From 1295c6a9d30b1f94185156ec8d9ca7c893c4f6c6 Mon Sep 17 00:00:00 2001
From: Yanli <mail@yanli.one>
Date: Thu, 13 Jun 2024 16:02:50 +0800
Subject: [PATCH] update clean text

---
 src/gpt_sovits/infer/text_utils.py | 21 +++++++++++++--------
 1 file changed, 13 insertions(+), 8 deletions(-)

diff --git a/src/gpt_sovits/infer/text_utils.py b/src/gpt_sovits/infer/text_utils.py
index 8edd312d..9a85da3a 100644
--- a/src/gpt_sovits/infer/text_utils.py
+++ b/src/gpt_sovits/infer/text_utils.py
@@ -16,7 +16,7 @@ def cut5(inp: str, splits: Sequence[str]):
     return mergeitems
 
 
-def merge_short_texts(texts: List[str], threshold: int = 32):
+def merge_short_texts(texts: List[str], threshold: int = 36):
     """Merge short texts to longer ones. Texts are generated by cut5."""
     result: List[str] = []
     text = ""
@@ -42,10 +42,15 @@ def clean_and_cut_text(text: str) -> List[str]:
     text = new_text
 
     lines = [line.strip() for line in text.split("\n") if line.strip()]
-    sents = [sent for line in lines for sent in cut5(line, tier1_punc)]
-    texts = [
-        "," + merged.strip()
-        for sent in sents
-        for merged in merge_short_texts(cut5(sent, ","))
-    ]
-    return texts
+    res: List[str] = []
+    for line in lines:
+        sents = [sent for sent in cut5(line, tier1_punc)]
+        texts = [
+            merged.strip()
+            for sent in sents
+            for merged in merge_short_texts(cut5(sent, ","))
+        ]
+        texts = merge_short_texts(texts)
+        res.extend(texts)
+    res = [("。" + x) if x[0] != "。" else x for x in res]
+    return res