Upload 75 files

.gitattributes

@@ -32,16 +32,3 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 **/*ckpt*.index filter=lfs diff=lfs merge=lfs -text
 *.safetensors filter=lfs diff=lfs merge=lfs -text
 *.ckpt filter=lfs diff=lfs merge=lfs -texttext/cmudict_cache.pickle filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_102.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_148.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_163.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_173.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_242.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_516.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_623.wav filter=lfs diff=lfs merge=lfs -text
-audio/Azuma/Azuma_664.wav filter=lfs diff=lfs merge=lfs -text
-audio/Taffy/t2~1_234.wav filter=lfs diff=lfs merge=lfs -text
-audio/Taffy/t2~1_260.wav filter=lfs diff=lfs merge=lfs -text
-audio/Taffy/Taffy_242.wav filter=lfs diff=lfs merge=lfs -text
-audio/Taffy/Taffy_250.wav filter=lfs diff=lfs merge=lfs -text
-text/cmudict_cache.pickle filter=lfs diff=lfs merge=lfs -text

app.py

@@ -10,9 +10,9 @@ logging.getLogger("torchaudio._extension").setLevel(logging.ERROR)
 import pdb
 
 gpt_path = os.environ.get(
-    "gpt_path", "models/Taffy/Taffy-e5.ckpt"
+    "gpt_path", "models/XingTong/XingTong-e10.ckpt"
 )
-sovits_path = os.environ.get("sovits_path", "models/Taffy/Taffy_e20_s1020.pth")
+sovits_path = os.environ.get("sovits_path", "models/XingTong/XingTong_e40_s3440.pth")
 cnhubert_base_path = os.environ.get(
     "cnhubert_base_path", "pretrained_models/chinese-hubert-base"
 )
@@ -61,7 +61,7 @@ def get_bert_feature(text, word2ph):
     with torch.no_grad():
         inputs = tokenizer(text, return_tensors="pt")
         for i in inputs:
-            inputs[i] = inputs[i].to(device)  #####输入是long不用管精度问题，精度随bert_model
+            inputs[i] = inputs[i].to(device)  
         res = bert_model(**inputs, output_hidden_states=True)
         res = torch.cat(res["hidden_states"][-3:-2], -1)[0].cpu()[1:-1]
     assert len(word2ph) == len(text)
@@ -446,17 +446,17 @@ def load_audio_text_mappings(folder_path, list_file_name):
                 audio_to_text_mappings[audio_file_path] = text
     return text_to_audio_mappings, audio_to_text_mappings
 
-audio_folder_path = 'audio/Taffy'
-text_to_audio_mappings, audio_to_text_mappings = load_audio_text_mappings(audio_folder_path, 'Taffy.list')
+audio_folder_path = 'audio/XingTong'
+text_to_audio_mappings, audio_to_text_mappings = load_audio_text_mappings(audio_folder_path, 'XingTong.list')
 
 with gr.Blocks(title="GPT-SoVITS WebUI") as app:
     gr.Markdown(value="""
-    # <center>【AI塔菲】在线语音生成（GPT-SoVITS）\n
+    # <center>【AI星瞳】在线语音生成（GPT-SoVITS）\n
     
     ### <center>模型作者：Xz乔希 https://space.bilibili.com/5859321\n
-    ### <center>GPT-SoVITS在线合集：https://www.modelscope.cn/studios/xzjosh/GPT-SoVITS\n
+    ### <center>【GPT-SoVITS】在线合集：https://www.modelscope.cn/studios/xzjosh/GPT-SoVITS\n
     ### <center>数据集下载：https://huggingface.co/datasets/XzJosh/audiodataset\n
-    ### <center>声音归属：永雏塔菲 https://space.bilibili.com/1265680561\n
+    ### <center>声音归属：星瞳_Official https://space.bilibili.com/401315430\n
     ### <center>GPT-SoVITS项目：https://github.com/RVC-Boss/GPT-SoVITS\n
     ### <center>使用本模型请严格遵守法律法规！发布二创作品请标注本项目作者及链接、作品使用GPT-SoVITS AI生成！\n
     ### <center>⚠️在线端不稳定且生成速度较慢，强烈建议下载模型本地推理！\n
@@ -482,7 +482,7 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
         prompt_language = gr.Dropdown(
             label="参考音频语种", choices=["中文", "英文", "日文"], value="中文"
         )
-        gr.Markdown(value="*请填写需要合成的目标文本")
+        gr.Markdown(value="*请填写需要合成的目标文本，中英混合选中文，日英混合选日文，暂不支持中日混合。")
         with gr.Row():
             text = gr.Textbox(label="需要合成的文本", value="")
             text_language = gr.Dropdown(