fix: 修复x一x类型分词合并时可能出现的吞字问题; (#1362)

GPT_SoVITS/text/tone_sandhi.py

@@ -681,6 +681,7 @@ class ToneSandhi:
                     and seg[i - 1][0] == "一"
                     and seg[i - 2][0] == word
                     and pos == "v"
+                    and seg[i - 2][1] == "v"
                 ):
                     continue
                 else: