Update app.py

app.py

@@ -24,8 +24,9 @@ def predict(text, speaker):
 
     speech = model.generate_speech(input_ids, speaker_embedding, vocoder=vocoder)
 
-    speech = (speech.numpy() * 32767).astype(np.int16)
-    return (16000, speech)
+    # speech = (speech.numpy() * 32767).astype(np.int16)
+    # return (16000, speech)
+    return speech
 
 gr.Interface(
     fn=predict,