We propose NTNU-SMIL’s system for SemEval-2026 Task 3 Track B Subtask 1 Dimensional Stance Analysis (DimStance). Our approach models target-conditioned valence–arousal regression using sentence-pair encoding, dual regression heads, and a logistic-loss regression formulation. For English and Chinese, we further leverage same-language transfer from Track A and apply lightweight out-of-fold calibration with multi-seed ensembling to reduce cross-lingual scale mismatch. Post-hoc analysis shows that same-language transfer and logistic-loss regression are the main drivers of performance gains, while arousal variance collapse remains a challenge in low-resource settings such as Swahili.

The Formosa Speech Recognition Challenge 2025 (FSR-2025) focuses on Taiwanese Hakka, a low-resource language with limited data diversity and channel coverage. To address this challenge, we propose a channel-aware, data-centric framework that leverages multilingual foundation models to mitigate mismatches between field recordings and training data. Our method integrates unsupervised anomaly detection and channel-conditioned augmentation to enhance data representativeness before ASR fine-tuning, aiming to explore the potential for improving robustness in low-resource Hakka speech recognition.