Dive into the world of speech security, anti-spoofing, and deepfake detection in this episode of Speech Pitch featuring Tomi Kinnunen, Professor of Computer Science at the University of Eastern Finland and a leading researcher in voice biometrics and audio spoofing countermeasures.
In this episode, Tomi breaks down the origins of the landmark ASVspoof initiative, shares insights from the SPEECHFAKES project, and discusses where speech synthesis research is headed next. We also touch on key academic advice for PhD students, mentoring strategies, and the big question: will better deep neural networks solve everything in speech processing?
Hosts: Priyansi Pal, Nikhil Raghav
Script: Priyansi Pal, Nikhil Raghav
Post production editors: Vishwas Shetty, Belu Ticona, Snigdha Banik, Dani Kazzy, Spyretta Leivaditi, Pascal Hecker
00:00 Intro
00:11 Introducing Tomi Kinnunen
01:17 Origins of ASVspoof
06:03 Perspective on the security arms race
09:35 Speech synthesis and deepfake detection research collaboration
14:24 Is anti-spoofing going to be more important than synthesis? Which modality is easier to detect?
17:17 The SPEECHFAKES project
24:31 Potential future research directions
25:42 Personal fun facts
28:27 Advice on mentoring and PhD research
31:13 Other interesting research areas
32:03 Question from the previous guests: Will better deep neural networks solve everything speech related? How can we convince students to write papers on why they did their research?
36:33 Question for the next guest: How to find the topics that were not yet addressed
37:21 Farewell and outro
Fler avsnitt från "Speech Pitch"
Missa inte ett avsnitt av “Speech Pitch” och prenumerera på det i GetPodcast-appen.
