arXiv subjects

Seong-Whan Lee

Publications and source records attributed to Seong-Whan Lee.

At least 73 records · Page 4Linked to original sources

arXiv2025

Spotlight-TTS: Spotlighting the Style via Voiced-Aware Style Extraction and Style Direction Adjustment for Expressive Text-to-Speech

Recent advances in expressive text-to-speech (TTS) have introduced diverse methods based on style embedding extracted from reference speech. However, synthesizing high-quality expressive speech remains challenging. We propose Spotlight-TTS, which exclusively emphasizes style via voiced-aware style extraction and style direction adjustment. Voiced-aware style extraction focuses on voiced regions highly related to style while maintaining continuity across different speech regions to improve expressiveness. We adjust the direction of the extracted style for optimal integration into the TTS model, which improves speech quality. Experimental results demonstrate that Spotlight-TTS achieves superior performance compared to baseline models in terms of expressiveness, overall speech quality, and style transfer capability. Our audio samples are publicly available.

cs.SD↗

arXiv2025

Diversify and Conquer: Open-set Disagreement for Robust Semi-supervised Learning with Outliers

Conventional semi-supervised learning (SSL) ideally assumes that labeled and unlabeled data share an identical class distribution, however in practice, this assumption is easily violated, as unlabeled data often includes unknown class data, i.e., outliers. The outliers are treated as noise, considerably degrading the performance of SSL models. To address this drawback, we propose a novel framework, Diversify and Conquer (DAC), to enhance SSL robustness in the context of open-set semi-supervised learning. In particular, we note that existing open-set SSL methods rely on prediction discrepancies between inliers and outliers from a single model trained on labeled data. This approach can be easily failed when the labeled data is insufficient, leading to performance degradation that is worse than naive SSL that do not account for outliers. In contrast, our approach exploits prediction disagreements among multiple models that are differently biased towards the unlabeled distribution. By leveraging the discrepancies arising from training on unlabeled data, our method enables robust outlier detection even when the labeled data is underspecified. Our key contribution is constructing a collection of differently biased models through a single training process. By encouraging divergent heads to be differently biased towards outliers while making consistent predictions for inliers, we exploit the disagreement among these heads as a measure to identify unknown concepts. Our code is available at https://github.com/heejokong/DivCon.

cs.CV↗

arXiv2025

MIRe: Enhancing Multimodal Queries Representation via Fusion-Free Modality Interaction for Multimodal Retrieval

Recent multimodal retrieval methods have endowed text-based retrievers with multimodal capabilities by utilizing pre-training strategies for visual-text alignment. They often directly fuse the two modalities for cross-reference during the alignment to understand multimodal queries. However, existing methods often overlook crucial visual information due to a text-dominant issue, which overly depends on text-driven signals. In this paper, we introduce MIRe, a retrieval framework that achieves modality interaction without fusing textual features during the alignment. Our method allows the textual query to attend to visual embeddings while not feeding text-driven signals back into the visual representations. Additionally, we construct a pre-training dataset for multimodal query retrieval by transforming concise question-answer pairs into extended passages. Our experiments demonstrate that our pre-training strategy significantly enhances the understanding of multimodal queries, resulting in strong performance across four multimodal retrieval benchmarks under zero-shot settings. Moreover, our ablation studies and analyses explicitly verify the effectiveness of our framework in mitigating the text-dominant issue. Our code is publicly available: https://github.com/yeongjoonJu/MIRe

cs.CV↗

arXiv2025

DiGIT: Multi-Dilated Gated Encoder and Central-Adjacent Region Integrated Decoder for Temporal Action Detection Transformer

