Update app.py

app.py

@@ -26,7 +26,7 @@ def check_port(port):
         except socket.error:
             return False
 
-def trim_silence(audio_clip, threshold=0.01):
+def trim_silence(audio_clip, threshold=0.005):
     """
     Trim silence from the start and end of an audio clip.
     Args:
@@ -35,32 +35,41 @@ def trim_silence(audio_clip, threshold=0.01):
     Returns:
         Trimmed AudioFileClip
     """
-    # Get audio data as numpy array
-    samples = audio_clip.to_soundarray(fps=44100)
-    # Compute amplitude (RMS)
-    if len(samples.shape) > 1:  # Stereo audio
-        amplitudes = np.sqrt(np.mean(samples**2, axis=1))
-    else:  # Mono audio
-        amplitudes = np.sqrt(samples**2)
-    
-    # Find non-silent regions
-    non_silent = amplitudes > threshold
-    if not np.any(non_silent):
-        logger.warning("Audio clip is completely silent")
+    try:
+        # Get audio data as numpy array
+        samples = audio_clip.to_soundarray(fps=44100)
+        # Compute amplitude (RMS)
+        if len(samples.shape) > 1:  # Stereo audio
+            amplitudes = np.sqrt(np.mean(samples**2, axis=1))
+        else:  # Mono audio
+            amplitudes = np.sqrt(samples**2)
+        
+        # Find non-silent regions
+        non_silent = amplitudes > threshold
+        if not np.any(non_silent):
+            logger.warning("Audio clip is completely silent; returning original clip")
+            return audio_clip
+        
+        # Find start and end indices
+        start_idx = np.argmax(non_silent)
+        end_idx = len(non_silent) - np.argmax(non_silent[::-1])
+        
+        # Convert indices to time (seconds)
+        start_time = start_idx / 44100
+        end_time = end_idx / 44100
+        
+        # Ensure the trimmed duration is reasonable
+        if end_time <= start_time:
+            logger.warning("Trimmed duration is zero or negative; returning original clip")
+            return audio_clip
+        
+        # Trim the audio
+        trimmed_audio = audio_clip.subclip(start_time, end_time)
+        logger.info(f"Trimmed audio from {start_time:.2f}s to {end_time:.2f}s (original duration: {audio_clip.duration:.2f}s)")
+        return trimmed_audio
+    except Exception as e:
+        logger.error(f"Error trimming silence: {str(e)}")
         return audio_clip
-    
-    # Find start and end indices
-    start_idx = np.argmax(non_silent)
-    end_idx = len(non_silent) - np.argmax(non_silent[::-1])
-    
-    # Convert indices to time (seconds)
-    start_time = start_idx / 44100
-    end_time = end_idx / 44100
-    
-    # Trim the audio
-    trimmed_audio = audio_clip.subclip(start_time, end_time)
-    logger.info(f"Trimmed audio from {start_time:.2f}s to {end_time:.2f}s (original duration: {audio_clip.duration:.2f}s)")
-    return trimmed_audio
 
 def merge_videos_and_audios(video_files=None, audio_files=None, orig_vol=1.0, music_vol=0.5):
     """
@@ -101,17 +110,35 @@ def merge_videos_and_audios(video_files=None, audio_files=None, orig_vol=1.0, mu
             audio_clips = []
             for audio in audio_files:
                 clip = AudioFileClip(audio).set_fps(44100)  # Normalize sample rate
-                logger.info(f"Original audio duration for {audio}: {clip.duration:.2f}s")
+                clip = clip.to_stereo() if clip.nchannels == 1 else clip  # Convert mono to stereo
+                logger.info(f"Original audio duration for {audio}: {clip.duration:.2f}s, channels: {clip.nchannels}")
                 trimmed_clip = trim_silence(clip)
-                audio_clips.append(trimmed_clip)
+                if trimmed_clip.duration > 0:
+                    audio_clips.append(trimmed_clip)
+                else:
+                    logger.warning(f"Skipping audio file {audio} as it has zero duration after trimming")
+            
+            # Check if we have enough clips to concatenate
+            if len(audio_clips) < 2:
+                error_msg = "Error: Fewer than 2 audio clips available after trimming (clips may be silent or too short)."
+                logger.error(error_msg)
+                for clip in audio_clips:
+                    clip.close()
+                return error_msg
             
             # Log durations after trimming
             for i, clip in enumerate(audio_clips):
-                logger.info(f"Trimmed audio {i+1} duration: {clip.duration:.2f}s")
+                logger.info(f"Trimmed audio {i+1} duration: {clip.duration:.2f}s, channels: {clip.nchannels}")
             
             # Concatenate audio clips
+            logger.info(f"Attempting to concatenate {len(audio_clips)} audio clips")
             final_audio_clip = concatenate_audioclips(audio_clips)
-            logger.info(f"Concatenated audio duration: {final_audio_clip.duration:.2f}s")
+            logger.info(f"Concatenated audio duration: {final_audio_clip.duration:.2f}s, channels: {final_audio_clip.nchannels}")
+            
+            # Verify concatenated duration
+            expected_duration = sum(clip.duration for clip in audio_clips)
+            if abs(final_audio_clip.duration - expected_duration) > 0.1:
+                logger.warning(f"Concatenated duration ({final_audio_clip.duration:.2f}s) does not match expected duration ({expected_duration:.2f}s)")
             
