Update app.py (#3)

- Update app.py (9cc994f62cb8dc75b4b8612f8225c4c36f4bb0e5)


Co-authored-by: wang <[email protected]>

app.py

@@ -16,6 +16,14 @@ from huggingface_hub import hf_hub_download
 # 繁简转换
 converter = opencc.OpenCC('s2t.json')
 
+# 预设参考音频选项
+PRESET_AUDIO_OPTIONS = {
+    "ZoengJyutGaai": "asset/张悦楷.wav",
+    "Trump": "asset/特朗普.wav",
+    "Taiyi Zhenren": "asset/太乙真人.wav",
+    "自定义上传": None
+}
+
 # 加载模型
 cosyvoice_base = CosyVoice2(
     'ASLP-lab/Cosyvoice2-Yue',
@@ -33,18 +41,24 @@ print('load model 2')
 # )
 
 @spaces.GPU 
-def tts_inference(model_choice, text, prompt_audio):
+def tts_inference(model_choice, text, preset_audio_choice, prompt_audio):
     # 选择模型和默认音频
-    if model_choice == "CosyVoice2-张悦楷粤语评书":
+    if model_choice == "CosyVoice2-Yue-ZoengJyutGaai":
         model = cosyvoice_zjg
         prompt_audio = "asset/sg_017_090.wav"
     elif model_choice == "CosyVoice2-精品女音":
         model = cosyvoice_base
         prompt_audio = "asset/F01_中立_20054.wav"
-    elif model_choice == "CosyVoice2-base":
+    elif model_choice == "CosyVoice2-Yue":
         model = cosyvoice_base
-        if prompt_audio is None:
-            return None, "请上传参考音频"
+        if preset_audio_choice == "自定义上传":
+            if custom_audio is None:
+                return None, "请上传参考音频或选择预设音频"
+            prompt_audio = custom_audio
+        else:
+            prompt_audio = PRESET_AUDIO_OPTIONS[preset_audio_choice]
+            if prompt_audio is None:
+                return None, "请选择有效的参考音频"
     else:
         return None, "未知模型"
     
@@ -74,13 +88,18 @@ demo = gr.Interface(
     fn=tts_inference,
     inputs=[
         gr.Dropdown(
-            ["CosyVoice2-base", "CosyVoice2-张悦楷粤语评书"],
+            ["CosyVoice2-Yue", "CosyVoice2-Yue-ZoengJyutGaai"],
             # ["CosyVoice2-base", "CosyVoice2-张悦楷粤语评书", "CosyVoice2-精品女音"],
-            label="选择模型", value="CosyVoice2-base"
+            label="select model", value="CosyVoice2-Yue"
+        ),
+        gr.Textbox(lines=2, label="input text"),
+        gr.Dropdown(
+            choices=list(PRESET_AUDIO_OPTIONS.keys()),
+            value="ZoengJyutGaai",
+            label="please select audio options(only for CosyVoice2-Yue)"
         ),
-        gr.Textbox(lines=2, label="输入文本"),
-        # gr.Audio(source="upload", type="filepath", label="上传参考音频（仅 CosyVoice2-base 必需）")
-        gr.Audio(sources=["upload"], type="filepath", label="上传参考音频（仅 CosyVoice2-base 必需）")
+        # gr.Audio(source="upload", type="filepath", label="上传参考音频（仅 CosyVoice2-Yue 必需）")
+        gr.Audio(sources=["upload"], type="filepath", label="upload prompt audio(only for CosyVoice2-Yue)")
     ],
     outputs=[
         gr.Audio(type="numpy", label="生成的语音"),