In this paper, we examine a key limitation in query-based detectors for temporal action detection (TAD), which arises from their direct adaptation of originally designed architectures for object detection. Despite the effectiveness of the existing models, they struggle to fully address the unique challenges of TAD, such as the redundancy in multi-scale features and the limited ability to capture sufficient temporal context. To address these issues, we propose a multi-dilated gated encoder and central-adjacent region integrated decoder for temporal action detection transformer (DiGIT). Our approach replaces the existing encoder that consists of multi-scale deformable attention and feedforward network with our multi-dilated gated encoder. Our proposed encoder reduces the redundant information caused by multi-level features while maintaining the ability to capture fine-grained and long-range temporal information. Furthermore, we introduce a central-adjacent region integrated decoder that leverages a more comprehensive sampling strategy for deformable cross-attention to capture the essential information. Extensive experiments demonstrate that DiGIT achieves state-of-the-art performance on THUMOS14, ActivityNet v1.3, and HACS-Segment. Code is available at: https://github.com/Dotori-HJ/DiGIT

cs.CV↗

arXiv2025

Integrating Locality-Aware Attention with Transformers for General Geometry PDEs

Neural operators have emerged as promising frameworks for learning mappings governed by partial differential equations (PDEs), serving as data-driven alternatives to traditional numerical methods. While methods such as the Fourier neural operator (FNO) have demonstrated notable performance, their reliance on uniform grids restricts their applicability to complex geometries and irregular meshes. Recently, Transformer-based neural operators with linear attention mechanisms have shown potential in overcoming these limitations for large-scale PDE simulations. However, these approaches predominantly emphasize global feature aggregation, often overlooking fine-scale dynamics and localized PDE behaviors essential for accurate solutions. To address these challenges, we propose the Locality-Aware Attention Transformer (LA2Former), which leverages K-nearest neighbors for dynamic patchifying and integrates global-local attention for enhanced PDE modeling. By combining linear attention for efficient global context encoding with pairwise attention for capturing intricate local interactions, LA2Former achieves an optimal balance between computational efficiency and predictive accuracy. Extensive evaluations across six benchmark datasets demonstrate that LA2Former improves predictive accuracy by over 50% relative to existing linear attention methods, while also outperforming full pairwise attention under optimal conditions. This work underscores the critical importance of localized feature learning in advancing Transformer-based neural operators for solving PDEs on complex and irregular domains.

cs.LG↗

arXiv2025

GRAIL: Gradient-Based Adaptive Unlearning for Privacy and Copyright in LLMs

Large Language Models (LLMs) trained on extensive datasets often learn sensitive information, which raises significant social and legal concerns under principles such as the "Right to be forgotten." Retraining entire models from scratch to remove undesired information is both costly and impractical. Furthermore, existing single-domain unlearning methods fail to address multi-domain scenarios, where knowledge is interwoven across domains such as privacy and copyright, creating overlapping representations that lead to excessive knowledge removal or degraded performance. To tackle these issues, we propose GRAIL (GRadient-based AdaptIve unLearning), a novel multi-domain unlearning framework. GRAIL leverages gradient information from multiple domains to precisely distinguish the unlearning scope from the retention scope, and applies an adaptive parameter-wise localization strategy to selectively remove targeted knowledge while preserving critical parameters for each domain. Experimental results on unlearning benchmarks show that GRAIL achieves unlearning success on par with the existing approaches, while also demonstrating up to 17% stronger knowledge retention success compared to the previous state-of-art method. Our findings establish a new paradigm for effectively managing and regulating sensitive information in large-scale pre-trained language models.

cs.CL↗

arXiv2025

EmoSphere++: Emotion-Controllable Zero-Shot Text-to-Speech via Emotion-Adaptive Spherical Vector

Emotional text-to-speech (TTS) technology has achieved significant progress in recent years; however, challenges remain owing to the inherent complexity of emotions and limitations of the available emotional speech datasets and models. Previous studies typically relied on limited emotional speech datasets or required extensive manual annotations, restricting their ability to generalize across different speakers and emotional styles. In this paper, we present EmoSphere++, an emotion-controllable zero-shot TTS model that can control emotional style and intensity to resemble natural human speech. We introduce a novel emotion-adaptive spherical vector that models emotional style and intensity without human annotation. Moreover, we propose a multi-level style encoder that can ensure effective generalization for both seen and unseen speakers. We also introduce additional loss functions to enhance the emotion transfer performance for zero-shot scenarios. We employ a conditional flow matching-based decoder to achieve high-quality and expressive emotional TTS in a few sampling steps. Experimental results demonstrate the effectiveness of the proposed framework.

