Soul of the Forest is an original game inspired by Undertale, Franken, and other comtemporary RPGs. Outcast from society and beaten by the king's shining champion, the injured Druid wanders into the forest. Discovering the magical nature of the woods and its impending doom at the hands of the tyrannical king, they embark on a quest to unite the forest. Can they save the Soul of the Forest?
As the Audio Lead on Wolverinesoft's Soul of the Forest project, I managed a team of 6 sound designers, composers, and technical audio specialists over 8 months. I was involved in all aspects of planning audio for the game. I helped facilitate brainstorming sessions with our team where we discussed how we could create an acoustic world with depth, realism, and the right character and emotional impact for our game. I also was responsible for tracking our progress and assigning tasks, helping my team learning new skills, and coordinating with the programmers, game designers, artists, and narrative team to make sure the audio represented and complemented their work. I also contacted professional sound designers in the industry and was able to schedule mentorship sessions with them for my team.
In terms of sound design and music, I contributed a significant amount of audio to Soul of the Forest. I wrote 3 musical tracks for the forest level, created dozens of original sound effects, and recorded and edited over 200 lines of dialogue read by 6 different voice actors. I recruited the voice actors and scheduled the recording sessions with them and my colleagues. Additionally I made field recordings and used them to create environmental ambience for the forest and castle. Using positioning in Wwise, 2D Box Colliders, and an EventPositionConfiner script in Unity, I made the environmental ambiences for all levels location specific. The forest birds and insects were different from those in the cornfields, the bustling village sounds were loudest in the center of the village, the blacksmith's iron emanated from a single stand, the river sound I recorded was located along the entire winding shape of the river, and the castle ambience contained randomized audio events, with their delay time also randomized between 4 and 12 seconds, to create the impression of a bustling castle occupied by nights and ladies occasionally passing through.
Additionally, I happened to have the chance to stay at the Anthenaeum, a historic building of brick, stone arches and old wood flooring built for Caltech in the 1930's. It was the perfect location to record castle footsteps on the granite halls, brick, wood flooring, and carpet. I recorded the sounds of swishing grass and leaves in my studio, and layered these sounds with some of the other footstep substrates. I randomized each set, for example a grass and dirt footstep would play simultaneously, but in a different combination (out of 6 samples) every time. This avoided the player's footsteps sounding repetitive and artificial.
For the dialogue of our non-English speaking forest creatures, I used a series of Random containers in Wwise to create a non-repeating 'language' for the mushroom characters as well. I recorded squeaks, snuffles, and snorts from pet store toys, cut them into 12 small samples of .5 - 1 seconds, and pitched them dramatically upwards. I created 2 random containers each with a playlist of 6 different audio objects. The first random container would play the 6 samples in a random order and a different number of repeats each time, for longer 'dialogue' lines. The second random container played every audio object on its list once, in a different order every time. I put these two containers within another random container so that the length of the "dialogue" would change each time.
I also used blend containers to create short 'words' which were only 2 to 3 samples long. I made 9 of these 'words' and put them in a random container set to play in Step Mode. Only one word would play for each dialogue line in the game, and I assigned these short words to the short responses in the game's dialogue.
I used similar techniques for most of the other forest animal dialogue. One of my team members made a roughly 5 second long sample of ants "chittering". I cut it into 6 pieces, put those into a random container, and randomized the pitch so that a different sample would play at a different pitch for each line of dialogue by the ant characters.
I also created mixers and buses to apply different processes to the various types of sounds in our game. Each type of audio had it's own mixer, so that I could adjust compression, EQ, gain, reverb, high/low pass filters to those groups of audio objects. These were routed to separate buses for Dialogue, Environment, Music, and SFX, with Aux buses with different effects for different contexts (castle vs outdoor, minigame sfx vs UI sfx). I used mixers (a type of container in Wwise with global settings for its child objects) to give all sfx the same treatment and volume, but then routed them to different auxillary buses based on their environments in the game's world.
I created more dramatic reverb and crunchy compression for battle SFX, and drier UI SFX. I created a subtle, short reverb with a low pass filter (to take out some of the highs that sound too much like walls in a room), and used different amounts of that for the outdoor ambiences and dialogue. Similarly I made a castle reverb with a longer tail, more early reflections, and more high end and applied it. I applied this in varying amounts to the castle ambience (footsteps, knights moving, scraping, etc) and the dialogue. My goal was to have an amount of reverb on the dialogue that is not noticeable, but simply sounds like it occupies a natural space rather than the ultra dry recordings we made in a small sound treated studio booth with a low gain dyamic mic.
About midway through the semester we realized that what we were hearing wasn't always obviously connected visually with what's on the screen. To fix this I learned how to use positioning in Wwise to locate sounds at their sources on the screen. Some were static, like the UI scroll with Quest updates at the upper left of the screen, so I assigned the sound to that area with balance-fade panning. Others sound sources weren't static and used the Direct Assignment setting. And some others had animations which I used automation of the sfx positioning to simulate.
In terms of balancing all the possible sounds that could be playing at once, we used kill switches to limit the number of sfx playing when they were attached to multiple objects (like all the knights in the castle), and we programmed the music and ambience to duck in volume or have a low pass filter applied during certain longer lasting sfx.
All of this work for Soul of the Forest was done in collaboration with a 55 person multidisciplinary team over two semesters at the University of Michigan's WolverineSoft Studio, a student games studio undertaking semester-long projects. Work on this project took place between August 2023 to April 2024.