Unifying the Sonic Experience of ‘Split Fiction’ – with Hazelight Studios’ Game Audio Team

Teamwork makes the dream work, they say. No statement could be truer for Hazelight Studios’ newest split-screen co-op action-adventure game Split Fiction (published by Electronic Arts).

In the game, players take on the roles of two novice writers, Mio and Zoe, who are caught in a virtual realm where their short story ideas are playing out. And their story aesthetics, by the way, couldn’t be more different. Mio’s ideas are futuristic sci-fi, while Zoe’s ideas are based in the fantasy realm. Since they’re sharing one simulated experience, they’re sent from one of Mio’s sci-fi stories into one of Zoe’s fantasy stories via portals and back again. Plus, there are twelve different side-stories. Escaping the simulation means relying on each other to solve the puzzles, overcome challenging scenarios, battle bosses, and ultimately, cooperate to achieve a shared goal.

In making the sound of the game, the Hazelight Studios audio team had to depend on each other to solve the puzzle of how to efficiently craft tons of sci-fi and fantasy sounds that felt part of a unified experience, overcome the unique challenges of mixing a split-screen game for two listeners, battle against a looming launch deadline, and ultimately, come together to deliver a satisfying sonic experience.

It was no small feat, but under the audio direction of Philip Eriksson, they accomplished all they set out to do. Here, Eriksson, Joakim Enigk Sjöberg (Senior Technical Sound Designer), Göran Kristiansson (Senior Audio Programmer), Gustav Landerholm (Senior Sound Designer), Filip Järnmark (Sound Designer), and Composer Gustaf Grefberg at Hazelight Studios talk about how their experience on the previous release It Takes Two helped to inform their process on Split Fiction, how building bespoke systems such as “SoundDefinitions” (a self-contained audio ecosystem built as an extension of Unreal Blueprint) allowed the sound designers to be fully self-sufficient in audio implementation, how they handled the creation of a huge amount of sound elements from two different genres using a “First pass only pass” approach, how they kept the mix (that is playing sounds for two separate listeners) from becoming a cacophony, how they used music to bridge the split-screen sounds and create a cohesive sonic experience, and much, much more!

Split Fiction | Official Co-op Gameplay Trailer

Coming off the success of It Takes Two, and taking what you’ve learned from that experience, how did you want to expand your use of sound for Split Fiction?

Philip Eriksson (PE): We basically threw away everything from our previous game It Takes Two. When it comes to audio tech (but also in general), we looked back at ITT as an experimental phase and our first step. We felt like ITT was only our “proof of concept” — we didn’t have anything before we started, so everything was made in a rushed way. When we started on Split Fiction, we felt it was time to show off our skills and abilities. And, we had the time to do our best on this game and make it the way we wanted.

We had long and tough discussions around our tech, audio implementation, and overall methods. During ITT, these things were basic and required more work from gameplay coders and designers. Now we wanted to break free and do as much as possible as sound designers. This enabled us to be fast, agile, and to own the sounds from creation to final implementation. We had laid the foundation for our split-screen listener solutions during ITT, so we kept all of our ideas, knowledge, and our solid foundation in some aspects from ITT and went on to expand, improve, and progress.

When it comes to the audio direction and the sound design work, we started from scratch, yet we had the experience of working together as a team for two to three years, as well as the benefit of starting together with the rest of the studio right away from the prototype stage. We wanted to make something spectacular.

we spent two wild years creating sounds at an intense pace while at the same time creating the tech needed to realise the vision in the most ad-hoc way imaginable

Joakim Enigk Sjöberg (JES): Making It Takes Two was such a non-stop ride for all of us involved with audio. The truth is that the team hadn’t formed until after the pre-production phase had just ended, which meant that we spent two wild years creating sounds at an intense pace while at the same time creating the tech needed to realise the vision in the most ad-hoc way imaginable. As the dust settled, we found ourselves with a pretty functional toolbox, although not a user-friendly one.

As we anticipated starting a new game, my goal was to establish a way for sound, VO, and music designers to be fully self-sufficient in the implementation of audio, giving them the ability to interface with any gameplay system without the immediate need to solicit game designers or programmers. This ambition was made real in the design of an asset called the Sound Definition, or SoundDef for short — a self-contained audio ecosystem built as an extension of Unreal Blueprint.

Unreal ‘Sound Definition’

SoundDef is modular, feature-rich with a hybrid approach of visual-based and text-based scripting, and most importantly, it’s fully owned by the individual audio designer. We put all of our audio eggs (i.e., sound effects, music, and VO) in the SoundDef-basket, as it became the one-stop shop for nearly all the sounds you hear in Split Fiction.

While some readers might think that this sounds similar to the workflow that MetaSounds offers inside of Unreal, the difference is actually quite stark. SoundDefs handle no audio generation — that all happens inside of Wwise. Instead, their job is to be a creative canvas where all of our designers on the audio team can pull data from the game and convert it to curated audio logic for Wwise to act upon.

Göran Kristiansson (GK): I entered the rollercoaster ride of It Takes Two in the middle of production, so it was all about making use of or expanding on what we had. At the same time, we were trying to find pain points in our processes from sound design tools to pushing play.