cs.SD↗

arXiv2025

CW-BASS: Confidence-Weighted Boundary-Aware Learning for Semi-Supervised Semantic Segmentation

Semi-supervised semantic segmentation (SSSS) aims to improve segmentation performance by utilizing large amounts of unlabeled data with limited labeled samples. Existing methods often suffer from coupling, where over-reliance on initial labeled data leads to suboptimal learning; confirmation bias, where incorrect predictions reinforce themselves repeatedly; and boundary blur caused by limited boundary-awareness and ambiguous edge cues. To address these issues, we propose CW-BASS, a novel framework for SSSS. In order to mitigate the impact of incorrect predictions, we assign confidence weights to pseudo-labels. Additionally, we leverage boundary-delineation techniques, which, despite being extensively explored in weakly-supervised semantic segmentation (WSSS), remain underutilized in SSSS. Specifically, our method: (1) reduces coupling via a confidence-weighted loss that adjusts pseudo-label influence based on their predicted confidence scores, (2) mitigates confirmation bias with a dynamic thresholding mechanism that learns to filter out pseudo-labels based on model performance, (3) tackles boundary blur using a boundary-aware module to refine segmentation near object edges, and (4) reduces label noise through a confidence decay strategy that progressively refines pseudo-labels during training. Extensive experiments on Pascal VOC 2012 and Cityscapes demonstrate that CW-BASS achieves state-of-the-art performance. Notably, CW-BASS achieves a 65.9% mIoU on Cityscapes under a challenging and underexplored 1/30 (3.3%) split (100 images), highlighting its effectiveness in limited-label settings. Our code is available at https://github.com/psychofict/CW-BASS.

cs.CV↗

arXiv2025

Personalized Continual EEG Decoding: Retaining and Transferring Knowledge

The significant inter-subject variability in electroen-cephalogram (EEG) signals often results in substantial changes to neural network weights as data distributions shift. This variability frequently causes catastrophic forgetting in continual EEG decoding tasks, where previously acquired knowledge is overwritten as new subjects are introduced. While retraining the entire dataset for each new subject can mitigate forgetting, this approach imposes significant computational costs, rendering it impractical for real-world applications. Therefore, an ideal brain-computer interface (BCI) model should incrementally learn new information without requiring complete retraining, thereby reducing computational overhead. Existing EEG decoding meth-ods typically rely on large, centralized source-domain datasets for pre-training to improve model generalization. However, in practical scenarios, data availability is often constrained by privacy concerns. Furthermore, these methods are susceptible to catastrophic forgetting in continual EEG decoding tasks, significantly limiting their utility in long-term learning scenarios. To address these issues, we propose the Personalized Continual EEG Decoding (PCED) framework for continual EEG decoding. The framework uses Euclidean Alignment for fast domain adap-tation, reducing inter-subject variability. To retain knowledge and prevent forgetting, it includes an exemplar replay mechanism that preserves key information from past tasks. A reservoir sampling-based memory management strategy optimizes exemplar storage to handle memory constraints in long-term learning. Experiments on the OpenBMI dataset with 54 subjects show that PCED balances knowledge retention and classification performance, providing an efficient solution for real-world BCI applications.

eess.SP↗

arXiv2025

DurFlex-EVC: Duration-Flexible Emotional Voice Conversion Leveraging Discrete Representations without Text Alignment

Emotional voice conversion (EVC) involves modifying various acoustic characteristics, such as pitch and spectral envelope, to match a desired emotional state while preserving the speaker's identity. Existing EVC methods often rely on text transcriptions or time-alignment information and struggle to handle varying speech durations effectively. In this paper, we propose DurFlex-EVC, a duration-flexible EVC framework that operates without the need for text or alignment information. We introduce a unit aligner that models contextual information by aligning speech with discrete units representing content, eliminating the need for text or speech-text alignment. Additionally, we design a style autoencoder that effectively disentangles content and emotional style, allowing precise manipulation of the emotional characteristics of the speech. We further enhance emotional expressiveness through a hierarchical stylize encoder that applies the target emotional style at multiple hierarchical levels, refining the stylization process to improve the naturalness and expressiveness of the converted speech. Experimental results from subjective and objective evaluations demonstrate that our approach outperforms baseline models, effectively handling duration variability and enhancing emotional expressiveness in the converted speech.

