Update app.py

app.py

@@ -1,11 +1,12 @@
 import torch
 import torchaudio
-from sgmse.model import ScoreModel
 import gradio as gr
+from sgmse.model import ScoreModel
 from sgmse.util.other import pad_spec
 import time  # Import the time module
+import os
 
-# Define parameters based on the argparse configuration in enhancement.py
+# Define parameters based on the configuration in enhancement.py
 args = {
     "test_dir": "./test_data",  # example directory, adjust as needed
     "enhanced_dir": "./enhanced_data",  # example directory, adjust as needed
@@ -24,7 +25,7 @@ def enhance_speech(audio_file):
     start_time = time.time()  # Start the timer
 
     # Load and process the audio file
-    y, sr = torchaudio.load(audio_file)
+    y, sr = torchaudio.load(audio_file.name)  # Gradio passes the file as a file-like object
     print(f"Loaded audio in {time.time() - start_time:.2f}s")
     T_orig = y.size(1)
 
@@ -51,19 +52,21 @@ def enhance_speech(audio_file):
     # Renormalize
     x_hat = x_hat * norm_factor
 
-    # Save the enhanced audio
-    output_file = 'enhanced_output.wav'
+    # Save the enhanced audio to a temporary file for Gradio output
+    output_file = "enhanced_output.wav"
     torchaudio.save(output_file, x_hat.cpu(), sr)
 
     print(f"Processed audio in {time.time() - start_time:.2f}s")
+    
+    # Return the path to the enhanced file for Gradio to handle
     return output_file
 
 # Gradio interface setup
-inputs = gr.Audio(label="Input Audio", type="filepath")
-outputs = gr.Audio(label="Output Audio", type="filepath")
+inputs = gr.Audio(label="Input Audio", type="file")  # Adjusted for file input
+outputs = gr.Audio(label="Enhanced Audio", type="file")  # Output as file
 title = "Speech Enhancement using SGMSE"
 description = "This Gradio demo uses the SGMSE model for speech enhancement. Upload your audio file to enhance it."
 article = "<p style='text-align: center'><a href='https://huggingface.co/SP-UHH/speech-enhancement-sgmse' target='_blank'>Model Card</a></p>"
 
-# Launch without share=True (as it's not supported on Hugging Face Spaces)
+# Launch the Gradio interface
 gr.Interface(fn=enhance_speech, inputs=inputs, outputs=outputs, title=title, description=description, article=article).launch()