chuban

batch_waveform.py

@@ -0,0 +1,112 @@
+"""
+批量处理音频文件生成波形图的脚本
+"""
+
+import os
+import matplotlib.pyplot as plt
+import librosa
+import librosa.display
+import argparse
+import glob
+
+def generate_waveform(audio_path, output_path):
+    """
+    生成音频波形图并保存
+    
+    Args:
+        audio_path: 音频文件路径
+        output_path: 输出图片保存路径
+    
+    Returns:
+        None
+    """
+    # 创建输出目录
+    os.makedirs(os.path.dirname(output_path), exist_ok=True)
+    
+    # 加载音频
+    y, sr = librosa.load(audio_path, sr=None)
+    
+    # 创建图形
+    plt.figure(figsize=(12, 4))
+    
+    # 绘制波形
+    librosa.display.waveshow(y, sr=sr)
+    plt.title('Waveform')
+    plt.xlabel('Time (s)')
+    plt.ylabel('Amplitude')
+    
+    # 保存图片
+    plt.tight_layout()
+    plt.savefig(output_path)
+    plt.close()
+
+def process_audio_directory(audio_dir, output_dir, limit=None, file_ext='wav'):
+    """
+    批量处理音频文件生成波形图
+    
+    Args:
+        audio_dir: 音频文件目录
+        output_dir: 输出图片保存目录
+        limit: 处理文件数量限制
+        file_ext: 音频文件扩展名
+    
+    Returns:
+        int: 处理的文件数量
+    """
+    # 创建输出目录
+    os.makedirs(output_dir, exist_ok=True)
+    
+    # 获取所有音频文件
+    audio_files = glob.glob(os.path.join(audio_dir, f"**/*.{file_ext}"), recursive=True)
+    
+    if limit and len(audio_files) > limit:
+        audio_files = audio_files[:limit]
+    
+    processed_count = 0
+    for audio_file in audio_files:
+        # 生成输出文件路径
+        rel_path = os.path.relpath(audio_file, audio_dir)
+        output_path = os.path.join(output_dir, f"{os.path.splitext(rel_path)[0]}.png")
+        
+        # 确保输出目录存在
+        os.makedirs(os.path.dirname(output_path), exist_ok=True)
+        
+        try:
+            # 生成波形图
+            generate_waveform(audio_file, output_path)
+            processed_count += 1
+            print(f"Processed {processed_count}/{len(audio_files)}: {rel_path}")
+        except Exception as e:
+            print(f"Error processing {audio_file}: {e}")
+    
+    return processed_count
+
+def main():
+    # 解析命令行参数
+    parser = argparse.ArgumentParser(description='Batch Audio Waveform Generator')
+    parser.add_argument('--audio_dir', type=str, default='./RAVDESS/Actor_01',
+                       help='Directory containing audio files')
+    parser.add_argument('--output_dir', type=str, default='./output/waveforms',
+                       help='Directory to save waveform images')
+    parser.add_argument('--limit', type=int, default=None,
+                       help='Limit the number of files to process')
+    parser.add_argument('--file_ext', type=str, default='wav',
+                       help='Audio file extension to process')
+    
+    args = parser.parse_args()
+    
+    # 检查目录是否存在
+    if not os.path.exists(args.audio_dir):
+        print(f"Error: Audio directory not found: {args.audio_dir}")
+        return
+    
+    # 批量处理音频文件
+    start_time = __import__('time').time()
+    count = process_audio_directory(args.audio_dir, args.output_dir, args.limit, args.file_ext)
+    end_time = __import__('time').time()
+    
+    print(f"\nProcessed {count} files in {end_time - start_time:.2f} seconds")
+    print(f"Waveform images saved to: {args.output_dir}")
+
+if __name__ == "__main__":
+    main()