Update app.py

app.py

@@ -114,23 +114,41 @@ def infer(audio_input_path):
 def load_chosen_audio(audio_path):
     return audio_path
 
-def overlay_audio(original_mp3: str, translated_wav: str) -> str:
+def overlay_audio(original_mp3: str, translated_wav: str, volume_reduction_db: int = 10) -> str:
+    """
+    Overlays translated audio on top of the original, reduces the original volume,
+    and ensures the final audio lasts as long as the longer of the two tracks.
+
+    :param original_mp3: Path to the original MP3 file.
+    :param translated_wav: Path to the translated WAV file.
+    :param volume_reduction_db: Volume reduction in dB (default is -10 dB).
+    :return: Path to the temporary output WAV file.
+    """
     # Load original MP3 and convert to WAV
     original = AudioSegment.from_mp3(original_mp3).set_frame_rate(16000).set_channels(1)
     
-    # Lower the volume (reduce by 6 dB, which is half)
-    original = original - 6
+    # Lower the volume (default is -10 dB)
+    original = original - volume_reduction_db
     
     # Load the translated WAV
     translated = AudioSegment.from_wav(translated_wav).set_frame_rate(16000).set_channels(1)
     
+    # Determine the final length (longer of the two)
+    final_length = max(len(original), len(translated))
+    
+    # Extend the shorter track with silence to match the longer track
+    if len(original) < final_length:
+        original = original + AudioSegment.silent(duration=final_length - len(original))
+    if len(translated) < final_length:
+        translated = translated + AudioSegment.silent(duration=final_length - len(translated))
+
     # Overlay the translated speech over the original
     combined = original.overlay(translated)
-    
+
     # Create a temporary file to save the output
     temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".wav")
     combined.export(temp_file.name, format="wav")
-    
+
     print(f"Final audio saved at: {temp_file.name}")
     return temp_file.name  # Return the temporary file path
 
@@ -164,7 +182,7 @@ with gr.Blocks(css=css) as demo:
                 visible=False,
                 scale=2
             )
-            choose_this_btn = gr.Button("Use this one", scale=1)
+            choose_this_btn = gr.Button("Use this one", scale=1, visible=False)
         combined_output = gr.Audio("Combined Outpu", visible=False)
         with gr.Accordion("Downloadable audio Output list", open=False, visible=False) as result_accordion:
             wav_list = gr.Files(label="Output Audio List", visible=False)