为api_v2和inference_webui_fast适配V3版本 (#2188)

* modified:   GPT_SoVITS/TTS_infer_pack/TTS.py
	modified:   GPT_SoVITS/TTS_infer_pack/TextPreprocessor.py
	modified:   GPT_SoVITS/inference_webui_fast.py

* 适配V3版本

* api_v2.py和inference_webui_fast.py的v3适配

* 修改了个远古bug,增加了更友好的提示信息

* 优化webui

* 修改为正确的path

* 修复v3 lora模型的载入问题

* 修复读取tts_infer.yaml文件时遇到的编码不匹配的问题

tools/audio_sr.py

@@ -39,6 +39,11 @@ class AP_BWE():
         self.model=model
         self.h=h
 
+    def to(self, *arg, **kwargs):
+        self.model.to(*arg, **kwargs)
+        self.device = self.model.conv_pre_mag.weight.device
+        return self
+
     def __call__(self, audio,orig_sampling_rate):
         with torch.no_grad():
             # audio, orig_sampling_rate = torchaudio.load(inp_path)