Update app.py

app.py

@@ -372,7 +372,6 @@ def infer(ref_audio_orig, ref_text, gen_text, exp_name, remove_silence, cross_fa
     gr.Info("Converting audio...")
     with tempfile.NamedTemporaryFile(delete=False, suffix=".wav") as f:
         aseg = AudioSegment.from_file(ref_audio_orig)
-        aseg = remove_silence_edges(aseg) + AudioSegment.silent(duration=50)
 
         non_silent_segs = silence.split_on_silence(
             aseg, min_silence_len=1000, silence_thresh=-50, keep_silence=1000
@@ -380,8 +379,10 @@ def infer(ref_audio_orig, ref_text, gen_text, exp_name, remove_silence, cross_fa
         non_silent_wave = AudioSegment.silent(duration=0)
         for non_silent_seg in non_silent_segs:
             non_silent_wave += non_silent_seg
+        
         aseg = non_silent_wave
-
+        aseg = remove_silence_edges(aseg) + AudioSegment.silent(duration=50)
+        
         audio_duration = len(aseg)
         if audio_duration > 7500:
             gr.Warning("Audio is over 8s, clipping to only first 8s.")