MEGA Hub

Partially Edited Audio: Generation, Defense, and Open Problems

Authors

Do you know Lin Zhang?You can claim authorship or link another user.Do you know You Zhang?You can claim authorship or link another user.Do you know Nicholas Evans?You can claim authorship or link another user.Do you know Xin Wang?You can claim authorship or link another user.Do you know David Harwath?You can claim authorship or link another user.Do you know Abhinav Dhall?You can claim authorship or link another user.Do you know Bowen Shi?You can claim authorship or link another user.Do you know Chengzhe Sun?You can claim authorship or link another user.Do you know Siwei Lyu?You can claim authorship or link another user.Do you know Sanjeev Khudanpur?You can claim authorship or link another user.

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.

Community

00

Publication notes

Author note
submitted to IEEE ICASSP 2027, 5 pages