So when starting anew, and with a mindset of no feature or ask being too small or too big […] I remade everything from the ground up

So when starting anew, and with a mindset of no feature or ask being too small or too big, especially since we wanted to empower each person’s processes, I remade everything from the ground up — remaking our backend to suit our needs and the two-listener setup.

For example, our music composer Gustaf Grefberg felt limited in ITT because we didn’t have a good systemic way of doing tonal stingers depending on the currently playing music. So I expanded our sound engine, Wwise’s music system, and UE’s backend with a hybrid approach. Gustaf could then create a playlist of stingers and the system selected what to play depending on which chord the current music was playing, with the option of overriding it too.

The Hazelight Studios Audio Team on ‘Split Fiction’

I love how different Mio and Zoe’s stories are! How did you approach the creation of these two different realms? Did you divide the sound work between realms (so, one team created the sci-fi realm sound, and another created the fantasy realm sound)? Or, did the whole sound team tackle one realm first and then the other? There’s such a variety of sound, and the action is non-stop, so there’s a lot of sound! How did you accomplish all of this?!?

PE: It would not have been possible with Hazelight’s game development methods to do all of the fantasy sound first and then move on to create everything for the sci-fi realm. We had to be agile. We jumped around, but we did try to stay as much as possible as one group. We went between the realms as one unit.

During the main production of the game, we tried to do these longer 4-6 week passes on a decent chunk of a level together as one unit. This was a very nice approach since we could make music and most of the sound effects for a whole part of the game. We had to leave certain holes in the soundscape, but we tried to make everything as final as possible.

The approach of bouncing between the realms made the work varied and fun. Just like playing Split Fiction, it did not become repetitive or mundane to work on either. We also wanted each chapter to have a unique style, so after working on fantasy sounds for a while, it was good to get back to sci-fi. We had fresh ideas every time a new pass started.

We jumped around, but we did try to stay as much as possible as one group. We went between the realms as one unit.

Towards the later parts of production, we had to split up our group more and more. This was because more parts of the game were finished and most of them finished at the same time, so we needed to tackle the game in a wider way. This for sure helped production, but it was easy to notice that we had not stayed as one unit later in the mixing stage. The levels that were made in the most spread out way had to be made in an ad-hoc “jump in, make a sound, and jump out” kind of way. This negatively affected the coherence, made the pre-mix rougher, and caused more bugs. But it had to be done; there was no way around it.

Our Hazelight audio team strives for and encourages sound design generalists. We’ve all done so many different types of sounds for Split Fiction as a whole, but also for all our realms. We had a few soft-owners for some parts of the audio work, but everyone has done everything together.

To be honest, I am not sure how we made all of these sounds. It certainly required a lot of hard work from everyone on the audio team, no question about it.

We had some external help within the sound design team for this game. Sebastian Pohle (Pole Position Production) worked with us full-time inhouse for over two years and Josh Smith (Respawn Entertainment) came on and did some heavy lifting! Funny story, when we did the onboarding of both of our external sound designers, they said, “This is impossible.” And we all agreed, but we managed anyway.

I will, though, be transparent and say that we would not have been able to pull this off if we had only worked normal working hours. A lot more effort has been put into this game’s sound. There is no magic trick. The incredible variation of Split Fiction and its insane amount of gameplay requires a lot of sound. Creating a lot of sounds takes a lot of time. The hard work of each individual who has made the sound for this game is the magic; my fellow colleagues are the magic.

we have developed a method we call “First pass only pass” within the audio team.

To expand on our strategies a bit, we have developed a method we call “First pass only pass” within the audio team. This has been a necessity so far with the projects I have made at Hazelight, but we are trying to reiterate that method to make it more sustainable for us. This method has been a necessary evil, yet it has taught us valuable skills as well: quickness and accuracy. We always try to finish a sound task instead of leaving it half done. We have saved a lot of time doing this. We never wait for the latest VFX or other things, but make it sound good from the start. In so many places in Split Fiction, the first iteration shipped. For example, we worked on The Megalith part pretty early on, when it was still in a rough visual state but was playable. 98% of those sounds were good enough and shipped.

So it is possible to lead with sound and not sit around and wait. If we had waited, then it would have been too late, and parts of the game would not have had sound.

One last thing on this “First pass only pass” topic, which I feel is something that makes the audio team at Hazelight special, is that our first pass is usually pretty good — we are fast but still manage to hit a quality bar that is usually good enough.

How did you want the fantasy realms to sound? Did you have references of fantasy projects (e.g., films, series, or games) that you liked? Were there fantasy sound tropes you wanted to avoid?

(I have to tell you that I loved the Donkey Kong nod in the first fantasy realm, with the troll throwing barrels down the road! And I loved the Giants/mountain climbing realm with the massive gong!)

PE: We did not want fantasy to sound too chimey and sparkly or even — to be a bit harsh — too “gamey.” We wanted a grown-up, natural, warm version with natural sounds that are slowly evolving. We use a lot of natural recorded sources in our designs and make them work for the scenarios through granular processing.

