Today's automotive vehicle design process is driven by brand image coupled with customer expectations. In order for designers to assess whether they will delight or even satisfy their customers, it is necessary to gain an appreciation of the performance of their components evaluated in the true context, not represented as functions on a graph. In short, being "quiet" no longer guarantees that a manufacturer will be ahead of the competition in the market; style and character have an important part to play.

An engineering process has been developed to enable customer requirements to drive the vehicle design. In this process, concept and hybrid modelling is used to make assessments (which are validated) of how our designs perform, enabling us to assemble simulations in a simple and timely manner (using readily available data).

This has caused a fundamental change in NVH engineering for the automotive industry. Where this process is not used, NVH departments essentially perform a monitoring and "damage limitation" role throughout the design process until sufficiently representative vehicles are available for test and assessment. At this point a lot of work is undertaken to engineer out NVH issues which have occurred. The new process enables NVH engineers to assess and evaluate designs and concepts from day 1 of the programme, and generic targets involving the brand values of the manufacturer can be prepared well in advance of a project as a general maintenance issue.

However, the results of these analysis techniques need to be interpreted, since they are usually of numerical or graphical form. Over the last 10 years, techniques have been developed to convert these graphs into sounds, which can be auditioned on headphones in a laboratory. There are some problems with this method of presentation, namely that the perception of sounds played on headphones in a room is quite different to the perception of the same sounds heard in a car being driven on the public road. Furthermore, these offline sound generation techniques are based on repeatable test measurements (for obvious reasons) - the problem is that no member of the general public ever drives his car in this manner - a classic example is the 3rd gear full load acceleration from 1000 to max rpm. Whilst these sounds are useful to engineers, and significantly better than graphs, they do not necessarily give a direct link to the customers' impression.

The perceptual differences between listening to sounds in a room and hearing sounds in a car on the road are due to the impoverished environment, i.e. lack of vibration, which especially inhibits the perception of low frequencies, and the fact that the subject is doing what they are asked to do - listen to sounds, as opposed to drive a car, which involves looking at the surroundings and making continual judgements about the environment. This impoverishment results in over concentration on the sounds and a bias in the results - some sounds become significant, when in the real vehicle they pass unnoticed.

What is needed therefore is somewhere where sounds can be evaluated as they are in the real world. But, the real world isn't perfect either - there are as many problems with real world testing as there are with jury appraisals in the laboratory, repeatability and the rather short auditory memory of humans to name but two. The half way house, where the best of both worlds can be found, is in a vehicle simulator.