Update app.py

app.py

@@ -24,14 +24,15 @@ def translate(audio):
 
 def synthesise(text):
     inputs = tokenizer(text=text, return_tensors="pt")
-    speech = model(**inputs).waveform
-    speech = speech.detach().cpu().numpy()
-    return speech
+    with torch.no_grad():
+        speech = model(**inputs).waveform
+    return speech.cpu()
 
 
 def speech_to_speech_translation(audio):
     translated_text = translate(audio)
     synthesised_speech = synthesise(translated_text)
+    synthesised_speech = (synthesised_speech.numpy() * 32767).astype(np.int16)
     return 16000, synthesised_speech