Diffusion-Link: Diffusion Probabilistic Model for Bridging the Audio-Text Modality Gap

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Nam, KiHyun, Choi, Jongmin, Lee, Hyeongkeun, Heo, Jungwoo, Chung, Joon Son
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915549959684096
author Nam, KiHyun
Choi, Jongmin
Lee, Hyeongkeun
Heo, Jungwoo
Chung, Joon Son
author_facet Nam, KiHyun
Choi, Jongmin
Lee, Hyeongkeun
Heo, Jungwoo
Chung, Joon Son
contents Contrastive audio-language pretraining yields powerful joint representations, yet a persistent audio-text modality gap limits the benefits of coupling multimodal encoders with large language models (LLMs). We present Diffusion-Link, a diffusion-based modality-bridging module that generatively maps audio embeddings into the text-embedding distribution. The module is trained at the output embedding from the frozen multimodal encoder and implemented as a lightweight network with three residual MLP blocks. To assess the effect of Diffusion-Link on multimodal encoder-LLM coupling, we evaluate on Automatic Audio Captioning (AAC); to our knowledge, this is the first application of diffusion-based modality bridging to AAC. We report two results. (1) Modality-gap analysis: on similarity and geometric criteria, Diffusion-Link reduces the modality gap the most among prior diffusion-based methods and shows a collective migration of audio embeddings toward the text distribution. (2) Downstream AAC: attaching Diffusion-Link to the same multimodal LLM baseline achieves state-of-the-art on AudioCaps in both zero-shot and fully supervised captioning without external knowledge, with relative gains up to 52.5% and 7.5%, respectively. These findings show that closing the modality gap is pivotal for effective coupling between multimodal encoders and LLMs, and diffusion-based modality bridging offers a promising direction beyond knowledge-retrieval-centric designs. Code will be released upon acceptance https://github.com/DevKiHyun/Diffusion-Link
format Preprint
id arxiv_https___arxiv_org_abs_2510_11330
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Diffusion-Link: Diffusion Probabilistic Model for Bridging the Audio-Text Modality Gap
Nam, KiHyun
Choi, Jongmin
Lee, Hyeongkeun
Heo, Jungwoo
Chung, Joon Son
Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
Contrastive audio-language pretraining yields powerful joint representations, yet a persistent audio-text modality gap limits the benefits of coupling multimodal encoders with large language models (LLMs). We present Diffusion-Link, a diffusion-based modality-bridging module that generatively maps audio embeddings into the text-embedding distribution. The module is trained at the output embedding from the frozen multimodal encoder and implemented as a lightweight network with three residual MLP blocks. To assess the effect of Diffusion-Link on multimodal encoder-LLM coupling, we evaluate on Automatic Audio Captioning (AAC); to our knowledge, this is the first application of diffusion-based modality bridging to AAC. We report two results. (1) Modality-gap analysis: on similarity and geometric criteria, Diffusion-Link reduces the modality gap the most among prior diffusion-based methods and shows a collective migration of audio embeddings toward the text distribution. (2) Downstream AAC: attaching Diffusion-Link to the same multimodal LLM baseline achieves state-of-the-art on AudioCaps in both zero-shot and fully supervised captioning without external knowledge, with relative gains up to 52.5% and 7.5%, respectively. These findings show that closing the modality gap is pivotal for effective coupling between multimodal encoders and LLMs, and diffusion-based modality bridging offers a promising direction beyond knowledge-retrieval-centric designs. Code will be released upon acceptance https://github.com/DevKiHyun/Diffusion-Link
title Diffusion-Link: Diffusion Probabilistic Model for Bridging the Audio-Text Modality Gap
topic Sound
Artificial Intelligence
Computation and Language
Machine Learning
Audio and Speech Processing
url https://arxiv.org/abs/2510.11330