Update app.py

app.py

@@ -1,6 +1,7 @@
 import gradio as gr
 import numpy as np
 from moviepy.editor import AudioFileClip, ImageClip, CompositeVideoClip
+import matplotlib.pyplot as plt
 import tempfile
 import os
 
@@ -8,29 +9,36 @@ def create_waveform_video(image, audio_path):
     # Save uploaded image temporarily
     image_path = tempfile.mktemp(suffix=".png")
     video_path = tempfile.mktemp(suffix=".mp4")
+    waveform_path = tempfile.mktemp(suffix=".png")
     
     image.save(image_path)
     
     # Load audio
     audio_clip = AudioFileClip(audio_path)
     
+    # Generate waveform as an image
+    waveform = audio_clip.to_soundarray(fps=22050).mean(axis=1)  # Convert stereo to mono
+    waveform = np.interp(waveform, (waveform.min(), waveform.max()), (-1, 1))  # Normalize
+
+    # Plot waveform using matplotlib
+    plt.figure(figsize=(10, 2))
+    plt.plot(waveform, color="blue")
+    plt.axis("off")
+    plt.savefig(waveform_path, bbox_inches="tight", pad_inches=0)
+    plt.close()
+
     # Load image and create a video clip of the same duration as the audio
-    img_clip = ImageClip(image_path).set_duration(audio_clip.duration)
-    img_clip = img_clip.resize(height=720)  # Resize image to a suitable video height
-    
-    # Generate waveform effect as a video
-    waveform = audio_clip.to_soundarray(fps=22050)  # Convert audio to waveform array
-    waveform = waveform.mean(axis=1)  # Convert stereo to mono by averaging channels
-    
-    # Normalize waveform for overlay
-    waveform = np.interp(waveform, (waveform.min(), waveform.max()), (-1, 1))
-    
-    # Combine image and audio into one video
-    final_clip = img_clip.set_audio(audio_clip)
+    img_clip = ImageClip(image_path).set_duration(audio_clip.duration).resize(height=720)
+    waveform_clip = ImageClip(waveform_path).set_duration(audio_clip.duration).resize(width=img_clip.w)
+    waveform_clip = waveform_clip.set_position(("center", img_clip.h - waveform_clip.h - 20))
+
+    # Combine image, waveform overlay, and audio into one video
+    final_clip = CompositeVideoClip([img_clip, waveform_clip.set_opacity(0.7)]).set_audio(audio_clip)
     final_clip.write_videofile(video_path, codec="libx264", fps=24)
     
     # Cleanup temporary files
     os.remove(image_path)
+    os.remove(waveform_path)
     
     return video_path