Update app.py

app.py

@@ -8,7 +8,8 @@ import gradio as gr
 import yt_dlp
 import subprocess
 from pydub import AudioSegment
-from pydub.effects import reverb
+from scipy.signal import convolve
+
 from audio_separator.separator import Separator
 from lib.infer import infer_audio
 import edge_tts
@@ -88,18 +89,28 @@ async def text_to_speech_edge(text, language_code):
 
 
 
-# Function to add reverb effect using pydub
+# Function to apply a basic reverb effect using convolution
 def add_simple_reverb(input_audio):
-    # Load the uploaded audio file
+    # Load the uploaded audio file using pydub
     sound = AudioSegment.from_file(input_audio)
-    # Apply reverb effect
-    reverbed_sound = reverb(sound)
+    
+    # Convert AudioSegment to numpy array
+    samples = np.array(sound.get_array_of_samples())
+    
+    # Define a simple impulse response for reverb (can be customized)
+    impulse_response = np.concatenate([np.zeros(5000), np.array([0.5**i for i in range(1000)])])
+    
+    # Apply convolution (reverb effect)
+    reverbed_samples = convolve(samples, impulse_response, mode='full')
+    reverbed_samples = reverbed_samples[:len(samples)]  # trim to original length
+    
+    # Convert numpy array back to AudioSegment
+    reverbed_sound = sound._spawn(reverbed_samples.astype(np.int16).tobytes())
     
     # Export the reverbed sound to a new file-like object (in-memory)
     output_path = "vocals_with_reverb.wav"
     reverbed_sound.export(output_path, format='wav')
     
-    # Return the output path for Gradio to handle
     return output_path