Update app.py

app.py

@@ -1,8 +1,8 @@
 import gradio as gr
 import torch
 import torchaudio
-from speechbrain.inference.enhancement import SpectralMaskEnhancement
 from speechbrain.inference.separation import SepformerSeparation as separator
+import os
 
 # Load the enhancement model
 model = separator.from_hparams(
@@ -12,16 +12,23 @@ model = separator.from_hparams(
 
 # Define the enhancement function
 def enhance_audio(noisy_audio):
+    # Convert MP3 to WAV
+    wav_audio = "temp_audio.wav"
+    torchaudio.save(wav_audio, *torchaudio.load(noisy_audio))
+
     # Load and add a batch dimension to the audio tensor
-    noisy = model.load_audio(noisy_audio).unsqueeze(0)
+    noisy = model.load_audio(wav_audio).unsqueeze(0)
 
     # Enhance the audio
     enhanced = model.enhance_batch(noisy, lengths=torch.tensor([1.0]))
 
-    # Save enhanced audio to a temporary file
+    # Save enhanced audio to a file
     enhanced_path = "enhanced.wav"
     torchaudio.save(enhanced_path, enhanced.cpu(), 16000)
     
+    # Clean up the temporary audio file
+    os.remove(wav_audio)
+
     return enhanced_path
 
 # Create the Gradio interface