Audio Deepfake Detection Using Xception Model - Dalam bentuk pengganti sidang - Artikel Jurnal
DZAMIR AKMAL
Informasi Umum
Kode
25.04.482
Klasifikasi
000 - General Works
Jenis
Karya Ilmiah - Skripsi (S1) - Reference
Subjek
Cyber Security
Dilihat
125 kali
Informasi Lainnya
Abstraksi
The proliferation of audio deepfakes, generated using advanced technologies such as WaveNet and Generative Adversarial Networks (GANs), poses significant threats to digital security, including identity theft, misinformation, and fraud. To address these challenges, this study proposes an end-to end framework for audio deepfake detection that leverages Mel Spectrograms as input features and the Xception model as the backbone architecture. The methodology includes optimized preprocessing techniques, such as normalization and resizing, and robust data augmentation strategies to enhance feature quality and model generalization. The framework was evaluated using the Automatic Speaker Verification (ASV) spoof 2021 dataset, achieving a high test accuracy of 95.86% with balanced precision, recall, and F1-scores for ‘real‘ and ‘fake‘ classifications. Comparative analysis demonstrated that the Xception model outperformed ResNet50 and MobileNetV2 in both accuracy and generalization. While the results highlight the robust
- CII3E3 - KEAMANAN SIBER
- CS3243 - KECERDASAN MESIN DAN ARTIFISIAL
- CCH4D4 - TUGAS AKHIR
Koleksi & Sirkulasi
Tersedia 1 dari total 1 Koleksi
Anda harus log in untuk mengakses flippingbook
Pengarang
| Nama | DZAMIR AKMAL |
| Jenis | Perorangan |
| Penyunting | Vera Suryani |
| Penerjemah |
Penerbit
| Nama | Universitas Telkom, S1 Informatika |
| Kota | Bandung |
| Tahun | 2025 |
Sirkulasi
| Harga sewa | IDR 0,00 |
| Denda harian | IDR 0,00 |
| Jenis | Non-Sirkulasi |
