Detect All-Type Deepfake Audio: Wavelet Prompt Tuning for Enhanced Auditory Perception DOI: XXXXXXX.XXXXXXX Conference: Proceedings of the 33rd ACM International Conference on Multimedia; October 27–31, 2025; Dublin, Ireland ISBN: 978-1-4503-XXXX-X/2018/06
Yuankun Xie, Ruibo Fu, Zhiyong Wang, Xiaopeng Wang, Songjun Cao, Long Ma, Haonnan Cheng, Long Ye · 2025
Read Detect All-Type Deepfake Audio: Wavelet Prompt Tuning for Enhanced Auditory Perception DOI: XXXXXXX.XXXXXXX Conference: Proceedings of the 33rd ACM International Conference on Multimedia; October 27–31, 2025; Dublin, Ireland ISBN: 978-1-4503-XXXX-X/2018/06 by Yuankun Xie, Ruibo Fu, Zhiyong Wang, Xiaopeng Wang, Songjun Cao, Long Ma, Haonnan Cheng, Long Ye (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06753.txt.
Description
Yuankun Xie and seven co-authors build a detector for fake audio of every kind. Their paper, from two thousand twenty-five, responds to fast gains in audio generation. Fake speech, sounds, singing, and music can now fool listeners. Old detectors work well on one kind of audio but fade on others. The team takes on the all-type task directly.
They first build a benchmark that mixes deepfakes across all four kinds. Then they propose a training scheme based on prompt tuning of a self-supervised front end. Small learned prompt tokens adapt the front end with far fewer trained values than full fine-tuning. The method also leans on how hearing itself picks out cues. The goal is one detector that holds up across kinds.
Catalog record
| CREATOR | Yuankun Xie, Ruibo Fu, Zhiyong Wang, Xiaopeng Wang, Songjun Cao, Long Ma, Haonnan Cheng, Long Ye |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 37 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-06753