Upload 3 files

app.py

@@ -10,9 +10,9 @@ logging.getLogger("torchaudio._extension").setLevel(logging.ERROR)
 import pdb
 
 gpt_path = os.environ.get(
-    "gpt_path", "models/XingTong/XingTong-e10.ckpt"
+    "gpt_path", "models/otto/otto-e10.ckpt"
 )
-sovits_path = os.environ.get("sovits_path", "models/XingTong/XingTong_e40_s3440.pth")
+sovits_path = os.environ.get("sovits_path", "models/otto/otto_e39_s1638.pth")
 cnhubert_base_path = os.environ.get(
     "cnhubert_base_path", "pretrained_models/chinese-hubert-base"
 )
@@ -427,6 +427,9 @@ def cut2(inp):
 def cut3(inp):
     inp = inp.strip("\n")
     return "\n".join(["%s。" % item for item in inp.strip("。").split("。")])
+def cut4(inp):
+    inp = inp.strip("\n")
+    return "\n".join(["%s." % item for item in inp.strip(".").split(".")])
 
 def scan_audio_files(folder_path):
     """ 扫描指定文件夹获取音频文件列表 """
@@ -446,17 +449,17 @@ 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/XingTong'
-text_to_audio_mappings, audio_to_text_mappings = load_audio_text_mappings(audio_folder_path, 'XingTong.list')
+audio_folder_path = 'audio/otto'
+text_to_audio_mappings, audio_to_text_mappings = load_audio_text_mappings(audio_folder_path, 'otto.list')
 
 with gr.Blocks(title="GPT-SoVITS WebUI") as app:
     gr.Markdown(value="""
-    # <center>【AI星瞳】在线语音生成（GPT-SoVITS）\n
+    # <center>【AI电棍】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>数据集下载：https://huggingface.co/datasets/XzJosh/audiodataset\n
-    ### <center>声音归属：星瞳_Official https://space.bilibili.com/401315430\n
+    ### <center>声音归属：电棍otto https://space.bilibili.com/628845081\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
@@ -464,9 +467,9 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
     # with gr.Tabs():
 
     with gr.Group():
-        gr.Markdown(value="*参考音频选择（必选）")
+        gr.Markdown(value="*参考音频选择（不建议选较长的）")
         with gr.Row():
-            audio_select = gr.Dropdown(label="选择参考音频（不建议选较长的）", choices=list(text_to_audio_mappings.keys()))
+            audio_select = gr.Dropdown(label="选择参考音频（必选）", choices=list(text_to_audio_mappings.keys()))
             ref_audio = gr.Audio(label="参考音频试听")
             ref_text = gr.Textbox(label="参考音频文本")
             
@@ -489,9 +492,9 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
                 label="需要合成的语种", choices=["中文", "英文", "日文"], value="中文"
             )
             how_to_cut = gr.Radio(
-                label=("怎么切"),
+                label=("自动切分（长文本建议切分）"),
                 choices=[("不切"),("凑五句一切"),("凑50字一切"),("按中文句号。切"),("按英文句号.切"),],
-                value=("凑50字一切"),
+                value=("不切"),
                 interactive=True,
             )
             inference_button = gr.Button("合成语音", variant="primary")
@@ -503,16 +506,18 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
         )
 
 
-    gr.Markdown(value="文本切分工具。太长的文本合成出来效果不一定好，所以太长建议先切。合成会根据文本的换行分开合成再拼起来。")
+    gr.Markdown(value="文本切分工具，需要复制。")
     with gr.Row():
         text_inp = gr.Textbox(label="需要合成的切分前文本", value="")
         button1 = gr.Button("凑五句一切", variant="primary")
         button2 = gr.Button("凑50字一切", variant="primary")
         button3 = gr.Button("按中文句号。切", variant="primary")
+        button4 = gr.Button("按英文句号.切", variant="primary")
         text_opt = gr.Textbox(label="切分后文本", value="")
         button1.click(cut1, [text_inp], [text_opt])
         button2.click(cut2, [text_inp], [text_opt])
         button3.click(cut3, [text_inp], [text_opt])
+        button4.click(cut4, [text_inp], [text_opt])
 
 app.queue(max_size=10)
 app.launch(inbrowser=True)

models/otto/otto-e10.ckpt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:02170064a054cc6e6532f884a00b8f509c2d345069c0b6c1c2dd9921f9e7b62f
+size 155084623

models/otto/otto_e39_s1638.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a3cd9a330d13c1d2ddaeb1ba4a6a2739cd8bcd386fbc2e6ec88f96f0e591c317
+size 84929843