Merge branch 'fast_inference_' of https://github.com/RVC-Boss/GPT-SoVITS

GPT_SoVITS/TTS_infer_pack/TTS.py

@@ -615,7 +615,7 @@ class TTS:
         fragment_interval = inputs.get("fragment_interval", 0.3)
         seed = inputs.get("seed", -1)
         seed = -1 if seed in ["", None] else seed
-        set_seed(seed)
+        actual_seed = set_seed(seed)
         
         if return_fragment:
             # split_bucket = False
@@ -824,13 +824,17 @@ class TTS:
 
             if not return_fragment:
                 print("%.3f\t%.3f\t%.3f\t%.3f" % (t1 - t0, t2 - t1, t_34, t_45))
-                yield self.audio_postprocess(audio, 
+                yield [
+                    self.audio_postprocess(audio, 
                                                 self.configs.sampling_rate, 
                                                 batch_index_list, 
                                                 speed_factor, 
                                                 split_bucket,
                                                 fragment_interval
-                                                )         
+                                                ), 
+                    f"<strong>text:</strong> {text} <strong>text_lang:</strong> {text_lang} <strong>prompt_text:</strong> {prompt_text} <strong>prompt_lang:</strong> {prompt_lang} <strong>top_k:</strong> {top_k} <strong>top_p:</strong> {top_p} <strong>temperature:</strong> {temperature} <strong>batch_size:</strong> {batch_size} <strong>batch_threshold:</strong> {batch_threshold} <strong>split_bucket:</strong> {split_bucket} <strong>return_fragment:</strong> {return_fragment} <strong>speed_factor:</strong> {speed_factor} <strong>fragment_interval:</strong> {fragment_interval} <strong>seed:</strong> {actual_seed}"
+                ]
+
         except Exception as e:
             traceback.print_exc()
             # 必须返回一个空音频, 否则会导致显存不释放。

GPT_SoVITS/TTS_infer_pack/text_segmentation_method.py

@@ -82,7 +82,8 @@ def cut1(inp):
     inp = inp.strip("\n")
     inps = split(inp)
     split_idx = list(range(0, len(inps), 4))
-    split_idx[-1] = None
+    # split_idx[-1] = None
+    split_idx.append(None)
     if len(split_idx) > 1:
         opts = []
         for idx in range(len(split_idx) - 1):
@@ -135,7 +136,8 @@ def cut5(inp):
     # if not re.search(r'[^\w\s]', inp[-1]):
     # inp += '。'
     inp = inp.strip("\n")
-    punds = r'[,.;?!、，。？！;：…]'
+    # punds = r'[,.;?!、，。？！;：…]'
+    punds = r'[,.;?!、，。？！；：:…]'
     items = re.split(f'({punds})', inp)
     mergeitems = ["".join(group) for group in zip(items[::2], items[1::2])]
     # 在句子不存在符号或句尾无符号的时候保证文本完整
@@ -149,4 +151,4 @@ def cut5(inp):
 if __name__ == '__main__':
     method = get_method("cut5")
     print(method("你好，我是小明。你好，我是小红。你好，我是小刚。你好，我是小张。"))
-    
+    

GPT_SoVITS/inference_webui.py

@@ -211,6 +211,8 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
                 with gr.Row():
                     inference_button = gr.Button(i18n("合成语音"), variant="primary")
                     stop_infer = gr.Button(i18n("终止合成"), variant="primary")
+                with gr.Row():
+                    inference_details = gr.Markdown()
                 
         
         inference_button.click(
@@ -224,7 +226,7 @@ with gr.Blocks(title="GPT-SoVITS WebUI") as app:
                 split_bucket,fragment_interval,
                 seed
              ],
-            [output],
+            [output, inference_details],
         )
         stop_infer.click(tts_pipline.stop, [], [])