Instrument for Soundscape recognition, identification and evaluation (ISRIE)

ISVR Consulting, the University of York and Newcastle University worked together over a 3-year period on a research project to develop an intelligent sound meter.  The principal idea was to develop a new type of meter which would not only measure noise levels, but would also characterise the soundscape in a given location and automatically identify or recognise, in real time, the types of sound contributing to the total acoustic environment.  The project was funded by the EPSRC.

The parameters that we use to define a sound environment are usually limited by what we can measure easily and quickly. Generally we measure the A-weighted sound pressure level. Most legislative controls and planning guidance are expressed as A-weighted levels. Often the levels are averaged over long periods to give a single number rating, or analysed statistically to show a distribution of levels. Rarely are the variations in the frequency and time domain considered in more detail.

The A-weighting is useful but not perfect. In general, people find some sound sources more annoying or intrusive than others, even if their A-weighted sound levels are the same. Often corrections need to be applied for impulsive or tonal sounds. But advances in technology may soon allow a sound level meter to discriminate between, and localise, sound sources enabling us to characterise ambient noise in terms of the relative contribution of different sources, such as road traffic, trains, aircraft, birdsong, industrial noise, etc.

 Although existing meters can provide measures of the total ambient noise, they are not yet able to provide information directly on the contribution of individual foreground and background source(s) to the overall mix of noise. Currently this sound identification is only possible with attended monitoring, where a person remains with the sound meter to identify sources, or by the time-consuming analysis of audio recordings, which is not always feasible. An advanced sound level meter that could recognise individual sources would have a significant impact on planning guidance.  It would, for example, enable planners to consider both ‘positive’ sound sources (e.g. natural sounds such as a waterfall or birdsong) and ‘negative’ noise sources (e.g. highways, industrial processes). The meter could also identify infrequent transient sounds, such as low flying military aircraft, which can cause great annoyance despite their relatively small contribution to the overall noise level.

A number of suggested platforms for measuring sounds along with promising techniques for advanced signal separation and classification have been proposed. The use of ISRIE could ultimately revolutionalise the way we currently perceive the sonic world around us and could directly affect the way we measure, record and assess sounds in the not-so-distant future.

Publications:

Oliver Bunting, Jon Stammers and David Chesmore (University of York); Omar Bouzid Gui Yun Tian (University of Newcastle-upon-Tyne); Christos Karatsovis and Stuart Dyne (ISVR Consulting, University of Southampton): Instrument for soundscape recognition, identification and evaluation (ISRIE): technology and practical uses.  Presented at EURONOISE 2009, October 26-28, Edinburgh, Scotland.

C. Karatsovis and S.J.C. Dyne, 2008: Instrument for soundscape recognition, identification and evaluation - an overview and potential use in legislative applicationsProceedings of the Institute of Acoustics, 30(2), 602-608, 2008.