Virtual Resident

What is going on

Long term annoyance

Expected Results by: January 2013

The data coming from the field studies performed around Arlanda Airport and London Heathrow will be adjusted to the necessary input data structure of the Virtual Resident model. Prediction models will be established and validated and finally most important parameters influencing annoyance will be identified.

Sound quality

Expected Results by: March 2013

The data of the Assessment of Single Overflight of the three field studies will be processed. All participants of the tests were sitting at home and judging during 2 times 1 hour long the actual aircraft flying over their houses. All flyovers of all recorded hours must be individually cut out, acoustically calibrated and then (psycho-)acoustical parameters have to be computed. Actually the data of the Cologne-Bonn Airport data is processed. Finally prediction models will be built to predict participants judgements.

Results achieved so far

The long-term predictor has been set up.

May 2010

First version of the long-term predictor based on the Frankfurt Study has been set up. Results show a mean prediction error below 10%. Based on parts of the conclusion, recommendation for WP2 field studies hase been formulated.

Best candidates as advanced aircraft sound quality descriptors have been identified.

May 2010

Automatic aircraft flyover detection algorithm have been set up.

May 2012

During the field studies, participants stayed most of the time at home during 4 days. At the end of each hour (daytime only) they had to rate the actual hour. Parallelly A-weighted Sound Pressure Levels (LA-logs) have been recorded outside their houses. An automatic aircraft flyover detection algorithm have been set up in order to identify those periods in participants' LA-logs, where flyovers occurred. For the flyovers acoustical parameters have been computed and then hourly dose parameters have been computed.

November 2012

The field study data of the Cologne-Bonn test have been completely processed. By means of the automatic aircraft flyover detection algorithm acoustical parameters have been computed for all participants' all rated hours. Neural Network—Classification And Regression Tree—and mixed models have been set up, which are able to predict hourly annoyance. It is assumed, that understanding and then lowering hourly annoyance will lead also to the lowering of long-term annoyance.


The main goal of WP4 is to develop a Virtual Resident (VRes) tool simulating human subjective perception of aircraft noise. As part of the model, the VRes tool will be able to predict human response to single aircraft sounds (e.g. sound quality), as well as the long-term annoyance of airport residents who are regularly exposed to multiple events. Better understanding of sound quality preference is expected to cover up a part of unexplained variance in long-term annoyance in current level based analyses. Acoustic and psycho-acoustic features of aircraft sounds (e.g. loudness, tonality, buzz-saw, etc.), multiple event descriptors (e.g. number of events, time duration between events, peak and overall loudness, etc.) and non-acoustic moderators (e.g. social status, mood, etc.) will all be taken into account to predict the virtual resident's response in terms of annoyance or preference.

The structure of the VRes tool will be illustrated through the following figure:

The intelligence, i.e. the knowledge base of the VRes tool will come from different sources:

  • Laboratory listening tests performed in the former EU project SEFA
  • Literature reviews and results of recent field studies (e.g. Frankfurt 2005 Study)
  • Laboratory tests, field studies and combined laboratory/field tests performed in WP2
  • Sound feature related threshold tests performed with the WP3 On-line Sound Synthesis Machine

The VRes tool will be developed into a standalone, executable program and will be extensively tested and validated. The tool will further be used in a close interaction with the WP3 Airport Noise Climate Synthesizer tool to predict the annoyance of various airport scenarios (e.g. current scenarios, optimized future ones, etc.) specified by WP5.

copyright | contact | sitemap