push

app.py

@@ -2185,6 +2185,55 @@ Improved lyrics with fixed rhythm:
         # No analysis found, add a minimal one
         lyrics = lyrics + "\n\n[Note: Rhythm Analysis]\nNo rhythm issues detected. All syllables align well with the beat pattern."
     
+    # Store the syllable guidance for later use
+    syllable_guidance_text = syllable_guidance
+    
+    # Before returning, add syllable analysis and prompt template
+    if isinstance(lyrics, str):
+        # Extract clean lyrics and analysis
+        if "[Note: Rhythm Analysis]" in lyrics:
+            clean_lyrics = lyrics.split("[Note: Rhythm Analysis]")[0].strip()
+            rhythm_analysis = lyrics.split("[Note: Rhythm Analysis]")[1]
+        elif "[Note: Potential rhythm mismatches" in lyrics:
+            clean_lyrics = lyrics.split("[Note:")[0].strip()
+            rhythm_analysis = "[Note:" + lyrics.split("[Note:")[1]
+        else:
+            clean_lyrics = lyrics
+            rhythm_analysis = "No rhythm analysis available"
+        
+        # Create syllable analysis
+        syllable_analysis = "=== SYLLABLE ANALYSIS ===\n\n"
+        if templates_for_verification:
+            syllable_analysis += "Template Analysis:\n"
+            for i, template in enumerate(templates_for_verification):
+                syllable_analysis += f"Line {i+1}:\n"
+                if isinstance(template, dict):
+                    if "syllable_template" in template:
+                        syllable_analysis += f"  Template: {template['syllable_template']}\n"
+                    if "syllable_count" in template:
+                        syllable_analysis += f"  Expected syllables: {template['syllable_count']}\n"
+                elif isinstance(template, str):
+                    syllable_analysis += f"  Template: {template}\n"
+                syllable_analysis += "\n"
+        
+        # Create prompt template
+        prompt_template = "=== PROMPT TEMPLATE ===\n\n"
+        prompt_template += "Genre: " + genre + "\n"
+        prompt_template += f"Duration: {duration:.1f} seconds\n"
+        prompt_template += f"Tempo: {tempo:.1f} BPM\n"
+        prompt_template += f"Key: {key} {mode}\n"
+        prompt_template += f"Primary Emotion: {primary_emotion}\n"
+        prompt_template += f"Primary Theme: {primary_theme}\n\n"
+        prompt_template += "Syllable Guidance:\n" + syllable_guidance_text
+        
+        # Return all components
+        return {
+            "lyrics": clean_lyrics,
+            "rhythm_analysis": rhythm_analysis,
+            "syllable_analysis": syllable_analysis,
+            "prompt_template": prompt_template
+        }
+    
     return lyrics
 
 def process_audio(audio_file):
@@ -2245,31 +2294,37 @@ def process_audio(audio_file):
         # Generate lyrics based on top genre, emotion analysis, and song structure
         try:
             primary_genre, _ = top_genres[0]
-            lyrics = generate_lyrics(primary_genre, audio_data["duration"], emotion_results, song_structure)
+            lyrics_result = generate_lyrics(primary_genre, audio_data["duration"], emotion_results, song_structure)
+            
+            # Handle both old and new return formats
+            if isinstance(lyrics_result, dict):
+                lyrics = lyrics_result["lyrics"]
+                rhythm_analysis = lyrics_result["rhythm_analysis"]
+                syllable_analysis = lyrics_result["syllable_analysis"]
+                prompt_template = lyrics_result["prompt_template"]
+            else:
+                lyrics = lyrics_result
+                rhythm_analysis = "No detailed rhythm analysis available"
+                syllable_analysis = "No syllable analysis available"
+                prompt_template = "No prompt template available"
+                
         except Exception as e:
             print(f"Error generating lyrics: {str(e)}")
             lyrics = f"Error generating lyrics: {str(e)}"
+            rhythm_analysis = "No rhythm analysis available"
+            syllable_analysis = "No syllable analysis available"
+            prompt_template = "No prompt template available"
         
         # Prepare results dictionary with additional rhythm analysis
         results = {
             "genre_results": genre_results,
             "lyrics": lyrics,
+            "rhythm_analysis": rhythm_analysis,
+            "syllable_analysis": syllable_analysis,
+            "prompt_template": prompt_template,
             "ast_results": ast_results
         }
         
