Partially edited audio enables flexible changes and raises defense challenges
Partially Edited Audio: Generation, Defense, and Open Problems
Sound
Summary
Editing just parts of an audio recording to change what is said or how it sounds is becoming easier with new AI tools. The authors explain that whether such changes are harmful or fake depends on how the edited audio is used, not on the edits themselves. They review how these partial audio edits are done, how to detect or defend against them, and highlight tools and tests available. They finish by pointing out what still needs to be solved in this area.
What this means in practice
- •For audio software developers: Integrate comprehensive editing and detection techniques for partially edited audio into media production and verification tools.
- •For cybersecurity teams: Use defense strategies from the survey to improve detection and interpretation of partially edited audio in security monitoring systems.
A survey. It maps existing work.
Authors
Lin Zhang, You Zhang, Nicholas Evans, Xin Wang, David Harwath, Abhinav Dhall, Bowen Shi, Chengzhe Sun, Siwei Lyu, Sanjeev Khudanpur
Abstract
Recent advances in generative audio models facilitate increasingly flexible and selective editing of linguistic content and paralinguistic, extralinguistic, and other attributes. This paradigm, named partially edited audio, differs from full audio generation as only portions of the real recording are edited. This technique can serve both beneficial and harmful applications. We therefore present a concise survey from both generation and defense perspectives, focusing on audio, as a contribution to the ICASSP 2027 special session entitled "Partially Edited Media: Perspectives from Synthesis and Defense." We first define partially edited audio and argue that what counts as ``fake'' or harmful is an application-dependent judgment, not a fixed property of the edit. We then survey audio editing by editing target, approaches, and evaluation, and then defense methods by detection, localization, diarization, and interpretability, summarizing representative benchmarks and toolkits. Finally, we discuss future directions and open problems.