Simplification of full spectrum soundscape analysis using specific frequencies

Abstract

Soundscapes are complex three-dimensional datasets that represent the sum total of sound within an environment, detailing the intensities and frequencies of signals within a time series. These data generates a three-dimensional graph that shows sound events generated by various sources of animals, anthropogenic sources, and environmental causes along a time series. These graphs are information dense and non-specific in nature, as it captures all sounds generated within the audible range of the receiver. Thus, the collection and processing of soundscape data remains a technological bottleneck due to the need for a large volume of acoustic datapoints to representatively map acoustic environments within time. Furthermore, the analysis of environmental acoustic data is difficult to comprehend without summarization or specialist knowledge.Acoustic events will occur as higher decibel pressure signals within a time series of an acoustic soundscape. Biological sounds are typically shown as complex groups of frequencies within a timeframe, these will occur and show as multiple frequencies of varying intensity on a soundscape. The strength of these signals and the distance from the receiver will determine how well these sounds are captured. Specific acoustic events are difficult to identify without knowledge of the acoustic signatures of potential causers, and may be challenging to separate from the soundscape noise depending on the ambient acoustic conditions. The ability to detect acoustic events is determined by how contrasted the relevant acoustic events are with the other environmental sound events and if there is any ambient masking in the environment.This paper demonstrates that the frequency bands of 63, 160, 400, 1000, 2500, 6250, and 16000 hertz can be used in place of a full frequency spectrum analysis. As acoustic events are complex, with varying intensity and frequency over a time period, furthermore these sounds generate harmonics, which further increases the range of frequencies which are detectable. The selection of the seven bands of frequency is obtained from commercially available frequency analysis standards, used to represent consumer electronic capabilities. The analytical output of the selected frequency bands is shown and compared to that of a full spectrum frequency analysis. Analysis of the selected frequencies are also presented, showing that both natural and anthropogenic sound events are detectable. An annotated graphical readout is also presented, showing that environmental acoustic sound events are detectable with specific frequency analysis.Comparative examples of this analysis methods are also shown, demonstrating the ability to detect differences between locations and relevant details for marine and terrestrial application are also discussed. These results show that a simplified level of data collection and processing can be carried out in lieu of a full frequency spectrum data and still provide the same level of detail for general acoustic analysis. This analysis method also reduces the amount of noise within the data and allows for faster processing times while still providing the same level of ability to detect changes within the acoustic environment. This paper is relevant to practitioners because it enables are greater volume of data to be processed as compared to the full spectrum analysis. It also makes the acoustic soundscapes more understandable to laypersons by reducing the amount of complexity within acoustic soundscape data.