From 87bea1fef2a655ec28674e45030c730e357a42e4 Mon Sep 17 00:00:00 2001
From: Dan Sadler <2017586+Sadler2@users.noreply.github.com>
Date: Thu, 6 Mar 2025 22:19:26 +0600
Subject: [PATCH] Ensure UTF-8 encoding for file reads to support Windows
 MaskGCT inference

This patch explicitly sets `encoding="utf-8"` when reading JSON files to prevent encoding issues on Windows.
---
 models/tts/maskgct/g2p/g2p/__init__.py   | 2 +-
 models/tts/maskgct/g2p/g2p_generation.py | 2 +-
 models/tts/maskgct/g2p/utils/g2p.py      | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/models/tts/maskgct/g2p/g2p/__init__.py b/models/tts/maskgct/g2p/g2p/__init__.py
index ca96b67e..6a750c56 100644
--- a/models/tts/maskgct/g2p/g2p/__init__.py
+++ b/models/tts/maskgct/g2p/g2p/__init__.py
@@ -25,7 +25,7 @@ def __init__(self, vacab_path="./models/tts/maskgct/g2p/g2p/vocab.json"):
         self.text_tokenizers = {}
         self.int_text_tokenizers()
 
-        with open(vacab_path, "r") as f:
+        with open(vacab_path, "r", encoding="utf-8") as f:
             json_data = f.read()
         data = json.loads(json_data)
         self.vocab = data["vocab"]
diff --git a/models/tts/maskgct/g2p/g2p_generation.py b/models/tts/maskgct/g2p/g2p_generation.py
index ca3d9841..69f99d20 100644
--- a/models/tts/maskgct/g2p/g2p_generation.py
+++ b/models/tts/maskgct/g2p/g2p_generation.py
@@ -114,7 +114,7 @@ def chn_eng_g2p(text: str):
 
 
 text_tokenizer = PhonemeBpeTokenizer()
-with open("./models/tts/maskgct/g2p/g2p/vocab.json", "r") as f:
+with open("./models/tts/maskgct/g2p/g2p/vocab.json", "r", encoding="utf-8") as f:
     json_data = f.read()
 data = json.loads(json_data)
 vocab = data["vocab"]
diff --git a/models/tts/maskgct/g2p/utils/g2p.py b/models/tts/maskgct/g2p/utils/g2p.py
index f71e0c8d..f586ba57 100644
--- a/models/tts/maskgct/g2p/utils/g2p.py
+++ b/models/tts/maskgct/g2p/utils/g2p.py
@@ -60,7 +60,7 @@
     "de": phonemizer_de,
 }
 
-with open("./models/tts/maskgct/g2p/utils/mls_en.json", "r") as f:
+with open("./models/tts/maskgct/g2p/utils/mls_en.json", "r", encoding="utf-8") as f:
     json_data = f.read()
 token = json.loads(json_data)