Date of Award

5-1-2022

Language

English

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

College/School/Department

Department of Electrical and Computer Engineering

Content Description

1 online resource (ix, 77 pages) : color illustrations.

Dissertation/Thesis Chair

Ming-Ching Chang

Committee Members

Daphney-Stavroula Zois, Hany Elgala

Keywords

AI-Synthesized, Bispectral Analysis, Self-Supervised Learning, Speech Synthesis, Spoken Language Generation, voice conversion, Speech synthesis, Natural language processing (Computer science), Speech processing systems, Computer sound processing, Deep learning (Machine learning), Artificial intelligence, Voice computing

Subject Categories

Computer Engineering

Abstract

From speech to images, and videos, advances in machine learning have led to dramatic improvements in the quality and realism of so-called AI-synthesized content. While there are many exciting and interesting applications, this type of content can also be used to create convincing and dangerous fakes. We seek to develop forensic techniques that can distinguish a real human voice from a synthesized voice. We observe that deep neural networks used to synthesize speech introduce specific and unusual artifacts not typically found in human speech. Although not necessarily audible, we develop various detection algorithms to measure these artifacts and be able to differentiate between human and synthesized speech.

Download

Share

COinS