875 papers
This collection focuses on Eess — As, a specialized area within electrical engineering and systems science dedicated to the analysis and design of complex systems. Here, researchers explore how individual components interact to create larger, functional networks, ranging from power grids to communication protocols. These studies often rely on rigorous mathematical models to predict system behavior and ensure stability in an increasingly interconnected world.
Gist.Science processes every new preprint in this category directly from arXiv, ensuring you have access to the very latest findings before formal publication. For each paper, we provide both a plain-language overview to make the core concepts accessible to a broad audience and a detailed technical summary for those seeking deeper methodological insights. Below are the latest papers in this field, carefully curated to keep you informed on the cutting edge of systems science.
This paper introduces WildElder, a Mandarin elderly speech dataset collected from online videos with fine-grained manual annotations, designed to address the limitations of existing controlled-environment datasets and serve as a robust benchmark for automatic speech recognition and speaker profiling research.
This study demonstrates that while a fixed Convolutional Neural Network achieves high sensitivity in detecting abnormal heart sounds using the PhysioNet 2016 dataset, employing PCEN or multi-resolution spectrogram front-ends yields slightly higher accuracy than standard logmel spectrograms, with Grad-CAM visualizations confirming the model's focus on physiologically relevant low-frequency heart sound components.
SALMONN-2 is a general-purpose audio large language model that leverages a unified self-supervised learning encoder with a novel multi-layer feature fusion adapter to achieve state-of-the-art performance across diverse audio tasks and enable multimodal in-context learning through targeted contextual biasing training.
This paper proposes a novel song aesthetics evaluation framework that utilizes Multi-Stem Attention Fusion to capture complex musical features and Hierarchical Granularity-Aware Interval Aggregation to model human perception nuances, demonstrating superior performance on both AI-generated and human-created song datasets compared to state-of-the-art models.
This paper proposes an unsupervised method for estimating the reliability of local segments in Dynamic Time Warping (DTW) alignments by measuring the agreement between the original path and a re-estimated path using FlexDTW with relaxed boundary conditions, achieving an aggregate AUROC of 0.97 in identifying reliable regions on audio-audio alignment tasks.
The paper introduces Audio-Visual Flamingo (AV-Flamingo), a fully open state-of-the-art large language model designed for joint understanding and reasoning over long, complex real-world videos by leveraging a massive new dataset, a progressive three-stage training curriculum, and a novel temporal interleaved chain-of-thought framework.
This paper proposes a reconstruction-based out-of-distribution detection method for vocoder recognition that utilizes an autoencoder architecture with contrastive learning and an auxiliary classifier to improve detection accuracy over existing probability-score approaches.
This paper introduces MIDI-RAE-JEPA, a self-supervised framework that combines a Swin Transformer V2 encoder with pitch- and time-shift equivariance objectives to learn hierarchical, semantically rich representations of symbolic music that excel in downstream tasks like emotion classification and enable high-quality generative modeling.
This paper proposes a filter-pruning strategy guided by norm-based criteria and lightweight finetuning to significantly reduce the computational cost of the AudioLDM model while maintaining or even improving its text-to-audio generation quality, including the recovery of specific sound events like gunshots and sirens.
This paper introduces "Bring Music The Horizon," an emotion-aware pipeline that generates immersive 360° music visualization videos by predicting a song's emotional trajectory and utilizing it to guide keyframe generation and video synthesis for a temporally continuous, immersive experience.