We always tried to differentiate our two main realms from each other. If fantasy is warm and slow, then sci-fi is colder and punchier. Another thing we played with was adding slight harmonic distortion to our asset choosing: even harmonic distortion for fantasy and odd harmonic distortion for sci-fi, to make a difference on the most microscopic level.

Split Fiction: Footage from the giants vocal recording sessions – conducted by composer Gustav Grefberg

we played with […] adding slight harmonic distortion to our asset choosing: even harmonic distortion for fantasy and odd harmonic distortion for sci-fi

Our mantra for this game was “Ear Candy” — always giving a new, fresh, and interesting sound to the players. We created a conveyor belt of interesting new sounds for the player through our constant progression through our levels and the alternation between fantasy and sci-fi.

We didn’t have one singular game or movie as our inspiration for fantasy. I personally always felt that fantasy was more difficult than sci-fi for some reason, but in the end, it was a mixture of inspirations and our own tastes that made our fantasy audio direction.

we never fought or struggled to find a unified audio direction within the audio team; we had similar ideas and tastes.

A few examples are the high fantasy style of Disney’s Maleficent from which we drew some inspiration for the magic abilities, but also the natural beauty in the ambiences and the sound for Maleficent’s wings. We listened to the later, darker Harry Potter films and found there to be a layer of realism to the magic spells; the impacts of them especially resonated with us. In the Harry Potter universe, there are a lot of very large and interesting creatures. This was one of many inspirations for the dragon sounds. We never wanted to “copy and paste” that, but rather put our own flavor on things. A last example of a fantasy masterpiece is The Lord of the Rings. The excellent character building through sound for creatures is one of the pieces that inspired us. I am referring to, for example, the Cave Troll sound design that is so beautifully made by David Farmer. We were very inspired by sound storytelling.

One thing to note is that we never fought or struggled to find a unified audio direction within the audio team; we had similar ideas and tastes.

How did you want the sci-fi realms to sound? Did you have references for sci-fi projects that you liked? Were there sci-fi sound tropes you wanted to avoid?

(I love the ‘Neon Revenge’ realm. It reminds me of Tron!)

PE: The Neon Revenge chapter to me is sooo Blade Runner, which I absolutely love. But I can also see the similarity to Tron, especially when it comes to the bikes.

As I said above, I always felt like sci-fi was easier for me to know what kind of style I wanted it to have. Maybe that stems from my love for bigness and my inability to make sounds that are more subtle and beautiful. I’m not sure. Or maybe it is because I have some experience of making sounds for at least one other science fiction-based universe through my work on the Star Wars Battlefront titles by EA DICE.

Through the immense variation of our gameplay and levels, the sci-fi references also became highly varied. We wanted to make each chapter within a realm have its unique flavour, to keep the soundscape fresh, and also match the constantly changing art style in our worlds.

We wanted to deepen the narrative and the characters by trying to imagine how Mio would have written sci-fi.

We wanted to deepen the narrative and the characters by trying to imagine how Mio would have written sci-fi. We wanted to portray her own grit, coolness, and personal past, and reflect that in the sound. She was the one who wrote these stories, so we needed to ask ourselves, “What does Mio like?” and “What inspired Mio?”

For The Neon Revenge specifically, we were definitely inspired by Blade Runner, as I’ve mentioned. Some inspirational parts from that universe were the city ambiences with the diverse variation of languages heard through the constant chaos of commercials, as were the spinner sound design and the blasters — lots of good stuff to get inspired by.

We wanted our vehicles and chase sections to have that same realistic worldizing and reverberation, even though the sounds were going to be very designed and unreal.

Another inspiration amongst many is the brilliant sound design for the Bat-bike in the film The Dark Knight. One part that was particularly interesting for us from that film was how believable the bike sounds when it is going through the shopping mall. We wanted our vehicles and chase sections to have that same realistic worldizing and reverberation, even though the sounds were going to be very designed and unreal.

We were also inspired by the sound character and style of The Matrix. We thought that style resonated with Mio as a character and included both a raw, gritty, realistic style at the same time as being very over-the-top superhero. We even honed in on things like the Sentinels from The Matrix and used that as an inspiration for the last boss fight in the Neon Realm.

What went into the sound of the “glitch” realm/space? (The voice echo/delay in here is really lovely and understated. You get a little taste of an echo/delay on their voices — so cool!)

PE: We had grand ideas for the “white space” from the earliest versions of the audio direction. We wanted to be bold with the mix! We wanted to create a vacuum feel, which we did, but maybe not as much as we had initially planned. Things change so quickly during the game’s production, but I still feel like we managed to make it effective.

The overall audio narrative for this space was to have it be disorienting at first, because Mio and Zoe don’t know what this place is in the simulation.

We wanted to create a disorienting, non-logical feeling the first time Mio and Zoe arrive at the “white space.” Then we wanted to create a strong progression with the sound of the “white space” and the glitches that followed the drama. This part is something that Uhørt (our cinematic audio outsourcing partner) and Pål Baglo (Lead Cinematic Sound Designer) really delivered on! The overall audio narrative for this space was to have it be disorienting at first, because Mio and Zoe don’t know what this place is in the simulation. It’s completely unattached to their own stories.

