Update app.py

app.py

@@ -172,6 +172,7 @@ def infer_batch(ref_audio, ref_text, gen_text_batches, exp_name, remove_silence,
     elif exp_name == "Deutsh":
         ema_model = E2TTS_ema_model2
     #ref_audio, ref_text = preprocess_ref_audio_text(ref_audio, ref_text, show_info=show_info)
+    ref_audio = remove_silence_edges(ref_audio) + AudioSegment.silent(duration=50)
 
     audio, sr = ref_audio
     if audio.shape[0] > 1:
@@ -183,7 +184,7 @@ def infer_batch(ref_audio, ref_text, gen_text_batches, exp_name, remove_silence,
     if sr != target_sample_rate:
         resampler = torchaudio.transforms.Resample(sr, target_sample_rate)
         audio = resampler(audio)
-    audio = remove_silence_edges(audio) + AudioSegment.silent(duration=50)
+    
 
     audio = audio.to(device)