This paper describes our system for SemEval-2026 Task 3, which focuses on predicting continuous valence and arousal scores. The task poses significant challenges due to variations in data scale and pragmatic ambiguities across languages. To address these disparities, we propose an Adaptive Dual-Track Framework that dynamically selects modeling strategies based on task characteristics. For semantically stable tasks, we apply a robust single baseline optimized with layer-wise learning rate decay (LLRD) to ensure stability. For high-ambiguity scenarios such as the Environmental Protection domain, we adopt a heterogeneous ensemble strategy to mitigate prediction variance. Experimental results demonstrate that our system consistently outperforms the initial standard baseline across all subtasks. Furthermore, our lightweight approach exhibits remarkable parameter efficiency, achieving highly competitive performance against newly introduced large language model (LLM) baselines. Additionally, ablation studies reveal that under regression settings, conventional regularization techniques, cross-lingual data transfer, and homogeneous ensemble learning can lead to negative transfer, confirming the necessity of strategically diverging approaches tailored to linguistic characteristics.

Mental health concerns have garnered increasing attention, highlighting the importance of timely and accurate identification of individual stress states as a critical research domain. This study employs the multimodal StressID dataset to evaluate the contributions of three modalities—physiological signals, video, and audio—in stress recognition tasks. A set of machine learning models, including Random Forests (RF), Support Vector Machines (SVM), Multi-Layer Perceptrons (MLP), and K-Nearest Neighbors (KNN), were trained and tested with optimized parameters for each modality. In addition, the effectiveness of different multimodal fusion strategies was systematically examined. The unimodal experiments revealed that the physiological modality achieved the highest performance in the binary stress classification task (F1-score = 0.751), whereas the audio modality outperformed the others in the three-class classification task (F1-score = 0.625). In the multimodal setting, feature-level fusion yielded stable improvements in the binary classification task, while decision-level fusion achieved superior performance in the three-class classification task (F1-score = 0.65). These findings demonstrate that multimodal integration can substantially enhance the accuracy of stress recognition. Future research directions include incorporating temporal modeling and addressing data imbalance to further improve the robustness and applicability of stress recognition systems.

This article presents an overview of the existing acoustical corpuses suitable for broadcast news automatic transcription task in the Slovak language. The TUKE-BNews-SK database created in our department was built to support the application development for automatic broadcast news processing and spontaneous speech recognition of the Slovak language. The audio corpus is composed of 479 Slovak TV broadcast news shows from public Slovak television called STV1 or “Jednotka” containing 265 hours of material and 186 hours of clean transcribed speech (4 hours subset extracted for testing purposes). The recordings were manually transcribed using Transcriber tool modified for Slovak annotators and automatic Slovak spell checking. The corpus design, acquisition, annotation scheme and pronunciation transcription is described together with corpus statistics and tools used. Finally the evaluation procedure using automatic speech recognition is presented on the broadcast news and parliamentary speeches test sets.