This report presents our systems submitted to the audio-only and audio-visual tracks of the DCASE2025 Task 3 Challenge: Stereo Sound Event Localization and Detection (SELD) in Regular Video Content. SELD is a complex task that combines temporal event classification with spatial localization, requiring reasoning across spatial, temporal, and semantic dimensions. The last is arguably the most challenging to model. Traditional SELD architectures rely on multichannel input, which limits their ability to leverage large-scale pre-training due to data constraints. To address this, we enhance standard SELD architectures with semantic information by integrating pre-trained, contrastive language-aligned models: CLAP for audio and OWL-ViT for visual inputs. These embeddings are incorporated into a modified Conformer module tailored for multimodal fusion, which we refer to as the Cross-Modal Conformer. Additionally, we incorporate autocorrelation-based acoustic features to improve distance estimation. We pre-train our models on curated synthetic audio and audio-visual datasets and apply a left-right channel swapping augmentation to further increase the training data. Both our audio-only and audio-visual systems substantially outperform the challenge baselines on the development set, demonstrating the effectiveness of our strategy. Performance is further improved through model ensembling and a visual post-processing step based on human keypoints. Future work will investigate the contribution of each modality and explore architectural variants to further enhance results.
Pre-training methods have achieved significant performance improvements in sound event localization and detection (SELD) tasks, but existing Transformer-based models suffer from high computational complexity. In this work, we propose a stereo sound event localization and detection system based on pre-trained PSELDnet and bidirectional Mamba sequence modeling. We replace the Conformer module with a BiMamba module and introduce asymmetric convolutions to more effectively model the spatiotemporal relationships between time and frequency dimensions. Experimental results demonstrate that the proposed method achieves significantly better performance than the baseline and the original PSELDnet with Conformer decoder architecture on the DCASE2025 Task 3 development dataset, while also reducing computational complexity. These findings highlight the effectiveness of the BiMamba architecture in addressing the challenges of the SELD task.
Sound event localization and detection (SELD) is a task for the classification of sound events and the identification of direction of arrival (DoA) utilizing multichannel acoustic signals. For effective classification and localization, a channel-spectro-temporal transformer (CST-former) was suggested. CST-former employs multidimensional attention mechanisms across the spatial, spectral, and temporal domains to enlarge the model's capacity to learn the domain information essential for event detection and DoA estimation over time. In this work, we present an enhanced version of CST-former with multiscale unfolded local embedding (MSULE) developed to capture and aggregate domain information over multiple time-frequency scales. Also, we propose finetuning and post-processing techniques beneficial for conducting the SELD task over limited training datasets. In-depth ablation studies of the proposed architecture and detailed analysis on the proposed modules are carried out to validate the efficacy of multidimensional attentions on the SELD task. Empirical validation through experimentation on STARSS22 and STARSS23 datasets demonstrates the remarkable performance of CST-former and post-processing techniques without using external data.
Sound event localization and detection (SELD) involves predicting active sound event classes over time while estimating their positions. The localization subtask in SELD is usually treated as a direction of arrival estimation problem, ignoring source distance. Only recently, SELD was extended to 3D by incorporating distance estimation, enabling the prediction of sound event positions in 3D space (3D SELD). However, existing methods lack input features designed for distance estimation. We argue that reverberation encodes valuable information for this task. This paper introduces two novel feature formats for 3D SELD based on reverberation: one using direct-to-reverberant ratio (DRR) and another leveraging signal autocorrelation to provide the model with insights into early reflections. Pre-training on synthetic data improves relative distance error (RDE) and overall SELD score, with autocorrelation-based features reducing RDE by over 3 percentage points on the STARSS23 dataset. The code to extract the features is available at github.com/dberghi/SELD-distance-features.
In traditional sound event localization and detection (SELD) tasks, the focus is typically on sound event detection (SED) and direction-of-arrival (DOA) estimation, but they fall short of providing full spatial information about the sound source. The 3D SELD task addresses this limitation by integrating source distance estimation (SDE), allowing for complete spatial localization. We propose three approaches to tackle this challenge: a novel method with independent training and joint prediction, which firstly treats DOA and distance estimation as separate tasks and then combines them to solve 3D SELD; a dual-branch representation with source Cartesian coordinate used for simultaneous DOA and distance estimation; and a three-branch structure that jointly models SED, DOA, and SDE within a unified framework. Our proposed method ranked first in the DCASE 2024 Challenge Task 3, demonstrating the effectiveness of joint modeling for addressing the 3D SELD task. The relevant code for this paper will be open-sourced in the future.
