Home // PREDICTION SOLUTIONS 2025, International Conference on Prediction Solutions for Technical and Societal Systems // View article
Authors:
Simin Mirzaei
Hamid Reza Tohidypour
Panos Nasiopoulos
Deepan Chakravarthy
Fei Kuan
Leo Kamino
Mabel Wang
Tayyib Chohan
Keywords: 911 systems; speech signal processing; machine learning; speech emotion recognition.
Abstract:
Overwhelmed 9-1-1 systems during large-scale emergencies can leave calls unanswered, delaying life-saving responses. This paper introduces a machine learning–based framework designed to prioritize calls by urgency. The proposed method integrates audio signal analysis and text transcription pipelines, fuses predictions using logistic regression, and applies a penalty to minimize false negatives. We constructed a dataset of 1351 labeled calls using a mix of public datasets, simulated calls, and data augmentation. Evaluations have shown that our proposed approach achieves 94% accuracy with a 1.5% false-negative rate, surpassing baseline models, and operates in real time. These results highlight the system’s potential to enhance the reliability of emergency response by ensuring that the most urgent calls are identified and addressed promptly, thereby reducing delays and improving outcomes during large-scale crises.
Pages: 1 to 5
Copyright: Copyright (c) IARIA, 2025
Publication date: October 26, 2025
Published in: conference
ISBN: 978-1-68558-314-9
Location: Barcelona, Spain
Dates: from October 26, 2025 to October 30, 2025