Heavy Vehicle Sound Study
In this sound study, I investigated the audio sources and specific characteristics of a military heavy tracked vehicle before recreating and implementing these sounds within a game-ready vehicle audio system.
I decided to design and implement audio for a heavy military tracked vehicle. I ended up creating a game-ready audio implementation for the BMP-3, a soviet era military vehicle with a loud diesel engine.
- Designed sounds for the various components of the vehicle.
- Created parameter-driven MetaSounds patches for the vehicle.
- Implemented audio for the vehicle.
- Programmed a drivable vehicle Pawn for the BMP-3 model.
Unreal Engine 5
The old Soviet V10 engine powering the BMP-3 is known to create quite a ruckus under load due to the lack of a muffler. In order to replicate this sound, I created layers for 1000, 2000 and 3000 rpm, along with a layer for when the engine is at idle. This approach was convenient because these rpm values are harmonically related. 1000 and 2000 rpm are an octave apart, with the 3000 rpm layer being the equivalent of one and a half times the frequency of the 1000 rpm layer. This way, I could guarantee layer crossfades without noticeable pitch deviation.
To accomplish this, I synthesized the majority of the engine sounds from brief waveforms of truck engines. This approach enabled me to achieve a perfectly stable rpm value for each layer. Within the MetaSound patch, the relative pitch and balance of all layers are governed by the same algorithm, which takes into account rpm, speed, torque, and throttle.
For the idle layer, I sampled individual combustion cycles from a truck engine recording and played them back rapidly in a random sequence at the correct speed. I also introduced some additional popping 'misfires' to keep the engine sound interesting at lower speeds.
The layers are filtered and attenuated based on distance and the azimuth to the engine's exhaust.
To recreate the recognizable sound of rattling tracks, I gathered a range of recordings featuring metal objects colliding or moving against each other. I then edited and processed these recordings into looping layers, each of which sonically represents a specific component of the track undercarriage, such as the sprocket wheel or track pins. These layers are blended together based on the vehicles speed and the roughness of the ground underneath the tracks.
To recreate the sound of the interaction between the track pads and the surface underneath, I collected field recordings of similar interactions: metal scraping on concrete, stones being crushed, and objects sliding through dirt. These were then processed into loopable layers and blended in the MetaSounds patch based on the surface underneath the tracks and the speed of the vehicle. To get the surface underneath the vehicle, a linetrace is performed each frame underneath each road wheel, this results in 12 unique probe points that are used to determine the correct blend of different layers. This way, the surface sounds of the tracks stay accurate if the vehicle is driving over different surfaces at once.