Sound event localization and detection with source distance estimation (3D SELD) involves not only identifying the sound category and its direction-of-arrival (DOA) but also predicting the source's distance, aiming to provide full information about the sound position. This paper proposes a multi-stage video attention network (MVANet) for audio-visual (AV) 3D SELD. Multi-stage audio features are used to adaptively capture the spatial information of sound sources in videos. We propose a novel output representation that combines the DOA with distance of sound sources by calculating the real Cartesian coordinates to address the newly introduced source distance estimation (SDE) task in the Detection and Classification of Acoustic Scenes and Events (DCASE) 2024 Challenge. We also employ a variety of effective data augmentation and pre-training methods. Experimental results on the STARSS23 dataset have proven the effectiveness of our proposed MVANet. By integrating the aforementioned techniques, our system outperforms the top-ranked method we used in the AV 3D SELD task of the DCASE 2024 Challenge without model ensemble. The code will be made publicly available in the future.
This paper investigates the feasibility of class-incremental learning (CIL) for Sound Event Localization and Detection (SELD) tasks. The method features an incremental learner that can learn new sound classes independently while preserving knowledge of old classes. The continual learning is achieved through a mean square error-based distillation loss to minimize output discrepancies between subsequent learners. The experiments are conducted on the TAU-NIGENS Spatial Sound Events 2021 dataset, which includes 12 different sound classes and demonstrate the efficacy of proposed method. We begin by learning 8 classes and introduce the 4 new classes at next stage. After the incremental phase, the system is evaluated on the full set of learned classes. Results show that, for this realistic dataset, our proposed method successfully maintains baseline performance across all metrics.
Sound event localization and detection (SELD) has seen substantial advancements through learning-based methods. These systems, typically trained from scratch on specific datasets, have shown considerable generalization capabilities. Recently, deep neural networks trained on large-scale datasets have achieved remarkable success in the sound event classification (SEC) field, prompting an open question of whether these advancements can be extended to develop general-purpose SELD models. In this paper, leveraging the power of pre-trained SEC models, we propose pre-trained SELD networks (PSELDNets) on large-scale synthetic datasets. These synthetic datasets, generated by convolving sound events with simulated spatial room impulse responses (SRIRs), contain 1,167 hours of audio clips with an ontology of 170 sound classes. These PSELDNets are transferred to downstream SELD tasks. When we adapt PSELDNets to specific scenarios, particularly in low-resource data cases, we introduce a data-efficient fine-tuning method, AdapterBit. PSELDNets are evaluated on a synthetic-test-set using collected SRIRs from TAU Spatial Room Impulse Response Database (TAU-SRIR DB) and achieve satisfactory performance. We also conduct our experiments to validate the transferability of PSELDNets to three publicly available datasets and our own collected audio recordings. Results demonstrate that PSELDNets surpass state-of-the-art systems across all publicly available datasets. Given the need for direction-of-arrival estimation, SELD generally relies on sufficient multi-channel audio clips. However, incorporating the AdapterBit, PSELDNets show more efficient adaptability to various tasks using minimal multi-channel or even just monophonic audio clips, outperforming the traditional fine-tuning approaches.
This paper describes sound event localization and detection (SELD) for spatial audio recordings captured by firstorder ambisonics (FOA) microphones. In this task, one may train a deep neural network (DNN) using FOA data annotated with the classes and directions of arrival (DOAs) of sound events. However, the performance of this approach is severely bounded by the amount of annotated data. To overcome this limitation, we propose a novel method of pretraining the feature extraction part of the DNN in a self-supervised manner. We use spatial audio-visual recordings abundantly available as virtual reality contents. Assuming that sound objects are concurrently observed by the FOA microphones and the omni-directional camera, we jointly train audio and visual encoders with contrastive learning such that the audio and visual embeddings of the same recording and DOA are made close. A key feature of our method is that the DOA-wise audio embeddings are jointly extracted from the raw audio data, while the DOA-wise visual embeddings are separately extracted from the local visual crops centered on the corresponding DOA. This encourages the latent features of the audio encoder to represent both the classes and DOAs of sound events. The experiment using the DCASE2022 Task 3 dataset of 20 hours shows non-annotated audio-visual recordings of 100 hours reduced the error score of SELD from 36.4 pts to 34.9 pts.
This report describes our systems submitted for the DCASE2024 Task 3 challenge: Audio and Audiovisual Sound Event Localization and Detection with Source Distance Estimation (Track B). Our main model is based on the audio-visual (AV) Conformer, which processes video and audio embeddings extracted with ResNet50 and with an audio encoder pre-trained on SELD, respectively. This model outperformed the audio-visual baseline of the development set of the STARSS23 dataset by a wide margin, halving its DOAE and improving the F1 by more than 3x. Our second system performs a temporal ensemble from the outputs of the AV-Conformer. We then extended the model with features for distance estimation, such as direct and reverberant signal components extracted from the omnidirectional audio channel, and depth maps extracted from the video frames. While the new system improved the RDE of our previous model by about 3 percentage points, it achieved a lower F1 score. This may be caused by sound classes that rarely appear in the training set and that the more complex system does not detect, as analysis can determine. To overcome this problem, our fourth and final system consists of an ensemble strategy combining the predictions of the other three. Many opportunities to refine the system and training strategy can be tested in future ablation experiments, and likely achieve incremental performance gains for this audio-visual task.
