Update app.py

app.py

@@ -14,7 +14,8 @@ from voice_synthesizer import VoiceSynthesizer
 from singing_converter import SingingConverter
 import setup
 import sys
-import nltk
+import subprocess
+
 nltk.download('punkt')
 nltk.download('punkt_tab')
 nltk.download('stopwords')
@@ -41,6 +42,38 @@ singing_converter = SingingConverter()
 # Setup sentiment analysis
 sentiment_analyzer = pipeline("sentiment-analysis")
 
+def create_placeholder_audio(output_path, duration=5, sample_rate=22050):
+    """Create a placeholder silence audio file"""
+    silence = np.zeros(int(duration * sample_rate))
+    sf.write(output_path, silence, sample_rate)
+    return output_path
+
+def convert_midi_to_wav(midi_path, wav_path, soundfont_path='/usr/share/sounds/sf2/FluidR3_GM.sf2'):
+    """Convert MIDI file to WAV using fluidsynth"""
+    # Check if the MIDI file exists
+    if not os.path.exists(midi_path):
+        print(f"MIDI file not found: {midi_path}")
+        print("Creating placeholder audio file instead")
+        return create_placeholder_audio(wav_path)
+        
+    try:
+        # Use fluidsynth to convert MIDI to WAV
+        subprocess.run([
+            'fluidsynth', 
+            '-a', 'file', 
+            '-F', wav_path, 
+            soundfont_path, 
+            midi_path
+        ], check=True)
+        
+        return wav_path
+    except subprocess.CalledProcessError as e:
+        print(f"Error converting MIDI to WAV: {e}")
+        return create_placeholder_audio(wav_path)
+    except FileNotFoundError:
+        print("fluidsynth not found. Using placeholder audio instead.")
+        return create_placeholder_audio(wav_path)
+
 def process_text_to_singing(text, voice_type="neutral", tempo=100, pitch_shift=0):
     """
     Convert text to singing voice with accompaniment based on mood
@@ -56,7 +89,7 @@ def process_text_to_singing(text, voice_type="neutral", tempo=100, pitch_shift=0
     """
     # Step 1: Analyze text for emotion/mood
     emotions = te.get_emotion(text)
-    dominant_emotion = max(emotions.items(), key=lambda x: x[1])[0]
+    dominant_emotion = max(emotions.items(), key=lambda x: x[1])[0] if emotions else "Happy"
     
     # Additional sentiment analysis
     sentiment_result = sentiment_analyzer(text)[0]
@@ -90,7 +123,7 @@ def process_text_to_singing(text, voice_type="neutral", tempo=100, pitch_shift=0
     )
     
     # Step 5: Generate musical accompaniment based on mood
-    accompaniment_path = "temp_accompaniment.wav"
+    accompaniment_midi_path = "temp_accompaniment.mid"
     
     # Map emotion to musical key and style
     emotion_key_map = {
@@ -106,27 +139,39 @@ def process_text_to_singing(text, voice_type="neutral", tempo=100, pitch_shift=0
     
     # Adjust tempo based on emotion if not explicitly set
     tempo_value = tempo
-    accompaniment_midi_path = "temp_accompaniment.mid"
     
-    generate_accompaniment(
-        lyrics=text,
-        melody_path=singing_audio_path,
-        output_path=accompaniment_path,
-        tempo_value=tempo_value,
-        key=key,
-        time_signature="4/4",
-        style=style
-    )
-
+    try:
+        # Try to generate the accompaniment MIDI
+        generate_accompaniment(
+            lyrics=text,
+            melody_path=singing_audio_path,
+            output_path=accompaniment_midi_path,
+            tempo_value=tempo_value,
+            key=key,
+            time_signature="4/4",
+            style=style
+        )
+    except Exception as e:
+        print(f"Error generating accompaniment: {e}")
+        # We'll handle this with the convert_midi_to_wav function that creates a placeholder
+    
+    # Convert MIDI to WAV
     accompaniment_path = "temp_accompaniment.wav"
     convert_midi_to_wav(accompaniment_midi_path, accompaniment_path)
     
     # Step 6: Mix singing voice with accompaniment
     final_output_path = "output_song.wav"
     
-    # Load audio files
+    # Load singing audio
     singing = AudioSegment.from_file(singing_audio_path)
-    accompaniment = AudioSegment.from_file(accompaniment_path)
+    
+    # Load accompaniment or create placeholder if loading fails
+    try:
+        accompaniment = AudioSegment.from_file(accompaniment_path)
+    except Exception as e:
+        print(f"Error loading accompaniment: {e}")
+        create_placeholder_audio(accompaniment_path)
+        accompaniment = AudioSegment.from_file(accompaniment_path)
     
     # Adjust volumes
     singing = singing - 3  # Reduce singing volume slightly
@@ -146,32 +191,6 @@ def process_text_to_singing(text, voice_type="neutral", tempo=100, pitch_shift=0
     
     return speech_audio_path, final_output_path
 
-def convert_midi_to_wav(midi_path, wav_path, soundfont_path='/usr/share/sounds/sf2/FluidR3_GM.sf2'):
-    """Convert MIDI file to WAV using fluidsynth"""
-    import subprocess
-    
-    # Check if the MIDI file exists
-    if not os.path.exists(midi_path):
-        raise FileNotFoundError(f"MIDI file not found: {midi_path}")
-        
-    try:
-        # Use fluidsynth to convert MIDI to WAV
-        subprocess.run([
-            'fluidsynth', 
-            '-a', 'file', 
-            '-F', wav_path, 
-            soundfont_path, 
-            midi_path
-        ], check=True)
-        
-        return wav_path
-    except subprocess.CalledProcessError as e:
-        print(f"Error converting MIDI to WAV: {e}")
-        raise
-    except FileNotFoundError:
-        print("fluidsynth not found. Please install it.")
-        raise
-
 # Create Gradio interface
 with gr.Blocks(title="Text2Sing-DiffSinger") as demo:
     gr.Markdown("# Text2Sing-DiffSinger")