fix

app.py

@@ -117,8 +117,16 @@ def generate_speech(
     else:
         audio_np = generated_audio
 
-    if audio_np.ndim == 1:
-        audio_np = np.expand_dims(audio_np, axis=0)
+    # Ensure audio is properly normalized and in the correct format
+    if audio_np.ndim == 2 and audio_np.shape[0] == 1:
+        audio_np = audio_np.squeeze(0)  # Remove batch dimension if present
+    
+    # Normalize audio to [-1, 1] range if needed
+    if np.abs(audio_np).max() > 1.0:
+        audio_np = audio_np / np.abs(audio_np).max()
+    
+    # Ensure audio is in float32 format
+    audio_np = audio_np.astype(np.float32)
 
     return (24000, audio_np)