arXiv subjects

Donghun Kang

Publications and source records attributed to Donghun Kang.

5 recordsLinked to original sources

arXiv2026

Phrase-Localized Language-Contrastive Guidance: Training-Free Localized Accent Control for Code-Switching Text-to-Speech

Current speech synthesis struggles with code-switching, which mixes a foreign language phrase into a primary language utterance, causing the phrase to be spoken with the primary language's accent rather than its native one. We propose Phrase-Localized Language-Contrastive Guidance (LCG), a training-free inference framework that restores a native accent to code-switched phrases in cross-lingual text-to-speech. LCG replaces the single language guidance applied across the whole utterance with a separate guidance for each region, so each part is guided by its own language. To choose where to apply this localized guidance, we propose a self-attention probing technique that finds the phrase boundaries without external alignments. Together, these components generate speech in which each region carries the accent of its own language, requiring no fine-tuning or auxiliary models. Across diverse language pairs, LCG robustly increases the nativeness of the code-switched phrase while suppressing accent leakage, and preserving overall speaker identity and naturalness.

cs.CL↗

arXiv2026

Verbal-R3: Verbal Reranker as the Missing Bridge between Retrieval and Reasoning

The conventional Retrieval-Augmented Generation (RAG) paradigm of injecting raw retrieved texts into the Large Language Model (LLM)'s context often results in suboptimal integration of retrieved information. This paper proposes to bridge retrieval results and the LLM's reasoning ability through Verbal Annotations, analytic narratives that explicitly articulate the logical connection between a search query and retrieved contexts. Our empirical investigation reveals the potential of Verbal Annotations to substantially enhance the LLM's ability to generate accurate, contextually-grounded responses. Motivated by this finding, we introduce Verbal-R3, a novel agentic RAG framework that consists of a Generator and a Verbal Reranker. The Generator performs iterative retrieval and reasoning, while the Verbal Reranker returns relevance scores and Verbal Annotations to guide the reasoning and answering process of the Generator. The inference process of Verbal-R3 is further refined through relevance-guided test-time scaling, which efficiently allocates test-time compute for effective trajectory expansion. Verbal-R3 achieves state-of-the-art performance on complex Question Answering benchmarks, validating the effectiveness of the proposed framework.

cs.CL↗

arXiv2026

Knowledge Integration Decay in Search-Augmented Reasoning of Large Language Models

Modern Large Language Models (LLMs) have demonstrated remarkable capabilities in complex tasks by employing search-augmented reasoning to incorporate external knowledge into long chains of thought. However, we identify a critical yet underexplored bottleneck in this paradigm, termed Knowledge Integration Decay (KID). Specifically, we observe that as the length of reasoning generated before search grows, models increasingly fail to integrate retrieved evidence into subsequent reasoning steps, limiting performance even when relevant information is available. To address this, we propose Self-Anchored Knowledge Encoding (SAKE), a training-free inference-time strategy designed to stabilize knowledge utilization. By anchoring retrieved knowledge at both the beginning and end of the reasoning process, SAKE prevents it from being overshadowed by prior context, thereby preserving its semantic integrity. Extensive experiments on multi-hop QA and complex reasoning benchmarks demonstrate that SAKE significantly mitigates KID and improves performance, offering a lightweight yet effective solution for knowledge integration in agentic LLMs.

cs.CL↗

arXiv2025

DoMIX: An Efficient Framework for Exploiting Domain Knowledge in Fine-Tuning

Domain-Adaptive Pre-training (DAP) has recently gained attention for its effectiveness in fine-tuning pre-trained models. Building on this, continual DAP has been explored to develop pre-trained models capable of incrementally incorporating different domain datasets. However, existing continual DAP methods face several limitations: (1) high computational cost and GPU memory usage during training; (2) sensitivity to incremental data order; and (3) providing a single, generalized model for all end tasks, which contradicts the essence of DAP. In this paper, we propose DoMIX, a novel approach that addresses these challenges by leveraging LoRA modules, a representative parameter-efficient fine-tuning (PEFT) method. Our approach enables efficient and parallel domain-adaptive pre-training that is robust to domain order and effectively utilizes accumulated knowledge to provide tailored pre-trained models for specific tasks. We also demonstrate that our method can be extended beyond the DAP setting to standard LLM fine-tuning scenarios. Code is available at https://github.com/dohoonkim-ai/DoMIX.

cs.CL↗

arXiv2023

Airfoil trailing-edge noise source identification using large-eddy simulation and wavelet transform

Airfoil noise is predicted and analyzed using wall-resolved large-eddy simulations and wavelet transforms for a NACA 0012 airfoil at a Mach number of 0.06 and a Reynolds number of 400,000 using a stair-strip forced transition and a natural transition. At a high angle of attack, vortex shedding and a laminar separation bubble (LSB) occur on the suction side. The LSB triggers the flow transition for both the forced and natural transition cases. The wavelet thresholding and denoising algorithm is used to decompose the pressure fields into the coherent or denoised pressure and the incoherent or background noise pressure. This denoising technique provides a clear picture of true noise generation and propagation. It also reveals the dominant noise source at specific frequencies when multiple noise sources are present. In another usage, the wavelet thresholding algorithm with down-sampling separates noise on the basis of flow structures. For example, the wavelet method separates noise between low-frequency vortex shedding noise and high-frequency LSB noise as well as trailing-edge noise. Finally, the wavelet transform is used to decompose the hydrodynamic and acoustic pressure components near the surface using the coherence between near-field pressure and far-field pressure. Overall, the wavelet-based decomposition is a valuable tool to study and reveal the mechanisms of airfoil noise generation.

physics.flu-dyn↗