Sound event localization and detection (SELD) is critical for various real-world applications, including smart monitoring and Internet of Things (IoT) systems. Although deep neural networks (DNNs) represent the state-of-the-art approach for SELD, their significant computational complexity and model sizes present challenges for deployment on resource-constrained edge devices, especially under real-time conditions. Despite the growing need for real-time SELD, research in this area remains limited. In this paper, we investigate the unique challenges of deploying SELD systems for real-world, real-time applications by performing extensive experiments on a commercially available Raspberry Pi 3 edge device. Our findings reveal two critical, often overlooked considerations: the high computational cost of feature extraction and the performance degradation associated with low-latency, real-time inference. This paper provides valuable insights and considerations for future work toward developing more efficient and robust real-time SELD systems
Humans can picture a sound scene given an imprecise natural language description. For example, it is easy to imagine an acoustic environment given a phrase like "the lion roar came from right behind me!". For a machine to have the same degree of comprehension, the machine must know what a lion is (semantic attribute), what the concept of "behind" is (spatial attribute) and how these pieces of linguistic information align with the semantic and spatial attributes of the sound (what a roar sounds like when its coming from behind). State-of-the-art audio foundation models which learn to map between audio scenes and natural textual descriptions, are trained on non-spatial audio and text pairs, and hence lack spatial awareness. In contrast, sound event localization and detection models are limited to recognizing sounds from a fixed number of classes, and they localize the source to absolute position (e.g., 0.2m) rather than a position described using natural language (e.g., "next to me"). To address these gaps, we present ELSA a spatially aware-audio and text embedding model trained using multimodal contrastive learning. ELSA supports non-spatial audio, spatial audio, and open vocabulary text captions describing both the spatial and semantic components of sound. To train ELSA: (a) we spatially augment the audio and captions of three open-source audio datasets totaling 4,738 hours of audio, and (b) we design an encoder to capture the semantics of non-spatial audio, and the semantics and spatial attributes of spatial audio using contrastive learning. ELSA is competitive with state-of-the-art for both semantic retrieval and 3D source localization. In particular, ELSA achieves +2.8% mean audio-to-text and text-to-audio R@1 above the baseline, and outperforms by -11.6{\deg} mean-absolute-error in 3D source localization over the baseline.
Sound event localization and detection (SELD) systems using audio recordings from a microphone array rely on spatial cues for determining the location of sound events. As a consequence, the localization performance of such systems is to a large extent determined by the quality of the audio features that are used as inputs to the system. We propose a new feature, based on neural generalized cross-correlations with phase-transform (NGCC-PHAT), that learns audio representations suitable for localization. Using permutation invariant training for the time-difference of arrival (TDOA) estimation problem enables NGCC-PHAT to learn TDOA features for multiple overlapping sound events. These features can be used as a drop-in replacement for GCC-PHAT inputs to a SELD-network. We test our method on the STARSS23 dataset and demonstrate improved localization performance compared to using standard GCC-PHAT or SALSA-Lite input features.
In the Sound Event Localization and Detection (SELD) task, Transformer-based models have demonstrated impressive capabilities. However, the quadratic complexity of the Transformer's self-attention mechanism results in computational inefficiencies. In this paper, we propose a network architecture for SELD called SELD-Mamba, which utilizes Mamba, a selective state-space model. We adopt the Event-Independent Network V2 (EINV2) as the foundational framework and replace its Conformer blocks with bidirectional Mamba blocks to capture a broader range of contextual information while maintaining computational efficiency. Additionally, we implement a two-stage training method, with the first stage focusing on Sound Event Detection (SED) and Direction of Arrival (DoA) estimation losses, and the second stage reintroducing the Source Distance Estimation (SDE) loss. Our experimental results on the 2024 DCASE Challenge Task3 dataset demonstrate the effectiveness of the selective state-space model in SELD and highlight the benefits of the two-stage training approach in enhancing SELD performance.
