@SounDoer：来自 Microsoft Research 的视频，声音设计师 Nick Wiswell 从声音录制、素材处理、系统结构等多个方面介绍了 Racing Game 声音设计，并对游戏混音做了详细的分析。Nick Wiswell 在游戏声音设计领域有着14年的从业经验，参与过 Forza Motorsport、Project Gotham Racing、Blur 和 MSR 等游戏项目。
Vroom, the interesting world of racing game sound design
A deep dive into the interesting, diverse and sometimes dangerous world of sound design for racing games. What makes a car sound the way it does, and how can we recreate that in real-time? How can games alter the focus of a mix when you don’t always know what will happen next? Games consoles support a myriad of surround formats, yet most games are heard via flat screen TV speakers. How can we make sure everyone gets the best experience possible? These questions, and hopefully many more from the audience, will be answered by a racing game audio veteran with 14 years’ experience of creating sound design for racing franchises including Forza Motorsport, Project Gotham Racing, Blur and MSR.
NOTE: This is a joint meeting organized by MSR and the Pacific Northwest Section of the Audio Engineering Society (AES). For more information about the section and future meetings announcement see http://www.aes.org/sections/pnw/. The meeting is public, feel free to forward the invitation outside Microsoft.
- Why do cars sound like they do?
Think of a car as a very large wind instrument.
The engine acts like a mouth and mouthpiece creating the airflow.
The intake and exhaust systems are tubes that create resonance.
The length and diameter of these tubes, combined with the number of cylinders they are connected to and the RPM of the engine, create a fundamental note and harmonics that determine the overall sound of the car.
RPM / 60 * N° Cylinders / 2 = Fundamental Hz
- Do you just play those sounds in the game? NO.
The audio needs to match what the car is doing in the game.
To do this we use the physics values that are driving the car in the game to determine what audio should play:
Boost Pressure (for Turbo and Supercharged cars)
Over 100 physics parameters are available to drive the audio system.
- How does that work? It’s complicated…
A simple version would be:
1) Record the car doing as many things as possible.
2) Take these recordings and chop them up into little pieces.
3) Piece them all back together again in game.
4) Map these sounds directly to the physics.
- Several ways to do this:
Crossfading looping sounds at fixed RPM’s.
- Are there any other things to consider? Lots of things.
Making sure the sound of each car component changes correctly based on the physics, using sample changes and real-time DSP.
Sounds for gear changes, backfires and exhaust back-pressure, turbos, superchargers and transmission whine for straight cut gearboxes in race cars.
Tire noise and skid sounds for over 30 different drivable surfaces. #可以说是游戏里最复杂的一个声音系统，一般会占用掉近一半的发声数。
Environmental reverb and early reflections.
Doppler shift for moving sounds.
Collisions, crowds and ambient world sounds. #真实的 Collisions 声音一点儿都不刺激。。。
- Cars are really loud in real life, games are not played at 120dB+
We need to find a way of simulating that volume.
We use distortion effects to help simulate the way people (and bad microphones) hear very loud things. #许多人都只是在油管视频里听到过赛车的声音，（由于实际拍摄质量和视频压缩的原因）他们会觉得这种“失真”的声音是真实的。
On FM4 we used custom version of iZotope Trash real-time in game. #Multiband Distortion
- Cars in the game move around a lot
Real-time surround panning to position the car based on the position of the car relative to the camera.
Real-time DSP to simulate distance, including a variety of filters, independent volume attenuation based on distance for environmental wet/dry and even a subtle phaser to give that distant “swirling” effect.
- Mixing for Interactive Media
In linear media everything happens the same way every time. With interactive media the user is in control so the experience can be different each time.
With linear media what you hear is very carefully crafted by the mix engineers to ensure everything that needs to be heard at each moment is clear and balanced in the mix.
With games you have no idea what is happening next, so not only are you playing back sounds in real-time but you also need to mix the game in real-time too.
- How can you mix interactive media?
1) Set it and hope: Give all your sounds a defined maximum volume level, and then let the audio properties of the sound mix the game for you.
2) Mixer Snapshots: Create separate mixes for different game scenarios and set up triggers in the game to switch between them at the appropriate time.
3) Compressors and side chains: Create a series of sub-mixes and use the fact that sound is playing through a bus to duck other less important sounds.
4) HDR Mixing: A fairly new concept where every sound in the game has a pre-determined relative volume. As louder sounds play, quieter sounds ducked or muted and then return as the louder sounds stop. #最初的设想来自 EA 某个内部引擎中的功能。
- Is there anything else to consider?
Almost all sounds in modern 3D games will move around relative to the player.
Games need to have a audio system that can deal with attenuating sounds in real time based on distance, direction, orientation and if there is anything obstructing the sound you are hearing.
This usually takes the form of volume, panning and filtering but other effects can be needed like Doppler shift for moving sounds and modulation effects for more distant sounds as previous mentioned.
Environmental audio, using real-time reverbs and delays to seat the sounds in their current environment.
- Which of these systems is best?
Different types of games work better with different systems.
If your game has very little audio or most of it is 2D “set it and hope” may well work just fine.
If your game is very scripted and you are generally aware of when mix changes will be required mixer snapshots is a good approach.
If your game is very dynamic with changes happening at any time then I would suggest side chaining or HDR to get the best results.
It really depends on the type of game you are making and how much of your performance budget you are prepared to allocate to it.
- What’s a performance budget?
One area that clearly differentiates games from linear media are the restrictions placed on audio by the capabilities of the target platform.
As all the audio is generated and mixed in real-time the audio system will have restrictions on CPU and memory usage, which vary greatly from system to system.
For example an iPhone game will probably have far less CPU and memory available than a AAA console title.
As consoles and computers have improved over the years audio quality has improved with more CPU available for voice count and real-time DSP, and more memory available for samples.
- Games, like many forms of media these days, can be played on a variety of different audio playback systems, from cell phone speakers, to flat screen TV’s to high end home theatre systems.
We need to make sure that the mix works as well as possible on as many types of playback systems as possible, focusing on the most common playback scenario.
Many audio professionals within the games industry will tell you this is a 5.1 surround system. #Nick Wiswell 认为目前最主流的重放设备还是电视机；游戏制作时以 5.1 系统为主。
- What do you think is the most common playback device? It’s a flat screen TV with poor quality tiny speakers.
And here’s more science, the laws of physics say that cheaply made, tiny TV speakers will not generate that 100Hz bass thump you added to all your sounds to add punch.
There is a way to maximize the sonic qualities of your mix for all playback systems, especially for games.
Make your sounds, roll off the extreme lows and test them on a low fi system to verify they have the low mid punch that will be reproduced on these small speakers. #Nick Wiswell 说他们会找 sounds-like-shit 的小音箱来试听测试，这些小音箱的特点就是 universal shit。
SounDoer – Focus On Sound Design