cs.SD↗

arXiv2025

Multi-Context Temporal Consistent Modeling for Referring Video Object Segmentation

Referring video object segmentation aims to segment objects within a video corresponding to a given text description. Existing transformer-based temporal modeling approaches face challenges related to query inconsistency and the limited consideration of context. Query inconsistency produces unstable masks of different objects in the middle of the video. The limited consideration of context leads to the segmentation of incorrect objects by failing to adequately account for the relationship between the given text and instances. To address these issues, we propose the Multi-context Temporal Consistency Module (MTCM), which consists of an Aligner and a Multi-Context Enhancer (MCE). The Aligner removes noise from queries and aligns them to achieve query consistency. The MCE predicts text-relevant queries by considering multi-context. We applied MTCM to four different models, increasing performance across all of them, particularly achieving 47.6 J&F on the MeViS. Code is available at https://github.com/Choi58/MTCM.

cs.CV↗

arXiv2025

SUGAR: Leveraging Contextual Confidence for Smarter Retrieval

Bearing in mind the limited parametric knowledge of Large Language Models (LLMs), retrieval-augmented generation (RAG) which supplies them with the relevant external knowledge has served as an approach to mitigate the issue of hallucinations to a certain extent. However, uniformly retrieving supporting context makes response generation source-inefficient, as triggering the retriever is not always necessary, or even inaccurate, when a model gets distracted by noisy retrieved content and produces an unhelpful answer. Motivated by these issues, we introduce Semantic Uncertainty Guided Adaptive Retrieval (SUGAR), where we leverage context-based entropy to actively decide whether to retrieve and to further determine between single-step and multi-step retrieval. Our empirical results show that selective retrieval guided by semantic uncertainty estimation improves the performance across diverse question answering tasks, as well as achieves a more efficient inference.

cs.CL↗

arXiv2025

JELLY: Joint Emotion Recognition and Context Reasoning with LLMs for Conversational Speech Synthesis

Recently, there has been a growing demand for conversational speech synthesis (CSS) that generates more natural speech by considering the conversational context. To address this, we introduce JELLY, a novel CSS framework that integrates emotion recognition and context reasoning for generating appropriate speech in conversation by fine-tuning a large language model (LLM) with multiple partial LoRA modules. We propose an Emotion-aware Q-former encoder, which enables the LLM to perceive emotions in speech. The encoder is trained to align speech emotions with text, utilizing datasets of emotional speech. The entire model is then fine-tuned with conversational speech data to infer emotional context for generating emotionally appropriate speech in conversation. Our experimental results demonstrate that JELLY excels in emotional context modeling, synthesizing speech that naturally aligns with conversation, while mitigating the scarcity of emotional conversational speech datasets.

cs.CL↗

arXiv2025

Towards Dynamic Neural Communication and Speech Neuroprosthesis Based on Viseme Decoding

Decoding text, speech, or images from human neural signals holds promising potential both as neuroprosthesis for patients and as innovative communication tools for general users. Although neural signals contain various information on speech intentions, movements, and phonetic details, generating informative outputs from them remains challenging, with mostly focusing on decoding short intentions or producing fragmented outputs. In this study, we developed a diffusion model-based framework to decode visual speech intentions from speech-related non-invasive brain signals, to facilitate face-to-face neural communication. We designed an experiment to consolidate various phonemes to train visemes of each phoneme, aiming to learn the representation of corresponding lip formations from neural signals. By decoding visemes from both isolated trials and continuous sentences, we successfully reconstructed coherent lip movements, effectively bridging the gap between brain signals and dynamic visual interfaces. The results highlight the potential of viseme decoding and talking face reconstruction from human neural signals, marking a significant step toward dynamic neural communication systems and speech neuroprosthesis for patients.

