An Intelligent Noise Meter
Jon Stammers
Research Student
Intelligent Systems Group
Department of Electronics
University of York
01904 432825
js185@york.ac.uk
Also Dr. David Chesmore,
University of York
(+44)1904 432394,
edc1@ohm.york.ac.uk .
Popular version of paper 1pSPa3
"Instrument for Soundscape Recognition, Identification and Evaluation (ISRIE): Signal Classification"
Wouldn’t it be nice to know what that sound was that kept you up until 3 in the morning? Maybe you know it was one of your neighbours having a party but will the authorities do anything about it? Unless an official recording was made at the time, probably not. The Instrument for Soundscape Recognition, Identification and Evaluation (or ISRIE) is proposed as an “intelligent” noise monitoring system and could provide the solution in circumstances such as this. The project is a collaboration between 3 UK institutions – the Universities of York and Newcastle and ISVR Consulting, Southampton – and its aim is to produce a device capable of distinguishing between the most prominent sounds found in a typical urban soundscape. The set of sounds which the project aims to recognise are based on current UK noise legislation and range from the bark of a dog to the constant hum emitted by an air conditioning unit.
Methods for separating out the various sources in a soundscape and identifying them are being investigated in York. The signals will first be captured by a microphone capable of listening in 3D. Novel processing techniques are then applied to these signals to break them into the sounds that they are made up from with information about their direction of origin being stored as well. Once the individual sounds have been separated a recognition system will then aim to place them in predefined categories. This system uses a type of analysis called Time-Domain Signal Coding which has shown previous success in the identification of birds, bats and insects. This technique requires less computation than more common signal processing methods so is ideally suited to a portable device. Coupled to the time-domain analysis will be a pattern recognition system more commonly used for analysis of speech.
Members of the project group at the Institute of Sound and Vibration Research (ISVR) are conducting research into how ISRIE could be used by sound consultants in their work. One way in which the system could be used is to aid in the classification required in Planning Policy Guidance (PPG) 24 and BS 4142 enquiries. In both of these cases a sonic nuisance has to be given a definite category which is normally chosen based on the contents of the sound. This can be quite subjective as it is up to the investigating consultant to decide which category a sound belongs to. This task could be performed by ISRIE and then quickly verified by a consultant.
The team at Newcastle University are developing methods for networking a number of ISRIE devices together. By doing so it could be possible to identify where a sound originated from making it possible to pinpoint a noise nuisance.
The future applications of the ISRIE system are by no means limited to examining the sounds emanating from an urban environment. It has already been suggested that this device could help in biodiversity studies to identify a number of different species in a particular area. Another option could be to use it in a large manufacturing plant to recognise the sound of a malfunctioning piece of machinery.