Abstract The Ring of Fire represents a constant threat in the case of possible eruptions due to its 160 active volcanoes. Consequently, several studies have determined that an increase of tectonic and long period events could indicate the beginning of a volcanic eruptive process. In this context, it is crucial to improve microearthquake detection by using real-time automatic recognition systems. For this purpose, this work aims to improve microearthquake detection based on Voice Activity Detection (VAD) and endpoint detection. Furthermore, we propose the use of signal segmentation, considering the signal-to-noise ratio, and adapting VAD techniques for microearthquake detection. The results tested in real-time have demonstrated the reliability of our detector by obtaining 97.89% precision, 97.10% accuracy, and a value of 0.015 of Balanced Error Rate (BER). We considered a dataset for real-time testing of 744 h of continuous data from Cotopaxi Volcano in Ecuador. In these tests, the MDA detector is 8.89% more precise than the classical STA/LTA detector.

中文翻译：

---

基于语音活动检测和端点检测的实时微地震检测器：科托帕希火山的一种方法

摘要 由于其 160 座活火山，火环在可能爆发的情况下代表着持续的威胁。因此，几项研究已经确定，构造和长期事件的增加可能表明火山喷发过程的开始。在这种情况下，使用实时自动识别系统改进微地震检测至关重要。为此，这项工作旨在改进基于语音活动检测 (VAD) 和端点检测的微地震检测。此外，我们建议使用信号分割，考虑信噪比，并采用 VAD 技术进行微地震检测。实时测试的结果通过获得 97.89% 的精度、97.10% 的准确度和 0 值证明了我们检测器的可靠性。015 平衡误码率 (BER)。我们考虑了一个数据集，用于实时测试来自厄瓜多尔科托帕希火山的 744 小时连续数据。在这些测试中，MDA 检测器的精度比经典 STA/LTA ​​检测器高 8.89%。