Gunshot Detection by STE and ZCR

Pei-Chen Lin ； Che-Yen Wen

《Forensic Science Journal》 18卷1期 (2019/12) Pp. 35-46

https://doi.org/10.6593/FSJ.201912_18(1).0004

摘要

Gunshot detection has been one of important research topics in military and law enforcement applications. It has been used to detect the direction and localization of shooting events. In audio analysis, it plays an important role of finding the possible gunshot time period, called gunshot audio segmentation. The short-time energy function (STE) and the zero-crossing rate function (ZCR) are useful methods for voiced and unvoiced detection. We can use STE and ZCR to detect signals with the strong energy and highlight the sound period with rich detail. The quality of audio data may be affected by recording devices. However, in previous gunshot detection research work, most of their audio data were not collected by cellphones. Since many of audio data collected from criminal investigation cases are recorded by cellphones in Taiwan, we need to evaluate the performance of detection methods. In this paper, we apply STE and ZCR to gunshot detection of audio recorded by cellphones (32 samples). From experimental results, we can obtain good results.

關鍵字

gunshot detection ； short-time energy ； zero-crossing rate