Update app.py

app.py

@@ -49,11 +49,11 @@ def predict(text, speaker):
     else:
         speaker_embedding = np.load(speaker_embeddings[speaker[:3]])
 
-    speaker_embedding = torch.tensor(speaker_embedding).unsqueeze(0)
+    #speaker_embedding = torch.tensor(speaker_embedding).unsqueeze(0)
 
     speech = model.generate_speech(input_ids, speaker_embedding, vocoder=vocoder)
 
-    speech = (speech.numpy() * 32767).astype(np.int16)
+    #speech = (speech.numpy() * 32767).astype(np.int16)
     return (16000, speech)