But as they learn about the glitches and how they can use them to escape and destroy the machine, we wanted it to progressively feel less and less frightening and disorienting and more “friendly.” Of course, towards the end, with Rader’s final violent efforts to maintain control, the “white space” and the glitches turn evil, until the inevitable destruction of the whole simulation.

It was such a pleasure to write up a bunch of weird words for the audio direction that described the “white space” and then have creative discussions with Pål around this. There was a bit of back and forth between us and Pål since this was one of the first things we started working on during our collaboration. But the results turned out so awesome!

We honed in on it being the raw form of the simulation, and relied heavily on synthesized sounds that we resampled and glitched out

Filip Järnmark FJ: In the System Failsafe Mode level (RedSpace), when Rader shows up and locks Mio and Zoe in and tethers them together, the “white space” is completely transformed and suddenly becomes hostile and more disorienting than ever. Up until this point, the glitch space is only seen in cutscenes, and with it becoming a playable environment, we needed to even more clearly distinguish it sonically from the main fantasy and sci-fi realms.

We honed in on it being the raw form of the simulation, and relied heavily on synthesized sounds that we resampled and glitched out — thank you KiloHeart’s Phase Plant and Reaper‘s global playback rate! We also had a fantastic recording available that Joakim (JES) had accidentally made when capturing gameplay with broken Vorbis decoders that was perfect for this purpose, as it basically captured a machine going haywire.

And as you mention, the sounds in this level are “worldized” with a bespoke IR that was a long reversed ping-pong delay/reverb with 5 taps that helped further enhance the disorienting and hostile feeling that we wanted here.

What went into the sounds of Rader Publishing’s lab, with the high-tech virtual reality ‘story-capturing’ machines and cool bubbles that the writers are contained in?

PE: We wanted The Machine to be one of the main characters in the room since the machine is the connection between Rader in the real world and Mio and Zoe in the simulation. One difficulty with the real world was not to make it sound too much like sci-fi, since there is quite a lot of high-tech in that environment as well.

One difficulty with the real world was not to make it sound too much like sci-fi

We wanted all our realms to have a really clear distinction from one another. One thing that Josef Fares (Creative Director) mentioned was that he wanted the players to instantly know which realm they were in. We took that with us and wanted our soundscapes to be so clearly fantasy and sci-fi that the players could close their eyes and still know perfectly clear which realm they were in. Josef had a fun way of describing this!

“Imagine that I have printed out a bunch of pictures of fantasy and sci-fi. Then I show you a random picture at a time, within the first second of seeing the picture that I hold up, you can tell me if it is either fantasy or sci-fi.”

This was the goal for all of us.

we described these ‘sound frames’ for each realm by specifying what identified each realm and also what did not belong at all within that specific realm.

In the Audio Direction presentation that we collaboratively created, we described these “sound frames” for each realm by specifying what identified each realm and also what did not belong at all within that specific realm. This helped our audio team to understand how each realm needed to sound, but also very clearly how it should not sound. These frames were just described with some adjectives, but even as little as that can paint a pretty clear frame if you choose the words carefully.

When it came to the real world, we described it as slightly older and more normal than it looked, maybe actually having slightly older phones ringing in the corridors of the Rader Corp building, as an example, to be able to build and progress towards an even newer and more advanced universe when going into the simulation of the sci-fi realm.

The crack in the “white space” that the glitches create on Mio/Zoe’s side and the machine on Rader’s side are two really important tools for the sound narrative. If a crack in the “white space” is created, then the machine needs to scream in agony, too. It is a very elegant way of the through-sound explaining who is in control of the situation: Mio, Zoe, or Rader.

What went into the sound of the portal?

JES: The role of the “Side Story Portal” was something that dramatically changed during production. The amount of these side stories, and their placement within the main levels, sometimes changed weekly. As their place in the game and the story started to solidify, one thing became clear — we wanted to use audio to give them a very strong identity within the surrounding soundscape. The idea is that their very existence is an anomaly, and as such, they needed to emit a sound that felt relatable to the glitches that Mio and Zoe traverse throughout the game. I wanted to give them a sound that pulled off the balance act of sounding both designed with intent, while at the same time coming across as stochastic and irregular.

I wanted to give them a sound that pulled off the balance act of sounding both designed with intent, while at the same time coming across as stochastic and irregular.

What I ended up doing was having them emanate music drawn from the actual world they lead into, heard in a diegetic manner with the portal as a spatialized source. This music would be heavily processed with custom DSP plugins that we built for Wwise, such as a Bitcrusher and also a very cool Stutter-effect that would chop up and quickly repeat random parts of the audio stream.

