Scholarlink

This work blends InceptionV3, MobileNetV2, and VGG16 models to present a deep learning ensemble framework for deepfake audio detection. Our approach uses data augmentation and converts two-second audio snippets from the Fake or Real (FoR) dataset into spectrogram pictures, therefore producing good binary classification ability between real and synthetic audio. On the test dataset, the suggested framework achieves an overall accuracy of 94.5%, showing improved performance in synthetic audio identification over single-model approaches

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deepfake Audio Detection Using Ensemble Learning

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • P. Kauser Ahmed,
  • Shaan Chandra,
  • Anishwar Chakraborty,
  • Uddipan Sarkar,
  • Tiasa Jana,
  • Sana Vaidya

摘要

This work blends InceptionV3, MobileNetV2, and VGG16 models to present a deep learning ensemble framework for deepfake audio detection. Our approach uses data augmentation and converts two-second audio snippets from the Fake or Real (FoR) dataset into spectrogram pictures, therefore producing good binary classification ability between real and synthetic audio. On the test dataset, the suggested framework achieves an overall accuracy of 94.5%, showing improved performance in synthetic audio identification over single-model approaches