There has been increasing interest in using automated bioacoustics analysis to monitor the environment. This involves using computational approaches to identify animals and other environmental phenomena from the sounds that they generate. The volume of data being recorded for bioacoustics analyses is increasing, as the scale of environmental surveys is increasing. This presents significant computational demands to perform analyses. These large-scale analyses cannot be performed at feasible speeds using traditional computing approaches. This research proposes AcoustiCloud: a system framework which represents bioacoustics processes as workflows and executes these across a Cloud-based system. It enables fast and efficient bioacoustics analysis for a variety of scenarios. The proposed system considers characteristics specific to bioacoustics processes resulting in fast execution times and high scalability. An implemented prototype is found to execute a bioacoustics workflow with 10 min of audio over 10 times faster than Pegasus, a widely used Workflow Management System.

使用自动生物声学分析监测环境的兴趣日益浓厚。这涉及使用计算方法从动物和其他动物产生的声音中识别它们。随着环境调查规模的增加，用于生物声学分析的记录数据量也在增加。这对执行分析提出了重要的计算要求。使用传统的计算方法无法以可行的速度执行这些大规模分析。这项研究提出了AcoustiCloud：一个系统框架，将生物声学过程表示为工作流程，并在基于Cloud的系统中执行这些工作流程。它可以针对各种场景进行快速有效的生物声学分析。所提出的系统考虑了生物声学过程特有的特性，从而导致了快速的执行时间和高可扩展性。发现一个已实现的原型可以执行生物声学工作流程，其音频处理时间为10分钟，比Pegasus（一种广泛使用的工作流程管理系统）快10倍。