Update app.py

app.py

@@ -149,14 +149,11 @@ def spectrogram_to_audio(magnitude_spectrogram):
     return audio
 
 
-import numpy as np
-
 def generate_audio_from_image(image):
     if image is None:
         raise ValueError("The uploaded image is 'None'. Please check the Gradio input.")
 
     # Ensure the image is in the right format
-    print(f"Image received: {type(image)}")  # Debugging: Check if image is received
     test_img = image_transform(image).unsqueeze(0).to(device)  # Preprocess image
 
     # Generate sound spectrogram from the image using the loaded generator
@@ -177,14 +174,11 @@ def generate_audio_from_image(image):
     # Convert to the required format (e.g., float32)
     generated_audio = generated_audio.astype(np.float32)
 
-    # Transpose to make it (samples, channels)
+    # Transpose to (samples, channels) for stereo
     generated_audio = generated_audio.T
 
-    # Debug: Print the shape and type of the generated audio
-    print(f"Generated audio shape after transpose: {generated_audio.shape}, type: {generated_audio.dtype}")
-
-    return generated_audio, sample_rate
-
+    # Return the audio and the sample rate (ensure sample rate is an integer)
+    return generated_audio, int(sample_rate)
 
 
 # Gradio Interface