@@ -18,7 +18,7 @@ def split_sentences_latin(text, min_len=10):
|
|
18 |
text = re.sub('[ββ]', '"', text)
|
19 |
text = re.sub('[ββ]', "'", text)
|
20 |
text = re.sub(r"[\<\>\(\)\[\]\"\Β«\Β»]+", "", text)
|
21 |
-
return txtsplit(text, 512, 512)
|
22 |
# ε°ζζ¬δΈηζ’θ‘符γη©Ίζ ΌεεΆθ‘¨η¬¦ζΏζ’δΈΊη©Ίζ Ό
|
23 |
# text = re.sub('[\n\t ]+', ' ', text)
|
24 |
# # ε¨ζ ηΉη¬¦ε·εζ·»ε δΈδΈͺη©Ίζ Ό
|
|
|
18 |
text = re.sub('[ββ]', '"', text)
|
19 |
text = re.sub('[ββ]', "'", text)
|
20 |
text = re.sub(r"[\<\>\(\)\[\]\"\Β«\Β»]+", "", text)
|
21 |
+
return [item.strip() for item in txtsplit(text, 512, 512) if item.strip()]
|
22 |
# ε°ζζ¬δΈηζ’θ‘符γη©Ίζ ΌεεΆθ‘¨η¬¦ζΏζ’δΈΊη©Ίζ Ό
|
23 |
# text = re.sub('[\n\t ]+', ' ', text)
|
24 |
# # ε¨ζ ηΉη¬¦ε·εζ·»ε δΈδΈͺη©Ίζ Ό
|