-        # Extract rhythm analysis if present in the lyrics
-        if isinstance(lyrics, str) and "[Note: Rhythm Analysis]" in lyrics:
-            clean_lyrics = lyrics.split("[Note: Rhythm Analysis]")[0].strip()
-            rhythm_analysis = "[Note: Rhythm Analysis]" + lyrics.split("[Note: Rhythm Analysis]")[1]
-            results["clean_lyrics"] = clean_lyrics
-            results["rhythm_analysis"] = rhythm_analysis
-        # Backwards compatibility with old format
-        elif isinstance(lyrics, str) and "[Note: Potential rhythm mismatches" in lyrics:
-            clean_lyrics = lyrics.split("[Note:")[0].strip()
-            rhythm_analysis = "[Note:" + lyrics.split("[Note:")[1]
-            results["clean_lyrics"] = clean_lyrics
-            results["rhythm_analysis"] = rhythm_analysis
-        
         return results
     
     except Exception as e:
@@ -2319,11 +2374,15 @@ with gr.Blocks(title="Music Genre Classifier & Lyrics Generator") as demo:
                 
                 with gr.TabItem("Rhythm Analysis"):
                     rhythm_analysis_output = gr.Textbox(label="Syllable-Beat Alignment Analysis", lines=16)
+                
+                with gr.TabItem("Syllable Analysis"):
+                    syllable_analysis_output = gr.Textbox(label="Detailed Syllable Analysis", lines=16)
+                    prompt_template_output = gr.Textbox(label="Prompt Template", lines=16)
     
     # Processing function with better handling of results
     def display_results(audio_file):
         if audio_file is None:
-            return "Please upload an audio file.", "No emotion analysis available.", "No audio classification available.", "No lyrics generated.", "No rhythm analysis available."
+            return "Please upload an audio file.", "No emotion analysis available.", "No audio classification available.", "No lyrics generated.", "No rhythm analysis available.", "No syllable analysis available.", "No prompt template available."
         
         try:
             # Process audio and get results
@@ -2331,9 +2390,9 @@ with gr.Blocks(title="Music Genre Classifier & Lyrics Generator") as demo:
             
             # Check if we got an error message instead of results
             if isinstance(results, str) and "Error" in results:
-                return results, "Error in analysis", "Error in classification", "No lyrics generated", "No rhythm analysis available"
+                return results, "Error in analysis", "Error in classification", "No lyrics generated", "No rhythm analysis available", "No syllable analysis available", "No prompt template available"
             elif isinstance(results, tuple) and isinstance(results[0], str) and "Error" in results[0]:
-                return results[0], "Error in analysis", "Error in classification", "No lyrics generated", "No rhythm analysis available"
+                return results[0], "Error in analysis", "Error in classification", "No lyrics generated", "No rhythm analysis available", "No syllable analysis available", "No prompt template available"
             
             # For backwards compatibility, handle both dictionary and tuple returns
             if isinstance(results, dict):
@@ -2344,6 +2403,10 @@ with gr.Blocks(title="Music Genre Classifier & Lyrics Generator") as demo:
                 # Use clean lyrics if available
                 clean_lyrics = results.get("clean_lyrics", lyrics)
                 rhythm_analysis = results.get("rhythm_analysis", "No detailed rhythm analysis available")
+                
+                # Extract syllable analysis and prompt template
+                syllable_analysis = results.get("syllable_analysis", "No syllable analysis available")
+                prompt_template = results.get("prompt_template", "No prompt template available")
             else:
                 # Handle the old tuple return format
                 genre_results, lyrics, ast_results = results
@@ -2360,6 +2423,10 @@ with gr.Blocks(title="Music Genre Classifier & Lyrics Generator") as demo:
                     elif "[Note: Potential rhythm mismatches" in lyrics:
                         clean_lyrics = lyrics.split("[Note:")[0].strip()
                         rhythm_analysis = "[Note:" + lyrics.split("[Note:")[1]
+                
+                # Default values for new fields
+                syllable_analysis = "No syllable analysis available"
+                prompt_template = "No prompt template available"
             
             # Format emotion analysis results
             try:
@@ -2412,19 +2479,19 @@ with gr.Blocks(title="Music Genre Classifier & Lyrics Generator") as demo:
             else:
                 ast_text = "No valid audio classification results available."
             
-            # Return all results for the tabbed interface
-            return genre_results, emotion_text, ast_text, clean_lyrics, rhythm_analysis
+            # Return all results including new fields
+            return genre_results, emotion_text, ast_text, clean_lyrics, rhythm_analysis, syllable_analysis, prompt_template
             
         except Exception as e:
             error_msg = f"Error: {str(e)}"
             print(error_msg)
-            return error_msg, "Error in emotion analysis", "Error in audio classification", "No lyrics generated", "No rhythm analysis available"
+            return error_msg, "Error in emotion analysis", "Error in audio classification", "No lyrics generated", "No rhythm analysis available", "No syllable analysis available", "No prompt template available"
     
-    # Connect the button to the display function
+    # Connect the button to the display function with updated outputs
     submit_btn.click(
         fn=display_results,
         inputs=[audio_input],
-        outputs=[genre_output, emotion_output, ast_output, lyrics_output, rhythm_analysis_output]
+        outputs=[genre_output, emotion_output, ast_output, lyrics_output, rhythm_analysis_output, syllable_analysis_output, prompt_template_output]
     )
     
     # Enhanced explanation of how the system works