From 9671d17d3162ed61e2d4904c81e3cfde8226eb23 Mon Sep 17 00:00:00 2001
From: litagin02 <litagin02@gmail.com>
Date: Sat, 6 Jan 2024 08:32:08 +0900
Subject: [PATCH 1/2] =?UTF-8?q?Fix=20=E3=82=83=20=E3=82=85=20=E3=82=87=20b?=
 =?UTF-8?q?ug?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

---
 text/japanese.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/text/japanese.py b/text/japanese.py
index af49eac..69301da 100644
--- a/text/japanese.py
+++ b/text/japanese.py
@@ -20,6 +20,7 @@ def hiragana2p(text: str) -> str:
     - avoid converting `o u` to `o o` (because the input is already actual `yomi`).
     - avoid using `N` for `ん` (for compatibility)
     - use `v` for `ゔ` related text.
+    - add bare `ゃ` `ゅ` `ょ` to `y a` `y u` `y o` (for compatibility).
     """
     # 3文字以上からなる変換規則
     text = text.replace("う゛ぁ", " v a")
@@ -321,6 +322,11 @@ def hiragana2p(text: str) -> str:
     text = text.replace("ゎ", " w a")
     text = text.replace("ぉ", " o")
 
+    # ここまでに処理されていないゅ等もそのまま大文字扱い（追加）
+    text = text.replace("ゃ", " y a")
+    text = text.replace("ゅ", " y u")
+    text = text.replace("ょ", " y o")
+
     # 長音の処理
     # for (pattern, replace_str) in JULIUS_LONG_VOWEL:
     #     text = pattern.sub(replace_str, text)
@@ -409,6 +415,8 @@ def text2sep_kata(text: str):
             else:
                 res.append(word)
         sep.append(word)
+    logger.debug(f"Text2sep_kata: {text} -> {res}")
+    logger.debug(f"get_accent: {parsed}")
     return sep, res, get_accent(parsed)
 
 
@@ -676,6 +684,9 @@ def rearrange_tones(tones, phones):
 
 def g2p(norm_text):
     sep_text, sep_kata, acc = text2sep_kata(norm_text)
+    logger.debug(f"Sep_text: {sep_text}")
+    logger.debug(f"Sep_kata: {sep_kata}")
+    logger.debug(f"Acc: {acc}")
     sep_tokenized = []
     for i in sep_text:
         if i not in punctuation:

From 896507f2cb1082f903a90595cf31a48cfe05ced7 Mon Sep 17 00:00:00 2001
From: litagin02 <litagin02@gmail.com>
Date: Sat, 6 Jan 2024 08:37:06 +0900
Subject: [PATCH 2/2] Fix wrong logger

---
 text/japanese.py | 5 -----
 1 file changed, 5 deletions(-)

diff --git a/text/japanese.py b/text/japanese.py
index 69301da..df6f59e 100644
--- a/text/japanese.py
+++ b/text/japanese.py
@@ -415,8 +415,6 @@ def text2sep_kata(text: str):
             else:
                 res.append(word)
         sep.append(word)
-    logger.debug(f"Text2sep_kata: {text} -> {res}")
-    logger.debug(f"get_accent: {parsed}")
     return sep, res, get_accent(parsed)
 
 
@@ -684,9 +682,6 @@ def rearrange_tones(tones, phones):
 
 def g2p(norm_text):
     sep_text, sep_kata, acc = text2sep_kata(norm_text)
-    logger.debug(f"Sep_text: {sep_text}")
-    logger.debug(f"Sep_kata: {sep_kata}")
-    logger.debug(f"Acc: {acc}")
     sep_tokenized = []
     for i in sep_text:
         if i not in punctuation: