diff --git a/GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py b/GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py
index 426929f8..7aa419f0 100644
--- a/GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py
+++ b/GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py
@@ -61,6 +61,7 @@ class TextPreprocessor:
         text = self.replace_consecutive_punctuation(text)
         texts = self.pre_seg_text(text, lang, text_split_method)
         result = []
+        # text_batch = []
         print(f"############ {i18n('提取文本Bert特征')} ############")
         for text in tqdm(texts):
             phones, bert_features, norm_text = self.segment_and_extract_feature_for_text(text, lang, version)
@@ -73,6 +74,34 @@ class TextPreprocessor:
             }
             result.append(res)
         return result
+        # for text in texts:
+        #     if text.strip():  # 忽略空句子
+        #         text_batch.append(text)
+        # phones_list, bert_list, norm_texts = self.batch_get_phones_and_bert(text_batch, lang, version)
+        # for phones, bert_features, norm_text in zip(phones_list, bert_list, norm_texts):
+        #     if phones is None or norm_text == "":
+        #         continue
+        #     res = {
+        #         "phones": phones,
+        #         "bert_features": bert_features,
+        #         "norm_text": norm_text,
+        #     }
+        #     result.append(res)
+        # return result
+
+
+        # for text in tqdm(texts):
+        #     phones, bert_features, norm_text = self.segment_and_extract_feature_for_text(text, lang, version)
+        #     if phones is None or norm_text == "":
+        #         continue
+        #     res = {
+        #         "phones": phones,
+        #         "bert_features": bert_features,
+        #         "norm_text": norm_text,
+        #     }
+        #     result.append(res)
+        
+        # return result
 
     def pre_seg_text(self, text: str, lang: str, text_split_method: str):
         text = text.strip("\n")
@@ -235,3 +264,42 @@ class TextPreprocessor:
         pattern = f"([{punctuations}])([{punctuations}])+"
         result = re.sub(pattern, r"\1", text)
         return result
+
+
+    def batch_get_phones_and_bert(self, texts: List[str], language: str, version: str):
+        phones_list = []
+        bert_list = []
+        norm_text_list = []
+
+        # 预处理文本，获取每句的 phones, word2ph, norm_text
+        format_texts = [self.clean_text_inf(t, language, version) for t in texts]
+        norm_texts = [x[2] for x in format_texts]
+        word2phs = [x[1] for x in format_texts]
+
+        # 批量送入 tokenizer
+        inputs = self.tokenizer(norm_texts, return_tensors="pt", padding=True, truncation=True)
+        inputs = {k: v.to(self.device) for k, v in inputs.items()}
+
+        with torch.no_grad():
+            outputs = self.bert_model(**inputs, output_hidden_states=True)
+            #  使用 last_hidden_state 是正确且高效的方式
+            hidden_states = outputs.last_hidden_state  # [batch_size, seq_len, hidden_dim]
+
+        for i in range(len(texts)):
+            res = hidden_states[i][1:-1].cpu()  # 去掉 [CLS] 和 [SEP]
+
+            word2ph = word2phs[i]
+            phone_level_feature = []
+            for j in range(len(word2ph)):
+                if j >= res.shape[0]:
+                    print(f"警告：BERT输出不足，跳过第 {i} 句中第 {j} 个 token")
+                    continue
+                phone_level_feature.append(res[j].repeat(word2ph[j], 1))
+            phone_level_feature = torch.cat(phone_level_feature, dim=0)
+
+            bert_list.append(phone_level_feature.T)
+            phones_list.append(cleaned_text_to_sequence(format_texts[i][0], version))
+            norm_text_list.append(norm_texts[i])
+
+        return phones_list, bert_list, norm_text_list
+
diff --git a/GPT_SoVITS/configs/tts_infer.yaml b/GPT_SoVITS/configs/tts_infer.yaml
index e2c13c28..20c41a20 100644
--- a/GPT_SoVITS/configs/tts_infer.yaml
+++ b/GPT_SoVITS/configs/tts_infer.yaml
@@ -1,8 +1,8 @@
 custom:
   bert_base_path: GPT_SoVITS/pretrained_models/chinese-roberta-wwm-ext-large
   cnhuhbert_base_path: GPT_SoVITS/pretrained_models/chinese-hubert-base
-  device: cpu
-  is_half: false
+  device: cuda
+  is_half: true
   t2s_weights_path: GPT_SoVITS/pretrained_models/gsv-v2final-pretrained/s1bert25hz-5kh-longer-epoch=12-step=369668.ckpt
   version: v2
   vits_weights_path: GPT_SoVITS/pretrained_models/gsv-v2final-pretrained/s2G2333k.pth
diff --git a/GPT_SoVITS/export_torch_script_v3.py b/GPT_SoVITS/export_torch_script_v3.py
index 2f57df0e..b34495a7 100644
--- a/GPT_SoVITS/export_torch_script_v3.py
+++ b/GPT_SoVITS/export_torch_script_v3.py
@@ -26,7 +26,7 @@ from inference_webui import get_spepc, norm_spec, resample, ssl_model
 logging.config.dictConfig(uvicorn.config.LOGGING_CONFIG)
 logger = logging.getLogger("uvicorn")
 
-is_half = False
+is_half = True
 device = "cuda" if torch.cuda.is_available() else "cpu"
 now_dir = os.getcwd()
 
diff --git a/GPT_SoVITS/inference_webui_fast.py b/GPT_SoVITS/inference_webui_fast.py
index 02686a00..b2a02685 100644
--- a/GPT_SoVITS/inference_webui_fast.py
+++ b/GPT_SoVITS/inference_webui_fast.py
@@ -69,7 +69,7 @@ if torch.cuda.is_available():
 else:
     device = "cpu"
 
-is_half = False
+is_half = True
 device = "cpu"
 
 dict_language_v1 = {
diff --git a/output.wav b/output.wav
index 8b3c38a8..e533dc26 100644
Binary files a/output.wav and b/output.wav differ