Improves audio generation robustness

Adds input validation and sanitization to audio generation functions to prevent errors caused by invalid or unexpected inputs.

Handles potential errors gracefully with fallback mechanisms, including a simplified audio tone as a last resort, enhancing the application's stability and user experience.

Also, adds limits to the random seed input to prevent errors.

app.py

@@ -6,12 +6,26 @@ def generate_audio_from_prompt(prompt, duration, seed):
     """
     Generate audio using simple synthesis based on prompt characteristics
     """
+    # Input validation and sanitization
+    if prompt is None:
+        prompt = "gentle melody"
+    if not isinstance(prompt, str):
+        prompt = str(prompt)
+    if duration is None or not isinstance(duration, (int, float)) or duration <= 0:
+        duration = 10.0  # Default duration
+    duration = min(max(duration, 1.0), 30.0)  # Clamp to reasonable range
+
     sample_rate = 44100
     duration_samples = int(duration * sample_rate)
 
-    # Set seed for reproducibility
+    # Set seed for reproducibility - handle None case explicitly
     if seed is not None:
-        np.random.seed(seed)
+        try:
+            seed_int = int(seed)
+            np.random.seed(seed_int)
+        except (ValueError, TypeError):
+            # If seed can't be converted to int, use system entropy
+            pass
 
     # Extract features from prompt to influence audio
     prompt_lower = prompt.lower()
@@ -77,6 +91,15 @@ def create_audio_generation_interface():
         Generate audio based on text prompt using intelligent synthesis
         """
         try:
+            # Input validation for main function
+            if prompt is None:
+                prompt = "gentle melody"
+            if not isinstance(prompt, str):
+                prompt = str(prompt)
+            if duration is None or not isinstance(duration, (int, float)):
+                duration = 10.0
+            duration = float(max(1.0, min(30.0, duration)))  # Ensure valid range
+
             print(f"Generating audio for prompt: '{prompt}', duration: {duration}s, seed: {seed}")
 
             # Use our intelligent synthesis function
@@ -86,13 +109,24 @@ def create_audio_generation_interface():
 
         except Exception as e:
             print(f"Error generating audio: {e}")
-            # Ultimate fallback
-            sample_rate = 44100
-            duration_samples = int(duration * sample_rate)
-            t = np.linspace(0, duration, duration_samples, endpoint=False)
-            audio = 0.3 * np.sin(2 * np.pi * 440 * t)  # Simple A4 tone
-
-            return (sample_rate, audio), f"Error: {str(e)}. Using simple fallback."
+            # Ultimate fallback with safety checks
+            try:
+                safe_duration = float(max(1.0, min(30.0, duration if isinstance(duration, (int, float)) else 10.0)))
+                sample_rate = 44100
+                duration_samples = int(safe_duration * sample_rate)
+                t = np.linspace(0, safe_duration, duration_samples, endpoint=False)
+                audio = 0.3 * np.sin(2 * np.pi * 440 * t)  # Simple A4 tone
+
+                return (sample_rate, audio), f"Error: {str(e)}. Using simple fallback."
+            except Exception as fallback_error:
+                print(f"Fallback also failed: {fallback_error}")
+                # Absolute minimum fallback
+                sample_rate = 44100
+                duration_samples = 441000  # 10 seconds
+                t = np.linspace(0, 10.0, duration_samples, endpoint=False)
+                audio = 0.3 * np.sin(2 * np.pi * 440 * t)
+
+                return (sample_rate, audio), "Critical error occurred. Using emergency fallback."
 
     # Create the Gradio interface
     with gr.Blocks(title="Stable Audio Open", theme=gr.themes.Soft()) as interface:
@@ -123,7 +157,9 @@ def create_audio_generation_interface():
                 seed_input = gr.Number(
                     label="Random Seed (optional)",
                     value=None,
-                    precision=0
+                    precision=0,
+                    minimum=0,
+                    maximum=999999  # Reasonable upper limit
                 )
 
                 generate_btn = gr.Button("🎵 Generate Audio", variant="primary")