The SunoCaps dataset aims to provide an innovative contribution to music data. Expert description of human-made musical pieces, from the widely used MusicCaps dataset, are used as prompts for generating complete songs for this dataset. This Automatic Music Generation is done with the state-of-the-art Suno generator of audio-based music. A subset of 64 pieces from MusicCaps is currently included, with a total of 256 generated entries. This total stems from generating four different variations for each human piece; two versions based on the original caption and two versions based on the original aspect description.

Background: Active Learning with AI‐tutoring in Higher Education tackles dropout rates.Objectives: To investigate teaching‐learning methodologies preferred by students. AHP is used to evaluate a ChatGPT‐based studented learning methodology which is compared to another active learning methodology and a traditional methodology. Study with Learning Analytics to evaluate alternatives, and help students elect the best strategies according to their preferences.Methods: Comparative study of three learning methodologies in a counterbalanced Single‐Group with 33 university students. It follows a pre‐test/post‐test approach using AHP and SAM. HRV and GSR used for the estimation of emotional states.Findings: Criteria related to in‐class experiences valued higher than test‐related criteria. Chat‐GPT integration was well regarded compared to well‐established methodologies. Student emotion self‐assessment correlated with physiological measures, validating used Learning Analytics.Conclusions: Proposed model AI‐Tutoring classroom integration functions effectively at increasing engagement and avoiding false information. AHP with the physiological measuring allows students to determine preferred learning methodologies, avoiding biases, and acknowledging minority groups.

The PaGER-Sync ADICVIDEO dataset is a multimodal, temporally synchronized repository of physiological and facial expression data recorded during controlled, immersive video game sessions designed to simulate realistic home gaming environments. It integrates biosignals from the Empatica E4 wristband —including Electrodermal Activity (EDA), Blood Volume Pulse (BVP), and Skin Temperature (TEMP) — with facial expression features extracted from video recordings using FaceReader software. Additionally, the dataset includes scores from pre-session psychometric questionnaires (Gaming Addiction Scale, Scale of Positive and Negative Experience, Emotion Regulation Questionnaire) and demographic gender data, providing psychological and individual difference context. A summary file detailing the two strongest emotions expressed by each participant with their respective percentages is included. A total of 25 participants played three commercial video games (Tetris, Sonic Racing, and Fall Guys) under controlled conditions, while their physiological responses were continuously recorded and their facial expressions captured on video for subsequent analysis. All data streams were precisely aligned using a common video-based timestamp, enabling frame-level synchronization across modalities, and the data were segmented by game. The dataset supports a wide range of research applications in affective computing, human-computer interaction, and behavioral analysis, and is particularly well-suited for the development and evaluation of multimodal affect detection models, as well as for exploring the interplay between psychological traits and real-time emotional responses.

Driver drowsiness is an important contributor to road accidents, but conventional multichannel electroencephalography (EEG) systems are difficult to deploy in everyday driving contexts. This study investigates whether band-power features obtained from a low-cost, single-channel commercial EEG device (NeuroSky MindWave Mobile 2, Fp1 position) can discriminate between the alert and drowsy labels provided in a public dataset. Frequency-band features were combined with physiologically motivated ratios, and several machine learning classifiers were evaluated using repeated nested stratified cross-validation. Because the dataset does not provide participant identifiers, the analysis estimates within-dataset discrimination and cannot establish generalization to unseen individuals. Random Forest achieved the highest mean area under the receiver operating characteristic curve (AUC=0.869), whereas the Soft-Voting Ensemble provided a compromise between accuracy (0.78), drowsiness recall (0.77), and AUC (0.863). SHAP analysis identified delta, highBeta, and highGamma as the most influential variables. A reduced model using these three features retained an AUC of 0.831, and the ensemble required approximately 7.52 ms per prediction on the evaluated computing platform. These findings support the computational feasibility of explainable drowsiness classification from consumer-grade single-channel EEG features, while subject-independent and embedded-hardware validation remain necessary.