Deepfake Detection for a Trustworthy Speech Communication

• Project/Area Number: 25K21245
• Research Category: Grant-in-Aid for Early-Career Scientists
• Allocation Type: Multi-year Fund
• Review Section: Basic Section 61020:Human interface and interaction-related
• Research Institution: Japan Advanced Institute of Science and Technology
• Principal Investigator: MAWALIM CandyOlivia 北陸先端科学技術大学院大学, 先端科学技術研究科, 助教 (10963720)
• Project Period (FY): 2025-04-01 – 2028-03-31
• Project Status: Granted (Fiscal Year 2025)
• Budget Amount: ¥4,810,000 (Direct Cost: ¥3,700,000、Indirect Cost: ¥1,110,000); Fiscal Year 2027: ¥1,560,000 (Direct Cost: ¥1,200,000、Indirect Cost: ¥360,000); Fiscal Year 2026: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000); Fiscal Year 2025: ¥1,820,000 (Direct Cost: ¥1,400,000、Indirect Cost: ¥420,000)
• Keywords: deepfake / spoof attacks / speaker verification / multilingual

Outline of Research at the Start

Advancements in deep learning have enabled the creation of highly realistic synthetic audio (deepfakes), posing threats to voice privacy and security. This research aims to address the limitations in existing research in deepfake detection by analyzing the physiological and acoustic characteristics of speech production mechanism that is unique from deepfakes. Robust deepfake detection methods will be developed that capable of handling diverse linguistic data, providing clear explanations for detection outcomes, and adapting to the evolving deepfake attacks.