Audio Deepfake Detection using Spectrograms

Surbhit Pratik , Hriday Eluru

audio, deepfake, Keras, TensorFlow, Conformer, ASVspoof, spectrogram

High-quality fake videos and audio generated by AI algorithms (the deep fakes) have started challenging the status of videos and audio as definitive evidence of events. These sounds and clippings, if used as evidence, could even become a threat to someone’s private life or even national security. In this project, we have developed a system that can accurately determine whether a given piece of evidence is real or artificially mutated using a Keras model with ConformerEncoder architecture as its backbone. The ASVspoof 2019 dataset has been used to perceive the accuracy of the model built.

Published Paper   E-Certificate

"Audio Deepfake Detection using Spectrograms", IJSDR - International Journal of Scientific Development and Research (www.IJSDR.org), ISSN:2455-2631, Vol.8, Issue 8, page no.488 - 492, August-2023, Available :https://ijsdr.org/papers/IJSDR2308078.pdf

Volume 8 Issue 8, August-2023

Pages : 488 - 492

Paper Reg. ID: IJSDR_206222

Published Paper Id: IJSDR2308078

Downloads: 000347239

Research Area: Computer Science & Technology 

Country: Ranchi, Jharkhand, India

Published Paper PDF: https://ijsdr.org/papers/IJSDR2308078

Published Paper URL: https://ijsdr.org/viewpaperforall?paper=IJSDR2308078

DOI: http://doi.one/10.1729/Journal.35856

ISSN: 2455-2631 | IMPACT FACTOR: 9.15 Calculated By Google Scholar | ESTD YEAR: 2016

An International Scholarly Open Access Journal, Peer-Reviewed, Refereed Journal Impact Factor 9.15 Calculate by Google Scholar and Semantic Scholar | AI-Powered Research Tool, Multidisciplinary, Monthly, Multilanguage Journal Indexing in All Major Database & Metadata, Citation Generator

Publisher: IJSDR(IJ Publication) Janvi Wave

This work is licensed under a Creative Commons Attribution 4.0 International License and The Open Definition