Additionally, when the players start to draw close to the portal, all sounds originating in their environment are gradually filtered in the speakers of their side of the screen, eventually becoming almost fully muffled as they stand directly in front of the portal. The effect is a bold shift to the sonic landscape — the simulation imprisoning Mio and Zoe distorting in the face of this unforeseen portal!

when the players start to draw close to the portal, all sounds originating in their environment are gradually filtered in the speakers of their side of the screen

A fantastic part of creating the sounds of the portal was the collaboration I had with Pål Baglo, our outsourcing-partnered sound designer. Pål had previously done some work for the portal and the glitches as they appear within cinematics, and as I was working on their implementation in gameplay, we were able to successfully begin a very fun process where I would first incorporate his sounds in my designs, after which he would take that result and feed it back into his new designs for cinematics, and around we went! This approach nurtures coherence in the soundscape between cinematics and gameplay.

What was your favorite realm to design? What went into your sound work on it?

Gustav Landerholm (GL): It’s hard to pick a favourite; there are so many unique settings in Split Fiction! But the soaking-wet, gritty city of Neon Revenge was an absolute treat to work with. Early on in the production, we knew we wanted it to sound like Blade Runner, with a heavy focus on ambience, especially thunder and rain, and what that rain interacted with.

I spent a fair amount of time capturing rain on different surfaces, such as metal railings, hollow metal, windshields, window panes, solid metal, concrete, gutters, containers, wood, etc. — all of it to create a rich palette of sounds to pick from when designing the ambience.

In the implementation stage, the majority of the rain sounds were emitter-based to make them very cinematic and directional in ambient traversal moments. In more combat-heavy situations, we relied on baked rain in the global exterior quad ambience.

When recording the rain and thunder, I utilised LOM Usi Pro Omni microphones in a broad A-B configuration, which yielded a very lovely width to the rain and captured fantastic slaps and rolls from the thunderclaps.

An area that sells this well is the alley that Mio and Zoe fall into after the highway combat. It’s a real feast for the ears just standing in that alley and soaking it all in.

FJ: I need to echo Gustav’s point on it being very difficult to pick a favourite. One of the most enjoyable aspects of working on Split Fiction was that we did sound design for both a sci-fi and fantasy game at the same time with their own unique sonic identities.

Breathing life into both our realms through ambience sound design was a true joy.

Breathing life into both our realms through ambience sound design was a true joy. You have this blank canvas in front of you when starting fresh in a level that has no audio implemented yet and it’s so satisfying hearing it coming alive. I remember jumping on the Rise of the Dragon Realm chapter and designing weird bird sounds that were both used as randomized spots and baked in our quads for those levels.

Recording Split Fiction game sound effects (poor robot!)

A previous sound designer at Hazelight, Anne-Sophie Mongeau, prior to leaving had created a construction kit for our in-house sound library with fantasy creatures. Her sounds were a great place to start. I also believe Joakim (JES) used some of them as building blocks for the dragon companions. I also used some of my recordings of chickens that my mother had at her home during this time of production, which I pitched and stretched in all sorts of ways. Some of those were also layered into parts of the Megalith’s vocalizations. It makes me happy knowing my mother’s old chickens and hens are part of our soundscape.

Another fun ambience task was the commercial billboards seen in Neon Revenge. I got to write taglines for our made-up brands, translate them into languages spoken by colleagues at Hazelight, and record my colleagues selling the Cosmic Burger, for example, in Norwegian, Polish, and Hungarian, to name a few.

There are so many cool weapons and gadgets in the game, like the gravity whip and gravity sword, the sparkly pig farts (ha!), laser assult rifle, laser bazooka, etc. What were some of your favorite weapons, gadgets, or abilities to design? What went into their sounds?

PE: I want to talk about the gravity bikes. Making sounds for those bikes was a lot of fun but I admit that I was pretty damn scared of the task before I started. My worries both stemmed from the technical challenge, since one of the most difficult tasks within interactive sound design is vehicles. Also, I needed to make them sound as good as all our references: the Bat-bike, Tron-bikes, and so on. One saying that helped me start this task in the right way was the wise words from Joakim (JES): “It needs to just sound before it can sound good.”

I set off by trying to find the most interesting motorcycle sounds in our internal audio library. I wanted a sound that already had a pretty futuristic, clean feeling, but I did not limit my searches to motorcycle sounds; a lot of other vehicles and other weird sounds were also in the mix.

I did not limit my searches to motorcycle sounds; a lot of other vehicles and other weird sounds were also in the mix.

I started in Reaper with a pretty simple ramp that had the right kind of energy and the right feel of speed from a motorcycle sound. At this point, I had not decided for myself if we needed to go down the route of implementing the sound in the game with granular vehicle playback systems or not. There was no need for that yet, since I did not know what I wanted the bikes to sound like at this point. I had, though, sat down with the programmer and the designer in charge of the bike driving feature to understand what you could do and could not do with the bikes.

Then I went on with my exploration in Reaper. Pretty early, I found that many of my attempts at adding different processing via plugins just removed a lot of the noisy aggression in the sound and made it too smooth and weird sounding.

