Abstract Large-scale vision-language models demonstrate strong multimodal alignment and generalization across diverse tasks. Among them, CLIP stands out as one of the most successful approaches. In this work, we extend the application of CLIP to sound source localization, proposing a self-supervised method operates without explicit text input. We introduce a framework that maps audios into tokens compatible with CLIP’s text encoder, producing audio-driven embeddings. These embeddings are used to generate sounding region masks, from which visual features are extracted and aligned with the audio embeddings through a contrastive audio-visual correspondence objective. Our findings show that alignment knowledge of pre-trained multimodal foundation model enables our method to generate more complete and compact localization for sounding objects. We further propose an LLM-guided extension that distills object-aware audio-visual scene understanding into the model during training to enhance alignment. Extensive experiments across five diverse tasks demonstrate that our method, in all variants, outperforms state-of-the-art approaches and achieves strong generalization in zero-shot settings.

We present a unified model capable of simultaneously grounding both spoken language and non-speech sounds within a visual scene, addressing key limitations in current audio-visual grounding models. Existing approaches are typically limited to handling either speech or non-speech sounds independently, or at best, together but sequentially without mixing. This limitation prevents them from capturing the complexity of real-world audio sources that are often mixed. Our approach introduces a "mix-and-separate" framework with audio-visual alignment objectives that jointly learn correspondence and disentanglement using mixed audio. Through these objectives, our model learns to produce distinct embeddings for each audio type, enabling effective disentanglement and grounding across mixed audio sources. Additionally, we created a new dataset to evaluate simultaneous grounding of mixed audio sources, demonstrating that our model outperforms prior methods. Our approach also achieves comparable or better performance in standard segmentation and cross-modal retrieval tasks, highlighting the benefits of our mix-and-separate approach.

Following the success of Large Language Models (LLMs), expanding their boundaries to new modalities represents a significant paradigm shift in multimodal understanding. Human perception is inherently multimodal, relying not only on text but also on auditory and visual cues for a complete understanding of the world. In recognition of this fact, audio-visual LLMs have recently emerged. Despite promising developments, the lack of dedicated benchmarks poses challenges for understanding and evaluating models. In this work, we show that audio-visual LLMs struggle to discern subtle relationships between audio and visual signals, leading to hallucinations and highlighting the need for reliable benchmarks. To address this, we introduce AVHBench, the first comprehensive benchmark specifically designed to evaluate the perception and comprehension capabilities of audio-visual LLMs. Our benchmark includes tests for assessing hallucinations, as well as the cross-modal matching and reasoning abilities of these models. Our results reveal that most existing audio-visual LLMs struggle with hallucinations caused by cross-interactions between modalities, due to their limited capacity to perceive complex multimodal signals and their relationships. Additionally, we demonstrate that simple training with our AVHBench improves robustness of audio-visual LLMs against hallucinations. Dataset: https://github.com/kaist-ami/AVHBench

We present MGAudio, a novel flow-based framework for open-domain video-to-audio generation, which introduces model-guided dual-role alignment as a central design principle. Unlike prior approaches that rely on classifier-based or classifier-free guidance, MGAudio enables the generative model to guide itself through a dedicated training objective designed for video-conditioned audio generation. The framework integrates three main components: (1) a scalable flow-based Transformer denoiser, (2) a dual-role alignment mechanism where the audio-visual encoder serves both as a conditioning module and as a feature aligner to improve generation quality, and (3) a model-guided objective that enhances cross-modal coherence and audio realism. MGAudio achieves state-of-the-art performance on VGGSound, reducing FAD to 0.40, substantially surpassing the best classifier-free guidance baselines, and consistently outperforms existing methods across FD, IS, and alignment metrics. It also generalizes well to the challenging UnAV-100 benchmark. These results highlight model-guided dual-role alignment as a powerful and scalable paradigm for conditional video-to-audio generation. Code is available at: https://github.com/pantheon5100/mgaudio

