Update app.py

app.py

@@ -129,11 +129,12 @@ def   modelspeech(text=TXT,name_model="wasmdashai/vits-ar-sa-huba-v2",speaking_r
      model=get_model(name_model)
      model.speaking_rate=speaking_rate
      with torch.no_grad():
-        wav=list(_inference_forward_stream(model,input_ids=inputs.input_ids,attention_mask=inputs.attention_mask,speaker_embeddings= None,is_streaming=False))[0]
+        wavs=_inference_forward_stream(model,input_ids=inputs.input_ids,attention_mask=inputs.attention_mask,speaker_embeddings= None,is_streaming=True)
      # with torch.no_grad():
      #      wav = model(input_ids=inputs["input_ids"].cuda()).waveform.cpu().numpy().reshape(-1)#.detach()
           
-     return  (model.config.sampling_rate,wav)
+     for   wav in wavs:
+          yield  (model.config.sampling_rate,wav)
 
 model_choices = gr.Dropdown(
                             choices=[