I […] split the signal to a few instances of Little AlterBoy to do the pitching in various steps

After that point, I focused my attempts on trying to maintain the bike sound and the noise, yet make it sound more sci-fi and less like an ordinary bike from our world. Then I tried out some pitch shifting, using the plugin Little AlterBoy from Soundtoys and got some pretty nice enhanced whining sounds when pitching the sound up but still maintaining quite a lot of the normal sound at the same time. I dove a little bit deeper into this method and split the signal to a few instances of Little AlterBoy to do the pitching in various steps and be able to blend all of that together and even automate it throughout the acceleration ramp. This yielded some desirable results, and with the automation, the vehicle sound still had a lot of story and variation. I also added a bunch of different distortions to the various pitched-up versions to glue them and create more overtones and a nicer whine.

At this point, I was pretty happy, but I only had one single acceleration ramp. This was a big limitation when it came to actually being able to make the whole range of sounds the bikes needed. So I went to the best place to source vehicle sounds: Pole Position Productions. I searched all of their bike libraries on their website and found at least three bikes that sounded similar to my current acceleration ramp, as well as a few more bike libraries that I could use to build all of the bikes that we needed. Then I called up my good friend Max Lachmann from Pole Position Productions and the next day I had a bunch of new sources to work with.

We did not use any specialized tools, such as REV, to make these bikes interactive.

I adapted my processing to the new source, designed all the sounds that were needed, and then started to build the implementation of the bikes in the game.

We did not use any specialized tools, such as REV, to make these bikes interactive. Due to the reason of our game being so linear and the bike behaviour being pretty simple, it was actually easier to just do it with a bunch of oneshots and loops, and to play these assets at the right moment.

Perhaps I am making it sound a bit simpler than the real implementation actually is, but you can come a very long way by using simple methods the right way instead of overcomplicating things and possibly affecting the end sound in a bad way, even though you are using very advanced tools. Another saying that we use within the Hazelight Audio Team: “The players will never see your fancy implementations or tools; they will just listen to what’s being played through the speakers and judge the sound that they hear. If it works, it works.”

GL: One of the most fun weapons I got to design came from the Dentist Nightmare bossfight, and that was the biting dentures! They posed a really fun challenge in that they needed to be both frightening and comical. They needed to pose a threat to the players but still suit the silliness of the situation. They also needed to feel satisfying to take control of, with varying gameplay styles. (Those teeth can be button-mashed like crazy, and I dare say that the sounds still work even if you are the fastest masher of them all!)

I ended up beefing up an actual wind-up toy of clattering teeth with mechanical ratcheting and groaning sounds alongside dog barks and pug yelps. I think they got a dimension of being “alive” and not just a weapon.

FJ: Haha, thanks for the pig mention! It was a breath of fresh air of sorts working on Pig World as we jumped on it quite late in production after having worked on levels with way heavier themes. So the silliness we got to play around with here was super fun. It was a hilarious level to sound design.

The farting pig abilities were built around recordings of real human farts (obviously) and layered with fire bursts, explosions, and flamethrowers to really nail the important explosive power of that pig’s belly. I mean it isn’t just simply passing gas, right?

For the slinky pig, I used recordings of bounce pads, slinky toys, and wire twangs from electric guitars. Then both the slinky and farting pig sounds were blended with magical chimes that were basically just granular wind chimes. The magical chime layer tied both pig abilities together nicely and, to my belief, dialed down the silliness of them (at least slightly). It was also a good space to use magical chimes in particular since we had decided to not have those sorts of sounds being part of the main fantasy realm, as Philip mentioned before.
 

What went into the sound for the ‘Parking Attendant’ in the Neon Revenge realm?

PE: All our bosses are so immensely packed with sound, but the hardest thing wasn’t making all of the assets (even though that was, of course, a large task too). The hardest thing was mixing the boss fights.

To help minimize this potential cacophony, we created a system that panned spatial sounds to the correct player.

The difficulty was trying not to make the mix a cacophony with all the sound effects, music, and voices on top of that. This was the single most difficult task for me in terms of game sound. To help minimize this potential cacophony, we created a system that panned spatial sounds to the correct player. This helped out a lot to create a bit more space and ownership in the mix for all the spatial sounds since our two players, especially during boss fights, have very different focuses and are in different locations.

For the Parking Attendant, I felt we got all of the pieces in place. First of all, the character of the Parking Attendant is hilarious. The writing, the voice acting, the voice processing (which was done by Johannes Karlebo (VO Designer), and also the implementation of the voice work so well.

Also, the music composition creates such a high-paced action feel, yet at the same time, leaves a lot of space for all of the crashing cars, lasers, and other weird things that are happening during the boss fight.

We split up the work in a way where Joakim (JES) made all of the sounds for the Ball and I took care of all of the sounds that happened or belonged to the platform

Joakim (JES) and I actually made all the sound effects for the Parking Attendant boss fight. We split up the work in a way where Joakim (JES) made all of the sounds for the Ball and I took care of all of the sounds that happened or belonged to the platform, with a few exceptions as always. The great thing about working together with a sound designer such as Joakim (JES) for 6 years is that we can say only a few words to each other to describe the inspiration or the style (or sometimes not even that) and then we get to it. I feel like we usually make sounds that fit together and sounds that follow a unified audio direction.

JES: Such a fun boss to make sounds for! I remember starting out by making an absolutely massive laser sound used in the initial chase sequence, then adding the cool motions and movements performed by the eye and its laser later in the fight. I really enjoyed pulling these sounds out of my soft-synths!

