Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Koh, Junyoung, Kim, Soo Yong, Choi, Gyu Hyeong, Choi, Yongwon
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2509.20891
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866909805851967488
author Koh, Junyoung
Kim, Soo Yong
Choi, Gyu Hyeong
Choi, Yongwon
author_facet Koh, Junyoung
Kim, Soo Yong
Choi, Gyu Hyeong
Choi, Yongwon
contents We present AIBA (Attention-In-Band Alignment), a lightweight, training-free pipeline to quantify where text-to-audio diffusion models attend on the time-frequency (T-F) plane. AIBA (i) hooks cross-attention at inference to record attention probabilities without modifying weights; (ii) projects them to fixed-size mel grids that are directly comparable to audio energy; and (iii) scores agreement with instrument-band ground truth via interpretable metrics (T-F IoU/AP, frequency-profile correlation, and a pointing game). On Slakh2100 with an AudioLDM2 backbone, AIBA reveals consistent instrument-dependent trends (e.g., bass favoring low bands) and achieves high precision with moderate recall.
format Preprint
id arxiv_https___arxiv_org_abs_2509_20891
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle AIBA: Attention-based Instrument Band Alignment for Text-to-Audio Diffusion
Koh, Junyoung
Kim, Soo Yong
Choi, Gyu Hyeong
Choi, Yongwon
Sound
We present AIBA (Attention-In-Band Alignment), a lightweight, training-free pipeline to quantify where text-to-audio diffusion models attend on the time-frequency (T-F) plane. AIBA (i) hooks cross-attention at inference to record attention probabilities without modifying weights; (ii) projects them to fixed-size mel grids that are directly comparable to audio energy; and (iii) scores agreement with instrument-band ground truth via interpretable metrics (T-F IoU/AP, frequency-profile correlation, and a pointing game). On Slakh2100 with an AudioLDM2 backbone, AIBA reveals consistent instrument-dependent trends (e.g., bass favoring low bands) and achieves high precision with moderate recall.
title AIBA: Attention-based Instrument Band Alignment for Text-to-Audio Diffusion
topic Sound
url https://arxiv.org/abs/2509.20891