This technical report details our systems submitted for Task 3 of the DCASE 2024 Challenge: Audio and Audiovisual Sound Event Localization and Detection (SELD) with Source Distance Estimation (SDE). We address only the audio-only SELD with SDE (SELDDE) task in this report. We propose to improve the existing ResNet-Conformer architectures with Squeeze-and-Excitation blocks in order to introduce additional forms of channel- and spatial-wise attention. In order to improve SELD performance, we also utilize the Spatial Cue-Augmented Log-Spectrogram (SALSA) features over the commonly used log-mel spectra features for polyphonic SELD. We complement the existing Sony-TAu Realistic Spatial Soundscapes 2023 (STARSS23) dataset with the audio channel swapping technique and synthesize additional data using the SpatialScaper generator. We also perform distance scaling in order to prevent large distance errors from contributing more towards the loss function. Finally, we evaluate our approach on the evaluation subset of the STARSS23 dataset.
Sound event localization and detection (SELD) aims to determine the appearance of sound classes, together with their Direction of Arrival (DOA). However, current SELD systems can only predict the activities of specific classes, for example, 13 classes in DCASE challenges. In this paper, we propose text-queried target sound event localization (SEL), a new paradigm that allows the user to input the text to describe the sound event, and the SEL model can predict the location of the related sound event. The proposed task presents a more user-friendly way for human-computer interaction. We provide a benchmark study for the proposed task and perform experiments on datasets created by simulated room impulse response (RIR) and real RIR to validate the effectiveness of the proposed methods. We hope that our benchmark will inspire the interest and additional research for text-queried sound source localization.
This study presents an audio-visual information fusion approach to sound event localization and detection (SELD) in low-resource scenarios. We aim at utilizing audio and video modality information through cross-modal learning and multi-modal fusion. First, we propose a cross-modal teacher-student learning (TSL) framework to transfer information from an audio-only teacher model, trained on a rich collection of audio data with multiple data augmentation techniques, to an audio-visual student model trained with only a limited set of multi-modal data. Next, we propose a two-stage audio-visual fusion strategy, consisting of an early feature fusion and a late video-guided decision fusion to exploit synergies between audio and video modalities. Finally, we introduce an innovative video pixel swapping (VPS) technique to extend an audio channel swapping (ACS) method to an audio-visual joint augmentation. Evaluation results on the Detection and Classification of Acoustic Scenes and Events (DCASE) 2023 Challenge data set demonstrate significant improvements in SELD performances. Furthermore, our submission to the SELD task of the DCASE 2023 Challenge ranks first place by effectively integrating the proposed techniques into a model ensemble.
Sound Event Localization and Detection (SELD) involves detecting and localizing sound events using multichannel sound recordings. Previously proposed Event-Independent Network V2 (EINV2) has achieved outstanding performance on SELD. However, it still faces challenges in effectively extracting features across spectral, spatial, and temporal domains. This paper proposes a three-stage network structure named Multi-scale Feature Fusion (MFF) module to fully extract multi-scale features across spectral, spatial, and temporal domains. The MFF module utilizes parallel subnetworks architecture to generate multi-scale spectral and spatial features. The TF-Convolution Module is employed to provide multi-scale temporal features. We incorporated MFF into EINV2 and term the proposed method as MFF-EINV2. Experimental results in 2022 and 2023 DCASE challenge task3 datasets show the effectiveness of our MFF-EINV2, which achieves state-of-the-art (SOTA) performance compared to published methods.
We aim to perform sound event localization and detection (SELD) using wearable equipment for a moving human, such as a pedestrian. Conventional SELD tasks have dealt only with microphone arrays located in static positions. However, self-motion with three rotational and three translational degrees of freedom (6DoF) shall be considered for wearable microphone arrays. A system trained only with a dataset using microphone arrays in a fixed position would be unable to adapt to the fast relative motion of sound events associated with self-motion, resulting in the degradation of SELD performance. To address this, we designed 6DoF SELD Dataset for wearable systems, the first SELD dataset considering the self-motion of microphones. Furthermore, we proposed a multi-modal SELD system that jointly utilizes audio and motion tracking sensor signals. These sensor signals are expected to help the system find useful acoustic cues for SELD on the basis of the current self-motion state. Experimental results on our dataset show that the proposed method effectively improves SELD performance with a mechanism to extract acoustic features conditioned by sensor signals.
The primary goal of the L3DAS23 Signal Processing Grand Challenge at ICASSP 2023 is to promote and support collaborative research on machine learning for 3D audio signal processing, with a specific emphasis on 3D speech enhancement and 3D Sound Event Localization and Detection in Extended Reality applications. As part of our latest competition, we provide a brand-new dataset, which maintains the same general characteristics of the L3DAS21 and L3DAS22 datasets, but with first-order Ambisonics recordings from multiple reverberant simulated environments. Moreover, we start exploring an audio-visual scenario by providing images of these environments, as perceived by the different microphone positions and orientations. We also propose updated baseline models for both tasks that can now support audio-image couples as input and a supporting API to replicate our results. Finally, we present the results of the participants. Further details about the challenge are available at https://www.l3das.com/icassp2023.