arXiv subjects

Taeyoung Jeong

Publications and source records attributed to Taeyoung Jeong.

3 recordsLinked to original sources

arXiv2026

Cross-lingual Retrieval-Augmented Classification for Dysarthria Severity Assessment

Automatic dysarthria severity assessment is limited by the scarcity of labeled pathological speech data. To address this, we propose Cross-lingual Retrieval-Augmented Classification (CRAC), which leverages speech from a different language via an align-retrieve-fuse pipeline. Supervised contrastive learning first shapes a severity-focused embedding space, then a vector database is built from the opposite-language corpus. During both training and inference, the classifier retrieves top-k references from the aligned space and fuses them with the input via cross-attention. Evaluated on Korean post-stroke and Italian ALS dysarthria datasets under a speaker-independent three-class protocol, CRAC achieves balanced accuracies of 87.3% on Korean and 86.7% on Italian, improving over monolingual baselines by 8.4 and 20.0 percentage points, respectively.

cs.SD↗

arXiv2026

CAF-Score: Calibrating CLAP with LALMs for Reference-free Audio Captioning Evaluation

While Large Audio-Language Models (LALMs) have advanced audio captioning, robust evaluation remains difficult. Reference-based metrics are expensive and often fail to assess acoustic fidelity, while Contrastive Language-Audio Pretraining (CLAP)-based approaches frequently overlook syntactic errors and fine-grained details. We propose CAF-Score, a reference-free metric that calibrates CLAP's coarse-grained semantic alignment with the fine-grained comprehension and syntactic awareness of LALMs. By combining contrastive audio-text embeddings with LALM reasoning, CAF-Score effectively detects syntactic inconsistencies and subtle hallucinations. Experiments on the BRACE benchmark demonstrate that our approach achieves the highest correlation with human judgments, even outperforming reference-based baselines in challenging scenarios. These results highlight the efficacy of CAF-Score for reference-free audio captioning evaluation. Code and results are available at https://github.com/inseong00/CAF-Score.

cs.SD↗

arXiv2026

VacHopPy: A Python package for vacancy hopping analysis based on molecular dynamics simulations

Multiscale modeling, which integrates material properties from ab initio calculations into continuum-scale simulations, is a promising strategy for optimizing semiconductor devices. However, a key challenge remains: while ab initio methods provide diffusion parameters specific to individual migration paths, continuum equations require a single effective set of parameters that captures the overall diffusion behavior. To address this issue, we present VacHopPy, an open-source Python package for vacancy hopping analysis based on molecular dynamics (MD). VacHopPy extracts an effective set of hopping parameters, including hopping distance, hopping barrier, number of effective paths, correlation factor, and attempt frequency, by statistically integrating energetic, kinetic, and geometric contributions across all paths. It also includes tools for tracking vacancy trajectories and for detecting phase transitions during MD simulations. The applicability of VacHopPy is demonstrated in three representative materials: face-centered cubic Al, rutile TiO2, and monoclinic HfO2. The extracted effective parameters reproduce temperature-dependent diffusion behavior and are in good agreement with previous experimental data. Provided in a simplified form, these parameters are well suited for continuum-scale models and remain valid over a wide temperature range spanning several hundred kelvins. Furthermore, VacHopPy inherently accounts for anisotropy in thermal vibrations, a factor often overlooked, making it suitable for simulating diffusion in complex crystals. Overall, VacHopPy establishes a robust bridge between atomic- and continuum-scale models, enabling more reliable multiscale simulation

cond-mat.mtrl-sci↗