q-bio.NC↗

arXiv2025

FLowHigh: Towards Efficient and High-Quality Audio Super-Resolution with Single-Step Flow Matching

Audio super-resolution is challenging owing to its ill-posed nature. Recently, the application of diffusion models in audio super-resolution has shown promising results in alleviating this challenge. However, diffusion-based models have limitations, primarily the necessity for numerous sampling steps, which causes significantly increased latency when synthesizing high-quality audio samples. In this paper, we propose FLowHigh, a novel approach that integrates flow matching, a highly efficient generative model, into audio super-resolution. We also explore probability paths specially tailored for audio super-resolution, which effectively capture high-resolution audio distributions, thereby enhancing reconstruction quality. The proposed method generates high-fidelity, high-resolution audio through a single-step sampling process across various input sampling rates. The experimental results on the VCTK benchmark dataset demonstrate that FLowHigh achieves state-of-the-art performance in audio super-resolution, as evaluated by log-spectral distance and ViSQOL while maintaining computational efficiency with only a single-step sampling process.

eess.AS↗

arXiv2024

Towards Personalized Brain-Computer Interface Application Based on Endogenous EEG Paradigms

In this paper, we propose a conceptual framework for personalized brain-computer interface (BCI) applications, which can offer an enhanced user experience by customizing services to individual preferences and needs, based on endogenous electroencephalography (EEG) paradigms including motor imagery (MI), speech imagery (SI), and visual imagery. The framework includes two essential components: user identification and intention classification, which enable personalized services by identifying individual users and recognizing their intended actions through EEG signals. We validate the feasibility of our framework using a private EEG dataset collected from eight subjects, employing the ShallowConvNet architecture to decode EEG features. The experimental results demonstrate that user identification achieved an average classification accuracy of 0.995, while intention classification achieved 0.47 accuracy across all paradigms, with MI demonstrating the best performance. These findings indicate that EEG signals can effectively support personalized BCI applications, offering robust identification and reliable intention decoding, especially for MI and SI.

cs.HC↗

arXiv2024

Dynamic Neural Communication: Convergence of Computer Vision and Brain-Computer Interface

Interpreting human neural signals to decode static speech intentions such as text or images and dynamic speech intentions such as audio or video is showing great potential as an innovative communication tool. Human communication accompanies various features, such as articulatory movements, facial expressions, and internal speech, all of which are reflected in neural signals. However, most studies only generate short or fragmented outputs, while providing informative communication by leveraging various features from neural signals remains challenging. In this study, we introduce a dynamic neural communication method that leverages current computer vision and brain-computer interface technologies. Our approach captures the user's intentions from neural signals and decodes visemes in short time steps to produce dynamic visual outputs. The results demonstrate the potential to rapidly capture and reconstruct lip movements during natural speech attempts from human neural signals, enabling dynamic neural communication through the convergence of computer vision and brain--computer interface.

cs.AI↗

arXiv2024

EmoSphere-TTS: Emotional Style and Intensity Modeling via Spherical Emotion Vector for Controllable Emotional Text-to-Speech

Despite rapid advances in the field of emotional text-to-speech (TTS), recent studies primarily focus on mimicking the average style of a particular emotion. As a result, the ability to manipulate speech emotion remains constrained to several predefined labels, compromising the ability to reflect the nuanced variations of emotion. In this paper, we propose EmoSphere-TTS, which synthesizes expressive emotional speech by using a spherical emotion vector to control the emotional style and intensity of the synthetic speech. Without any human annotation, we use the arousal, valence, and dominance pseudo-labels to model the complex nature of emotion via a Cartesian-spherical transformation. Furthermore, we propose a dual conditional adversarial network to improve the quality of generated speech by reflecting the multi-aspect characteristics. The experimental results demonstrate the model ability to control emotional style and intensity with high-quality expressive speech.

cs.SD↗