Update app.py

app.py

@@ -32,7 +32,11 @@ def translate(audio):
 def synthesise(text):
     inputs = tokenizer(text, return_tensors="pt")
     # inputs = processor(text=text, return_tensors="pt")
-    speech = model.generate_speech(inputs["input_ids"].to(device), speaker_embeddings.to(device), vocoder=vocoder)
+    with torch.no_grad():
+        outputs = model(inputs["input_ids"])
+
+    speech = outputs.audio[0]
+    # speech = model.generate_speech(inputs["input_ids"].to(device), speaker_embeddings.to(device), vocoder=vocoder)
     return speech.cpu()