Fusion Segment Transformer for AI Generated Music Detection

Fusion Segment Transformer: Bi-directional attention guided fusion network for AI Generated Music Detection

Authors: Yumin Kim and Seonghyeon Go

Submitted to ICASSP 2026. Detects AI-generated music by modeling full audio segments with content-structure fusion.

⚠️ Note: On Zero GPU environment, processing may take ~30 seconds per audio file.

⚠️ This demo uses a research checkpoint trained on data up to Jan 2025, which is outdated. Results on music from newer AI generators may be wrong. For an up-to-date, production AI music detection service, please visit mippia.com.