Update app.py

app.py

@@ -9,6 +9,8 @@ import librosa
 import numpy as np
 import soundfile
 from scipy.io import wavfile
+import edge_tts
+import tempfile
 
 from inference.infer_tool import Svc
 
@@ -50,21 +52,28 @@ def create_fn(model, spk):
         os.remove(temp_path)
         return sr, out_audio
 
-    def tts_fn(input_text, gender, tts_rate, vc_transform, auto_f0, f0p):
+    async def tts_fn(input_text, gender, tts_rate, vc_transform, auto_f0, f0p):
         if input_text == '':
             return 0, None
         input_text = re.sub(r"[\n\,\(\) ]", "", input_text)
         voice = tts_voice[gender]
         ratestr = "+{:.0%}".format(tts_rate) if tts_rate >= 0 else "{:.0%}".format(tts_rate)
-        temp_path = "temp.wav"
-        p = subprocess.Popen("edge-tts " +
-                             " --text " + input_text +
-                             " --write-media " + temp_path +
-                             " --voice " + voice +
-                             " --rate=" + ratestr, shell=True,
-                             stdout=subprocess.PIPE,
-                             stdin=subprocess.PIPE)
-        p.wait()
+        # temp_path = "temp.wav"
+        # p = subprocess.Popen("edge-tts " +
+        #                      " --text " + input_text +
+        #                      " --write-media " + temp_path +
+        #                      " --voice " + voice +
+        #                      " --rate=" + ratestr, shell=True,
+        #                      stdout=subprocess.PIPE,
+        #                      stdin=subprocess.PIPE)
+        # p.wait()
+        communicate = edge_tts.Communicate(text=input_text,
+                                           voice=voice,
+                                           rate=ratestr)
+        with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as tmp_file:
+            temp_path = tmp_file.name
+        await communicate.save(temp_path)
+
         audio, sr = librosa.load(temp_path)
         audio = librosa.resample(audio, orig_sr=sr, target_sr=sampling_rate)
         os.remove(temp_path)
@@ -93,7 +102,7 @@ if __name__ == '__main__':
         gr.Markdown(
             "# <center> 罪恶都市角色语音生成，即将上线...\n"
             "## <center> 模型作者：B站[Cyber蝈蝈总](https://space.bilibili.com/37706580)\n"
-            "<center> 圣安地列斯人物AI语音请移步[GTA_SOVITS](https://huggingface.co/spaces/GroveStreet/GTA_SOVITS)，使用此资源创作的作品请标明出处\n"
+            "<center> 圣安地列斯人物AI语音请移步[GTASA](https://huggingface.co/spaces/GroveStreet/GTA_SOVITS)，使用此资源创作的作品请标明出处\n"
         )
         with gr.Tabs():
             for (name, cover, (svc_fn, tts_fn)) in models: