Update app.py

app.py

@@ -4,9 +4,7 @@
 
 import gradio as gr
 import numpy as np
-import matplotlib.pyplot as plt
 import librosa
-import librosa.display
 import os
 import cv2
 
@@ -24,29 +22,21 @@ def generate_frequency_visualization(audio_path, fps, num_bars):
         # Perform Short-Time Fourier Transform (STFT)
         hop_length = int(sr / fps)  # Hop length to match the desired fps
         S = np.abs(librosa.stft(y, n_fft=2048, hop_length=hop_length))
-        frequencies = librosa.fft_frequencies(sr=sr)
+        S = S[:num_bars, :]  # Limit the frequency bands to match the number of bars
 
-        # Create frequency bins for the bars
-        bins = np.linspace(0, len(frequencies), num_bars + 1, dtype=int)
-        bar_heights = []
-
-        # Aggregate power for each bar
-        for i in range(len(S[0])):
-            frame = S[:, i]
-            bar_frame = [np.mean(frame[bins[j]:bins[j+1]]) for j in range(num_bars)]
-            bar_heights.append(bar_frame)
+        # Normalize frequency power
+        S = S / np.max(S)
 
         # Create a directory to save the frames
         os.makedirs('frames', exist_ok=True)
 
         # Generate and save each frame
-        for i, heights in enumerate(bar_heights):
+        for i in range(S.shape[1]):
             # Create black background
             img = np.zeros((720, 1280, 3), dtype=np.uint8)
 
-            # Normalize heights to fit the frame
-            heights = np.array(heights)
-            heights = (heights / np.max(heights) * 600).astype(int)
+            # Get the bar heights for the current frame
+            heights = (S[:, i] * 600).astype(int)
 
             # Calculate bar positions
             bar_width = 80
@@ -61,14 +51,14 @@ def generate_frequency_visualization(audio_path, fps, num_bars):
             frame_path = f'frames/frame_{i:04d}.png'
             cv2.imwrite(frame_path, img)
 
-        print(f"Generated {len(bar_heights)} frames for visualization.")
+        print(f"Generated {S.shape[1]} frames for visualization.")
         return 'frames', duration
     except Exception as e:
         print(f"Error generating frequency visualization: {e}")
         return None, None
 
 # Function to create a video from the generated frames
-def create_video_from_frames(frames_directory, audio_path, fps, duration):
+def create_video_from_frames(frames_directory, audio_path, fps):
     try:
         # Get the list of frame files
         frame_files = [os.path.join(frames_directory, f) for f in os.listdir(frames_directory) if f.endswith('.png')]
@@ -109,7 +99,7 @@ def process_audio(audio):
     num_bars = 12
     frames_directory, duration = generate_frequency_visualization(audio_path, fps, num_bars)
     if frames_directory:
-        video_path = create_video_from_frames(frames_directory, audio_path, fps, duration)
+        video_path = create_video_from_frames(frames_directory, audio_path, fps)
         return video_path
     else:
         return None
@@ -135,6 +125,6 @@ if __name__ == "__main__":
 # The following dependencies are required to run this app:
 # - librosa
 # - numpy
-# - matplotlib
 # - opencv-python
+# - matplotlib
 # - ffmpeg (installed separately)