Recent studies on learning-based sound source localization have primarily focused on localization performance. However, prior work and existing benchmarks often overlook a crucial aspect: cross-modal interaction, which is essential for interactive sound source localization. This interaction is vital for understanding semantically matched or mismatched audio-visual events, such as silent objects or true sound sources among multiple objects. In this work, we comprehensively examine the cross-modal interaction of existing methods, benchmarks, evaluation metrics, and cross-modal understanding tasks. We identify the overlooked points of previous studies and make several contributions to address them. First, we propose a learning framework that incorporates retrieval-based and hand-crafted augmentation techniques, enhancing cross-modal interaction through cross-modal alignment. Second, we introduce new evaluation metrics to accurately and rigorously assess localization methods, focusing on both localization performance and cross-modal interaction. Third, to thoroughly analyze interactive sound source localization, we present a new semi-synthetic benchmark with diverse categorical combinations. Finally, we evaluate both interactive sound source localization and auxiliary cross-modal retrieval tasks, benchmarking competing methods alongside our own. Our new benchmark and evaluation metrics reveal that previous methods struggle with interactive sound source localization tasks, largely due to their limited cross-modal interaction capabilities. Our method, which features enhanced cross-modal alignment, demonstrates superior sound source localization and cross-modal interaction performance. This work provides the most comprehensive analysis of sound source localization to date, with extensive validation of competing methods on both existing and new benchmarks using both new and standard evaluation metrics.

Transformers have become central to recent advances in audio classification. However, training an audio spectrogram transformer, e.g. AST, from scratch can be resource and time-intensive. Furthermore, the complexity of transformers heavily depends on the input audio spectrogram size. In this work, we aim to optimize AST training by linking to the resolution in the time-axis. We introduce multi-phase training of audio spectrogram transformers by connecting the seminal idea of coarse-to-fine with transformer models. To achieve this, we propose a set of methods for temporal compression. By employing one of these methods, the transformer model learns from lowerresolution (coarse) data in the initial phases, and then is fine-tuned with high-resolution data later in a curriculum learning strategy. Experimental results demonstrate that the proposed training mechanism for AST leads to improved (or on-par) performance with faster convergence, i.e. requiring fewer computational resources and less time. This approach is also generalizable to other AST-based methods regardless of their learning paradigms.

The objective of this study is to investigate the learning process of Visually Grounded Speech (VGS) models through joint learning that combines contrastive learning and masked image modeling. Typically, VGS models ahn to establish audio-visual alignment between images and then spoken captions within a contrastive learning framework. Building upon this seminal concept, in this work, we explore whether visual reconstruction with the help of cross-modality can enhance alignment, given that spoken captions describe visual appearances. To achieve this, we extend the contrastive learning-based VGS models by incorporating a masked autoencoder that utilizes cross-attention in the decoder. Through this cross-modal interaction in the decoder, spoken caption features guide the model to reconstruct the masked patches and capture correspondence between the two modalities. Our findings suggest that integrating cross-modal reconstruction within the contrastive learning framework enhances audio-visual feature alignment. Consequently, our proposed method gives comparable performance to existing models that utilize prior knowledge or other modalities, such as object region proposals or Contrastive Language-Image Pretraining (CLIP).

Large-scale pre-trained image-text models demonstrate remarkable versatility across diverse tasks, benefiting from their robust representational capabilities and effective multimodal alignment. We extend the application of these models, specifically CLIP, to the domain of sound source localization. Unlike conventional approaches, we employ the pre-trained CLIP model without explicit text input, relying solely on the audio-visual correspondence. To this end, we introduce a framework that translates audio signals into tokens compatible with CLIP’s text encoder, yielding audio-driven embeddings. By directly using these embeddings, our method generates audio-grounded masks for the provided audio, extracts audio-grounded image features from the highlighted regions, and aligns them with the audio-driven embeddings using the audio-visual correspondence objective. Our findings suggest that utilizing pre-trained image-text models enable our model to generate more complete and compact localization maps for the sounding objects. Extensive experiments show that our method outperforms state-of-the-art approaches by a significant margin.