It’s also truly incredible the amount of character that VO can infuse into a large metal ball. While Philip and I were working on our sounds, Viktor Israelsson (Senior Technical VO Designer) was implementing the dialogue for the boss. When we started to hear the amazing personality and tone that he had hit for this truly unique encounter, we instantly felt that we wanted it to be at the forefront of the experience, making sure to have plenty of space for it in the mix, which was easier said than done with regards to just how wild of a scenario this part of the game is.

Add Gustaf’s absolutely gnarly music on top and I truly feel that the result is one of the many highlights of the game.

The music is such an important part of this game’s sound. It bridges the gap between the different sounds in the split screens so well, making the game feel whole (instead of split-screenish). Can you talk about mixing the music and the effects to achieve this connected feeling in a split-screen co-op game?

PE: I agree with your statement. Music is the collective motion forward through the game for both players, and it acts as a red thread through all of our realms and chapters. Even if the music changes in character and instrumentation, there is always something in the music that you can latch on to and that will hold your hand and move you forward through the game.

Music is the collective motion forward through the game for both players, and it acts as a red thread through all of our realms and chapters.

We didn’t experiment with splitting the music between the players, but that has been a deliberate decision. We want the two players to share the experience and put them in the same emotional feeling instead of two different emotions. I have been preaching a few things to the audio team. One of those things is that we really need red threads in our game mix since our games are so varied and messy. So, for example, character sounds such as movement and voices, and also the music, can act as a guide and a reference to all the other madness. I strongly believe that we need to have a solid ground to stand on and a clear reference throughout the game as listeners.

Even though the characters have different personalities and attitudes throughout the game, they are experiencing a shared journey[…] This enables the music to serve not just as audio glue, but also keep the emotional journey coherent.

Gustaf Grefberg (GG): Even though the characters have different personalities and attitudes throughout the game, they are experiencing a shared journey. Whenever there’s a dialogue, cutscene, or gameplay event happening it conveys their shared story. This enables the music to serve not just as audio glue, but also keep the emotional journey coherent.

We could have gone down the path of separate stems and all kinds of things but it would have overcomplicated a lot of things and not added much benefit. Of course, for those of you who have reached the end part of the game, there are a couple of unique music structures where we separate sci-fi from fantasy and pan them in different ways depending on where the players are located on the screen. This is where a more advanced music mix system made sense, both in terms of art and narration.

What were some other challenges you faced while mixing the game? Or, could you talk about your overall approach to mixing two different simultaneous perspectives?

PE: Mixing a game such as Split Fiction is a challenge, but a really fun one. An interesting aspect is that we need two players and a minimum of four ears to mix a game like this. It is simply not possible to perform the mix as it is intended without playing it together with another person. Of course, we pre-mix our individual sounds by ourselves, but to test those sounds properly, we need to play the game as it is meant to be played. It is also very important to play as both characters, because it can be really hard to stay objective and you can easily start favouring the sounds of your half of the screen. We as sound designers need to understand both sides of the game equally well. Due to these reasons, even mixing a game like Split Fiction becomes a true co-op experience.

Split Fiction: Mixing split screen game audio – here’s what it sounds like:

even mixing a game like ‘Split Fiction’ becomes a true co-op experience.

JES: Just as Philip says, one cannot overstate the challenge of mixing audio for a game like this one. Representing two perspectives through one shared output is a paradox, and one wouldn’t be wrong for saying that it cannot fully be done.

The truth is that we have to look beyond conventional game mixing techniques. In a single-listener scenario one is often able to make many assumptions about player perspectives and listener-emitter relationships and leverage this to approach mixing more deterministically. The mix of Split Fiction is a much more volatile subject, where two listener perspectives are constantly battling for each player’s attention.

We approach every scene with a very holistic mindset, built around the idea of the ‘Sonic Real-Estate’

The way we carve an understandable mix out of this storm of sounds is by combining technology in the form of a specialized implementation of low-level attenuation rulesets and spatialization rulesets in Wwise with methodically designed sound content. We approach every scene with a very holistic mindset, built around the idea of the Sonic Real-Estate — the screen is a canvas with finite space, and both players’ perspective must be given its due space within this room. We play the right sound for the right player by using conscious panning techniques with customized spatialization and speaker distribution.

What were some of the technical challenges you had in creating the sound of Split Fiction? What were some of your novel solutions for handling those challenges?

JES: As established, the single greatest challenge of creating audio for a Hazelight split-screen game is the fact that we have two listeners for audio, which has many complicated knock-on effects. It influences every single sound we play, and every single decision we make as designers. Providing answers to all the intricate situations that this complex setup presents is far from trivial, but by expanding upon many of the ideas around spatialization that we first put into play during It Takes Two we were able to craft a very exciting sound experience inside of Split Fiction.

the rapid changes in gameplay scenarios come with a myriad of different and unique camera perspectives, which presents its own challenges in providing the player with a lens through which to view the world.

