Update app.py

app.py

@@ -204,11 +204,14 @@ def predict(prompt, language, audio_file_pth, mic_file_path, use_mic, voice_clea
         wav = torch.cat(wav_chunks, dim=0)
         torchaudio.save("output.wav", wav.squeeze().unsqueeze(0).cpu(), 24000)
         
+        second_of_silence = AudioSegment.silent() # use default
+        second_of_silence.export("sil.wav", format='wav')
+        
         yield (
             gr.make_waveform(
                 audio="output.wav",
             ),
-            "output.wav",
+            "sil.wav",
             speaker_wav,
         )                
     else: