MUMBAI, India, Oct. 31 -- Intellectual Property India has published a patent application (202411030361 A) filed by Indian Institute Of Technology, Jodhpur, Rajasthan, on April 15, 2024, for 'system and method for detecting deepfakes in multi-lingual audio files.'
Inventor(s) include Mayank Vatsa; Richa Singh; and Rishabh Ranjan.
The application for the patent was published on Oct. 31, under issue no. 44/2025.
According to the abstract released by the Intellectual Property India: "A system (100) for deepfake detection is disclosed. The system (100) includes an input unit (102) and processing circuitry (108). The input unit (102)receives audio signals, while the processing circuitry (108) extracts one or more semantic tokens and one or more acoustic tokens from the audio signals. The processing circuitry (108) identifies spoofed semantic tokens from the extracted semantic tokens, indicating deceptive linguistic content, and spoofed acoustic tokens from the acoustic tokens, indicating manipulated sound characteristics. The spoofed semantic and spoofed acoustic tokens are combined to generate a unified set of spoofed tokens, to which a value is assigned based on the degree of genuineness of the audio signals. Subsequently, the assigned value is compared with a predefined threshold to classify the audio signals as either genuine or spoofed."
Disclaimer: Curated by HT Syndication.