Transformers have rapidly become the preferred choice for audio classification, surpassing methods based on CNNs. However, Audio Spectrogram Transformers (ASTs) exhibit quadratic scaling due to self-attention. The removal of this quadratic self-attention cost presents an appealing direction. Recently, state space models (SSMs), such as Mamba, have demonstrated potential in language and vision tasks in this regard. In this study, we explore whether reliance on self-attention is necessary for audio classification tasks. By introducing Audio Mamba (AuM), the first self-attention-free, purely SSM-based model for audio classification, we aim to address this question. We evaluate AuM on various audio datasets - comprising six different benchmarks - where it achieves comparable or better performance compared to well-established AST model.

How does audio describe the world around us? In this paper, we propose a method for generating an image of a scene from sound. Our method addresses the challenges of dealing with the large gaps that often exist between sight and sound. We design a model that works by scheduling the learning procedure of each model component to associate audio-visual modalities despite their information gaps. The key idea is to enrich the audio features with visual information by learning to align audio to visual latent space. We translate the input audio to visual features, then use a pre-trained generator to produce an image. To further improve the quality of our generated images, we use sound source localization to select the audio-visual pairs that have strong cross-modal correlations. We obtain substantially better results on the VEGAS and VGGSound datasets than prior approaches. We also show that we can control our model's predictions by applying simple manipulations to the input waveform, or to the latent space.

The goal of this work is to localize sound sources in visual scenes with a self-supervised approach. Contrastive learning in the context of sound source localization leverages the natural correspondence between audio and visual signals where the audio-visual pairs from the same source are assumed as positive, while randomly selected pairs are negatives. However, this approach brings in noisy correspondences; for example, positive audio and visual pair signals that may be unrelated to each other, or negative pairs that may contain semantically similar samples to the positive one. Our key contribution in this work is to show that using a less strict decision boundary in contrastive learning can alleviate the effect of noisy correspondences in sound source localization. We propose a simple yet effective approach by slightly modifying the contrastive loss with a negative margin. Extensive experimental results show that our approach gives on-par or better performance than the state-of-the-art methods. Furthermore, we demonstrate that the introduction of a negative margin to existing methods results in a consistent improvement in performance.

The objective of this work is to explore the learning of visually grounded speech models (VGS) from multilingual perspective. Bilingual VGS models are generally trained with an equal number of spoken captions from both languages. However, in reality, there can be an imbalance among the languages for the available spoken captions. Our key contribution in this work is to leverage the power of a high-resource language in a bilingual visually grounded speech model to improve the performance of a low-resource language. We introduce two methods to distill the knowledge of high-resource language into low-resource languages: (1) incorporating a strong pre-trained high-resource language encoder and (2) using semantically similar spoken captions. Our experiments show that combining these two approaches effectively enables the low-resource language to surpass the performances of monolingual and bilingual counterparts for cross-modal retrieval tasks.

Humans can easily perceive the direction of sound sources in a visual scene, termed sound source localization. Recent studies on learning-based sound source localization have mainly explored the problem from a localization perspective. However, prior arts and existing benchmarks do not account for a more important aspect of the problem, cross-modal semantic understanding, which is essential for genuine sound source localization. Cross-modal semantic understanding is important in understanding semantically mismatched audio-visual events, e.g., silent objects, or off-screen sounds. To account for this, we propose a cross-modal alignment task as a joint task with sound source localization to better learn the interaction between audio and visual modalities. Thereby, we achieve high localization performance with strong cross-modal semantic understanding. Our method outperforms the state-of-the-art approaches in both sound source localization and cross-modal retrieval. Our work suggests that jointly tackling both tasks is necessary to conquer genuine sound source localization.