             # Write the final audio
             logger.info(f"Writing output audio to {output_path}")
@@ -143,38 +170,52 @@ def merge_videos_and_audios(video_files=None, audio_files=None, orig_vol=1.0, mu
             audio_clips = []
             for audio in audio_files:
                 clip = AudioFileClip(audio).set_fps(44100)  # Normalize sample rate
-                logger.info(f"Original audio duration for {audio}: {clip.duration:.2f}s")
+                clip = clip.to_stereo() if clip.nchannels == 1 else clip  # Convert mono to stereo
+                logger.info(f"Original audio duration for {audio}: {clip.duration:.2f}s, channels: {clip.nchannels}")
                 trimmed_clip = trim_silence(clip)
-                audio_clips.append(trimmed_clip)
+                if trimmed_clip.duration > 0:
+                    audio_clips.append(trimmed_clip)
+                else:
+                    logger.warning(f"Skipping audio file {audio} as it has zero duration after trimming")
             
             # Log durations after trimming
             for i, clip in enumerate(audio_clips):
-                logger.info(f"Trimmed audio {i+1} duration: {clip.duration:.2f}s")
-            
-            # Concatenate audio clips
-            concatenated_audio = concatenate_audioclips(audio_clips)
-            logger.info(f"Concatenated audio duration: {concatenated_audio.duration:.2f}s")
+                logger.info(f"Trimmed audio {i+1} duration: {clip.duration:.2f}s, channels: {clip.nchannels}")
             
-            # Adjust concatenated audio duration to match video duration (trim or loop)
-            if concatenated_audio.duration > video_duration:
-                concatenated_audio = concatenated_audio.subclip(0, video_duration)
-                logger.info(f"Trimmed concatenated audio to match video duration: {concatenated_audio.duration:.2f}s")
-            elif concatenated_audio.duration < video_duration:
-                # Loop the audio to match video duration
-                concatenated_audio = concatenated_audio.fx(lambda clip: clip.loop(duration=video_duration))
-                logger.info(f"Looped concatenated audio to match video duration: {concatenated_audio.duration:.2f}s")
-            
-            # Apply volume to concatenated audio
-            concatenated_audio = concatenated_audio.volumex(music_vol)
-            
-            # Get original video audio (if any) and apply volume
-            original_audio = final_video_clip.audio.volumex(orig_vol) if final_video_clip.audio else None
-            
-            # Composite the audio tracks
-            if original_audio:
-                final_audio = CompositeAudioClip([original_audio, concatenated_audio])
+            if not audio_clips:
+                logger.warning("No valid audio clips after trimming; using original video audio only")
+                final_audio = final_video_clip.audio.volumex(orig_vol) if final_video_clip.audio else None
             else:
-                final_audio = concatenated_audio
+                # Concatenate audio clips
+                logger.info(f"Attempting to concatenate {len(audio_clips)} audio clips")
+                concatenated_audio = concatenate_audioclips(audio_clips)
+                logger.info(f"Concatenated audio duration: {concatenated_audio.duration:.2f}s, channels: {concatenated_audio.nchannels}")
+                
+                # Verify concatenated duration
+                expected_duration = sum(clip.duration for clip in audio_clips)
+                if abs(concatenated_audio.duration - expected_duration) > 0.1:
+                    logger.warning(f"Concatenated duration ({concatenated_audio.duration:.2f}s) does not match expected duration ({expected_duration:.2f}s)")
+                
+                # Adjust concatenated audio duration to match video duration (trim or loop)
+                if concatenated_audio.duration > video_duration:
+                    concatenated_audio = concatenated_audio.subclip(0, video_duration)
+                    logger.info(f"Trimmed concatenated audio to match video duration: {concatenated_audio.duration:.2f}s")
+                elif concatenated_audio.duration < video_duration:
+                    # Loop the audio to match video duration
+                    concatenated_audio = concatenated_audio.fx(lambda clip: clip.loop(duration=video_duration))
+                    logger.info(f"Looped concatenated audio to match video duration: {concatenated_audio.duration:.2f}s")
+                
+                # Apply volume to concatenated audio
+                concatenated_audio = concatenated_audio.volumex(music_vol)
+                
+                # Get original video audio (if any) and apply volume
+                original_audio = final_video_clip.audio.volumex(orig_vol) if final_video_clip.audio else None
+                
+                # Composite the audio tracks
+                if original_audio:
+                    final_audio = CompositeAudioClip([original_audio, concatenated_audio])
+                else:
+                    final_audio = concatenated_audio
         else:
             logger.info("No audio files provided; using original video audio if available")
             # If no audio files provided, retain original video audio (if any)
@@ -191,10 +232,11 @@ def merge_videos_and_audios(video_files=None, audio_files=None, orig_vol=1.0, mu
         final_video_clip.close()
         for clip in video_clips:
             clip.close()
-        if audio_files:
+        if audio_files and audio_clips:
             for clip in audio_clips:
                 clip.close()
-            concatenated_audio.close()
+            if 'concatenated_audio' in locals():
+                concatenated_audio.close()
         
         logger.info("Video merge completed successfully")
         return output_path