Source Decomposition for Vehicle Sound Simulation

With the advent of Brand Engineering, the requirement for vehicles to have the right sound has never been more important. Given the scale of this engineering task, it is essential to follow a logical process which combines test and CAE data to create a “virtual vehicle” that can be used to simulate sound. A key step in this process is the ability to decompose the sound experienced by a vehicle occupant into its constituent parts (engine, intake, road, tyre, wind etc.). Once the source contributions have been separated, subjectively evaluated and objectively characterised, meaningful targets can be set and the task of engineering a particular sound quality becomes a possibility. This paper sets out a philosophy and methods by which the sound recorded inside the passenger compartment of a vehicle during normal driving conditions can be decomposed into the contribution of each major source. It also describes the process for resynthesising those separated source sound contributions back into the original binaural sound. Finally, it illustrates these concepts with examples and discusses the use of these techniques. INTRODUCTION Today’s goal for NVH engineers is to impart product image or “Brand Values” to the customer through the sounds of the vehicle by delivering not only customer acceptability, but customer delight. Since palliative treatments will rarely achieve the desired sound of the brand or delight the customer, NVH must now lead the design, as well as react to the inevitable issues. “Sound Engineering” is the process which allows us to define and engineer a Brand sound into products. As it is a top down approach to vehicle engineering, initially targets are set for the sound of the whole vehicle which reflect the Brand values and then cascade them to the sounds of its subsystems. The first step along this path is to decompose whole vehicle sounds into individual noise source contributions, i.e. road, wind, engine etc. This allows the balance of the whole vehicle sound to be evaluated by varying the contributions of each source using sound modification tools. This technique is often used to develop the whole vehicle sound target itself, since we have a far greater control and understanding of the sound by working at source level than whole vehicle level. The process of sound decomposition has two requirements. Firstly, the resultant sounds must be adequate for subjective (i.e. listening), as well as objective, evaluation. Secondly, to set targets for the vehicle and its subsystems, and gain information useful for engineering the vehicle, the techniques we use for sound decomposition must separate the sound into the contribution of physical, as opposed to virtual, sources. Once the interior sound has been separated into its physical source contributions, further decomposition can be performed; for example, the engine noise into structure-borne or airborne contributions, or the structure-borne engine noise into the inputs through different mounts / bushes, or the contribution of each body panel inside the vehicle from critical transfer paths. These analyses produce separated sounds that can be listened to individually or in groups. In all these cases, source decomposition is the first step. Finally, because the customer never hears the separate sources in isolation, we need the ability to resynthesise the whole vehicle sound from the separate contributions for any driving condition. Sound engineering therefore requires the use of tools and techniques in an integrated process which undertakes the decomposition, modification, re-synthesis, and presentation of vehicle sounds. Source Decomposition for Vehicle Sound Simulation Page 2 of 11 SOUND SOURCE DECOMPOSITION