To understand our surrounding world, our brain is continuously inundated with multisensory information and their complex interactions coming from the outside world at any given moment. While processing this information might seem effortless for human brains, it is challenging to build a machine that can perform similar tasks since complex interactions cannot be dealt with a single type of integration but require more sophisticated approaches. In this paper, we propose a new simple method to address the multisensory integration in video understanding. Unlike previous works where a single fusion type is used, we design a multi-head model with individual event-specific layers to deal with different audio-visual relationships, enabling different ways of audio-visual fusion. Experimental results show that our event-specific layers can discover unique properties of the audio-visual relationships in the videos, e.g., semantically matched moments, and rhythmic events. Moreover, although our network is trained with single labels, our multi-head design can inherently output additional semantically meaningful multi-labels for a video. As an application, we demonstrate that our proposed method can expose the extent of event-characteristics of popular benchmark datasets.

The objective of this work is to localize the sound sources in visual scenes. Existing audio-visual works employ contrastive learning by assigning corresponding audio-visual pairs from the same source as positives while randomly mismatched pairs as negatives. However, these negative pairs may contain semantically matched audio-visual information. Thus, these semantically correlated pairs, "hard positives", are mistakenly grouped as negatives. Our key contribution is showing that hard positives can give similar response maps to the corresponding pairs. Our approach incorporates these hard positives by adding their response maps into a contrastive learning objective directly. We demonstrate the effectiveness of our approach on VGG-SS and SoundNet-Flickr test sets, showing favorable performance to the state-of-the-art methods.

In this paper, we tackle sound localization as a natural outcome of the audio-visual video classification problem. Differently from the existing sound localization approaches, we do not use any explicit sub-modules or training mechanisms but use simple cross-modal attention on top of the representations learned by a classification loss. Our key contribution is to show that a simple audio-visual classification model has the ability to localize sound sources accurately and to give on par performance with state-of-the-art methods by proving that indeed "less is more". Furthermore, we propose potential applications that can be built based on our model. First, we introduce informative moment selection to enhance the localization task learning in the existing approaches compare to mid-frame usage. Then, we introduce a pseudo bounding box generation procedure that can significantly boost the performance of the existing methods in semi-supervised settings or be used for large-scale automatic annotation with minimal effort from any video dataset.

Visual events are usually accompanied by sounds in our daily lives. However, can the machines learn to correlate the visual scene and sound, as well as localize the sound source only by observing them like humans? To investigate its empirical learnability, in this work we first present a novel unsupervised algorithm to address the problem of localizing sound sources in visual scenes. In order to achieve this goal, a two-stream network structure which handles each modality with attention mechanism is developed for sound source localization. The network naturally reveals the localized response in the scene without human annotation. In addition, a new sound source dataset is developed for performance evaluation. Nevertheless, our empirical evaluation shows that the unsupervised method generates false conclusions in some cases. Thereby, we show that this false conclusion cannot be fixed without human prior knowledge due to the well-known correlation and causality mismatch misconception. To fix this issue, we extend our network to the supervised and semi-supervised network settings via a simple modification due to the general architecture of our two-stream network. We show that the false conclusions can be effectively corrected even with a small amount of supervision, i.e., semi-supervised setup. Furthermore, we present the versatility of the learned audio and visual embeddings on the cross-modal content alignment and we extend this proposed algorithm to a new application, sound saliency based automatic camera view panning in 360 degree videos.

Visual events are usually accompanied by sounds in our daily lives. We pose the question: Can the machine learn the correspondence between visual scene and the sound, and localize the sound source only by observing sound and visual scene pairs like human? In this paper, we propose a novel unsupervised algorithm to address the problem of localizing the sound source in visual scenes. A two-stream network structure which handles each modality, with attention mechanism is developed for sound source localization. Moreover, although our network is formulated within the unsupervised learning framework, it can be extended to a unified architecture with a simple modification for the supervised and semi-supervised learning settings as well. Meanwhile, a new sound source dataset is developed for performance evaluation. Our empirical evaluation shows that the unsupervised method eventually go through false conclusion in some cases. We also show that even with a few supervision, i.e., semi-supervised setup, false conclusion is able to be corrected effectively.