Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
October 3, 2025Open Access

Evaluating Fake Music Detection Performance Under Audio Augmentations

View Full Paper
Ask AI
Bookmark
Share

Authors

TSTomasz SrokaTWTomasz WężowiczDSDominik Sidorczuk

Discussion

Loading...

Member takes

Overview

This work explores how audio augmentations impact classification accuracy in detecting fake music, suggesting a decline in model performance.

Key Points

  • The classification accuracy of deepfake detection models decreases significantly with audio augmentations.
  • Testing revealed that even light audio augmentations notably impacted the model's performance on detecting synthetic music.
  • A dataset was constructed of both real and synthetic music to better evaluate model generalization under different audio conditions.
  • The study emphasizes the challenges posed by generative audio models in distinguishing human-composed music from generated compositions.

Cite This Study

Sroka et al. (2025) studied this question.

synapsesocial.com/papers/68e040eda99c246f578b331fhttps://doi.org/10.48550/arxiv.2507.10447
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Detecting music deepfakes is easy but actually hard2024 · 3 citations
  2. 2Targeted Augmented Data for Audio Deepfake Detection2024
  3. 3Detecting Musical Deepfakes2025
  4. 4FakeSound: Deepfake General Audio Detection2024
  5. 5Exploring Self-supervised Embeddings and Synthetic Data Augmentation for Robust Audio Deepfake Detection2024 · 12 citations