Update app.py

app.py

@@ -201,7 +201,7 @@ def   modelspeech(text):
      audio_bytes = query({"inputs":text   })
      wav, sr = torchaudio.load(audio_bytes)
        
-     return sr,wav.squeeze().cpu().numpy()
+     yield sr,wav.squeeze().cpu().numpy()
      with torch.no_grad():
           inputs = tokenizer(text, return_tensors="pt")#.cuda()
 
@@ -236,11 +236,12 @@ def text_to_speech(text,session_ai):
            out=pad_text
            pad_text=''
            k+=1
-           yield modelspeech(out),session_ai
-           # for   stream_wav in generate_audio(out):
-           #      yield stream_wav
+           # yield modelspeech(out),session_ai
+           for   outmodel in modelspeech(out):
+                 yield outmodel,session_ai
     if pad_text!='':
-        yield modelspeech(pad_text),session_ai
+        for outmodel in modelspeech(pad_text):
+            yield outmodel,session_ai
 
        # for   stream_wav in generate_audio(pad_text):
        #          yield stream_wav