feat: Add Support for v2 Model in Web UI

- Added support for the v2 model in the web UI.
- Implemented logic to handle v2-specific features, including the handling of prompt audio and disabling streaming inference for v2 models.
- Updated UI instructions to ensure users are properly guided when selecting the v2 model.

Author: lsby

webui.py

@@ -30,10 +30,10 @@
 instruct_dict = {'预训练音色': '1. 选择预训练音色\n2. 点击生成音频按钮',
                  '3s极速复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 输入prompt文本\n3. 点击生成音频按钮',
                  '跨语种复刻': '1. 选择prompt音频文件，或录入prompt音频，注意不超过30s，若同时提供，优先选择prompt音频文件\n2. 点击生成音频按钮',
-                 '自然语言控制': '1. 选择预训练音色\n2. 输入instruct文本\n3. 点击生成音频按钮'}
+                 '自然语言控制': '1. 选择预训练音色(v2模型需要选择或录入prompt音频)\n2. 输入instruct文本\n3. 点击生成音频按钮'}
 stream_mode_list = [('否', False), ('是', True)]
 max_val = 0.8
-
+model_versions = None
 
 def generate_seed():
     seed = random.randint(1, 100000000)
@@ -61,6 +61,10 @@ def change_instruction(mode_checkbox_group):
 
 def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, prompt_wav_upload, prompt_wav_record, instruct_text,
                    seed, stream, speed):
+    if model_versions == 'v2':
+        if stream == True:
+            stream = False
+            gr.Warning('您正在使用v2版本模型, 不支持流式推理, 将使用非流式模式.')
     if prompt_wav_upload is not None:
         prompt_wav = prompt_wav_upload
     elif prompt_wav_record is not None:
@@ -69,13 +73,13 @@ def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, pro
         prompt_wav = None
     # if instruct mode, please make sure that model is iic/CosyVoice-300M-Instruct and not cross_lingual mode
     if mode_checkbox_group in ['自然语言控制']:
-        if cosyvoice.instruct is False:
+        if cosyvoice.instruct is False and model_versions == 'v1':
             gr.Warning('您正在使用自然语言控制模式, {}模型不支持此模式, 请使用iic/CosyVoice-300M-Instruct模型'.format(args.model_dir))
             yield (cosyvoice.sample_rate, default_data)
         if instruct_text == '':
             gr.Warning('您正在使用自然语言控制模式, 请输入instruct文本')
             yield (cosyvoice.sample_rate, default_data)
-        if prompt_wav is not None or prompt_text != '':
+        if prompt_wav is not None or prompt_text != '' and model_versions == 'v1':
             gr.Info('您正在使用自然语言控制模式, prompt音频/prompt文本会被忽略')
     # if cross_lingual mode, please make sure that model is iic/CosyVoice-300M and tts_text prompt_text are different language
     if mode_checkbox_group in ['跨语种复刻']:
@@ -128,12 +132,20 @@ def generate_audio(tts_text, mode_checkbox_group, sft_dropdown, prompt_text, pro
         set_all_random_seed(seed)
         for i in cosyvoice.inference_cross_lingual(tts_text, prompt_speech_16k, stream=stream, speed=speed):
             yield (cosyvoice.sample_rate, i['tts_speech'].numpy().flatten())
-    else:
+    elif mode_checkbox_group == '自然语言控制':
         logging.info('get instruct inference request')
         set_all_random_seed(seed)
-        for i in cosyvoice.inference_instruct(tts_text, sft_dropdown, instruct_text, stream=stream, speed=speed):
-            yield (cosyvoice.sample_rate, i['tts_speech'].numpy().flatten())
-
+        if model_versions == 'v1':
+            for i in cosyvoice.inference_instruct(tts_text, sft_dropdown, instruct_text, stream=stream, speed=speed):
+                yield (cosyvoice.sample_rate, i['tts_speech'].numpy().flatten())
+        elif model_versions == 'v2':
+            prompt_speech_16k = postprocess(load_wav(prompt_wav, prompt_sr))
+            for i in cosyvoice.inference_instruct2(tts_text, instruct_text, prompt_speech_16k, stream=stream):
+                yield (cosyvoice.sample_rate, i['tts_speech'].numpy().flatten())
+        else:
+            gr.Warning('非预期的模型版本！')
+    else:
+        gr.Warning('非预期的选项！')
 
 def main():
     with gr.Blocks() as demo:
@@ -186,9 +198,11 @@ def main():
     args = parser.parse_args()
     try:
         cosyvoice = CosyVoice(args.model_dir)
+        model_versions = 'v1'
     except Exception:
         try:
             cosyvoice = CosyVoice2(args.model_dir)
+            model_versions = 'v2'
         except Exception:
             raise TypeError('no valid model_type!')