David Lammers

My responsibilities

On this project my main responsibility was audio, in every aspect. In addition to that I was also responsible for making sure all of the software worked together. This meant I had to do extensive testing with engine versions to ensure compatibility as well as ensuring that we could lock in each version.

I was also responsible for our choice of audio engine, Wwise. This is where most of my compatibility testing was focused, because there was no other major software that was requested to be integrated. The reason we went with Wwise was because of the length of the project and Wwise’s scalability. During this project, the idea of changing versions was floated around due to engine issues, so when this happened, I was a significant factor in deciding whether to actually swap versions or not. I managed to keep the versions of all software involved the same throughout the entire project, but this was through extensive testing and problem-solving.

Pipeline

On previous projects, the pipeline ended up messy due to my attempts at creating consistent documentation that the team did not use.
Because of this, I decided to change the way the audio pipeline worked massively for this project. I made generic placeholder sounds for everybody to use in every feature they would make. Then, during sprint planning, I would examine which placeholders were still in the current build, what the team would be working on next, and the ideas on the board, and make my sprint planning based on that. This kept the audio pipeline very short and always on point. Sadly, this also made the pipeline very volatile and short-sighted. Despite this, the new pipeline ended up being much better for ensuring audio quality across the board as well as ensuring that every asset had the appropriate audio to go along with it. While this pipeline is not at all scalable for bigger audio teams, this is definitely a pipeline I will keep using if I am one of the very few or the only one responsible for audio.

Wwise dynamic mixing

While most of the advanced spatial features of Wwise ended up not being usefull for our project, the better mixing tools of Wwise allowed me to create a proper dynamic mix.

I attended a talk by Ninja Kiwi about how they did the audio for Hellblade 2. In this talk they briefly showed a screenshot of their dynamic mix setup. Using this screenshot I reverse-engineered thier dynamic mixing setup and implemented it into our own game. During this process I decided to simplify the dynamic mix system because our music is steady, meaning that the frequency usage is constantly known. The Hellblade dynamic mixing actively compared different frequency bands constantly due to the vastly changing ambience. Because the frequencies of the music don’t change much I could skip this step to save on processing power.

Using this knowledge I made the following dynamic mix system: All of the combat sounds go into 1 bus, I measure the volume of this bus to guage how deep into combat the player is. Using this measured volume I duck the music, and apply an EQ to duck certain frequencies of the music harder than others. This makes space for all of the combat sounds while keeping the drums of the music mostly audible to maintain consistency in the music. The parts of the music with a very short attack also get a compressor applied to them at high enough combat volume to prevent these parts from overpowering the ducking and coming through in the combat mix.

Boss sounds

This game’s boss sounds to me are the proof of my capabilities as an audio designer.

The story of the boss sounds is a rough one. The boss was added quite late in development, and 1 day before our build deadline. The boss caught me completely by surprise, which highlights the downsides of my established pipeline. Despite this I managed to make all of the sounds for the boss in time for the build the next day. This took me about 11 hours of straight work.

Due to the interactivity of the boss sounds as well as the real-time changes I would have to make to the sounds in order to make them feel right I decided to heavily utilize synthesizers for the boss sounds. Synthesizing these sounds would allow me to keep the limitations of Wwise real-time changes in mind at the very beginning of asset creation, massively simplifying the implementation step.

Next I used several of the tricks I had learned over the course of the project in Wwise to provide additional detail to the sounds that would take really long to otherwise bake into the sounds themselves. This mainly involved having the boss’ ambient sounds change depending on which attack its using.