Another aspect of Split Fiction is that the rapid changes in gameplay scenarios come with a myriad of different and unique camera perspectives, which presents its own challenges in providing the player with a lens through which to view the world. The perceived distance to the player as seen from the view of the screen is constantly changing, which means that we needed to build a system that would allow us to have any sounds be spatialized and made believable for any custom camera perspective demanded by gameplay.

The way we solved this was by building a backend in Unreal, dubbed Proxy Emitters. This allowed us to re-route sounds playing on an emitter to a duplicated source, which, by using auxiliary routing inside of Wwise, allowed us to create spatialized submixes containing all of the sounds playing on the original emitter. This turned out to have many useful applications throughout our game, but more than anything, we used it to dynamically mix, spatialize, and process player sounds based on their immediate relation to the camera and the player’s view.

GK: Not to sound like a broken record but having more than one listener is the biggest challenge since we need to adapt every part of our tech AND existing tech that many others take for granted.

For instance, we always assign ownership of the left or right side of the mix to one of the players, but for certain sounds, we still want to have it spatialized. This was realized by weighting the mix for the closest or specific listener for that specific sound or object, thus creating an audible perceived ownership to either the left or the right side. We termed this Spatial Panning Gain.

we always assign ownership of the left or right side of the mix to one of the players, but for certain sounds, we still want to have it spatialized.

Not all features are about new tech, but rather making it easier to use the tools in the toolbox, such as what we called Runtime Effects and BusMixers. Runtime effects are simply Wwise sharesets with a specific DSP plugin with a duration and design set in Unreal instead of Wwise, making it possible to sound design a specific behaviour for any part of the game with only needing a start and a stop.

BusMixers are another such concept where we don’t want to leave our canvas in game but continue to mix by reading or reacting to what’s happening on the screen. Their job is essentially the same as States in Wwise, applying, for example, volume changes to a specific ActorMixer or Bus. But, they can handle their own lifetime and clean up. The designer decides whether to keep it for a set duration or control it themselves by gameplay information without any pre-setup using the already existing hierarchy in Wwise.

PE: There was at least one pretty big technical challenge that we created for ourselves. During so many discussions, Joakim (JES) told me, “No, we should not do this.” But I did not listen to him. That challenge was to support Object Audio/3D Audio. The Hazelight Audio Team has never really let any obstacle be too big or too difficult, so we did not want to give up this time either.

The fact that just making a game sound good with two listeners has been a big challenge for us from the start, but we have never viewed it as an obstacle, more as an opportunity.

The fact that just making a game sound good with two listeners has been a big challenge for us from the start, but we have never viewed it as an obstacle, more as an opportunity. We have always aimed to make our split-screen games sound as good as any other game.

So, with games supporting Object Audio becoming more and more standard, we also wanted to be able to provide our players with the most immersive mix possible. In the end, that’s what we managed to do! Don’t quote me on this but maybe as the first split-screen game ever to do so.
 

Split Fiction uses Unreal Engine 5. Was this a good fit for the sound team? And, can you talk about your use of middleware?

JES: As previously mentioned, both It Takes Two and Split Fiction leverage Wwise for audio playback and processing. Both of these titles come with massively varied and unique gameplay scenarios, which, in extension, also means that we have to design, implement, and maintain large amounts of sound assets.

we have consciously progressed towards moving almost all conditional audio logic out of Wwise, and instead, designing these behaviours inside our SoundDef systems inside of Unreal Engine.

In crafting audio for these games, we have consciously progressed towards moving almost all conditional audio logic out of Wwise, and instead, designing these behaviours inside our SoundDef systems inside of Unreal Engine. The reasoning behind this stems from our need for very situational control of gameplay sounds based on whatever unique scenario is currently unfolding on the screen, and each audio designer is very empowered in their creation of complex sound implementations thanks to the large degree of control the custom workflows built in-engine offers them. The return on investment of the time spent expanding our core systems in both Unreal and Wwise has been massive for us.

GK: As Joakim mentioned, we use both Unreal Engine 5 and Wwise, but one can’t say it just works out of the box. We have to customize or fix most things to suit our splitscreen games.

We want every submix instance in Wwise to be its own entity and not affect its siblings or source of origin.

An example with Wwise is that I’ve changed how it routes signal. We want every submix instance in Wwise to be its own entity and not affect its siblings or source of origin. If you have more than one listener and use listener relative routing, the signal is increased by 6dB, which is not such a nice thing when the number of listeners can be dynamic and if not fixed, creates a very unbalanced and hard to work with mix.
 

What are you most proud of in terms of your sound work on Split Fiction? What do you feel was your biggest personal achievement in making the game’s sound?

JES: To me, the most fascinating aspect of Split Fiction is the mix itself. The game is absolutely full to the brim of spectacular sounds covering all manners of archetypes, from soft ambiences and critters to earth-shattering events, and having them all come alive and be represented in two listener perspectives is no small task. Even though every member of our team has lived inside of this soundspace for years, we’ll still come to new realizations around small nuances of this very volatile mix. It is truly unique in its presentation!

The game is absolutely full to the brim of spectacular sounds covering all manners of archetypes […] and having them all come alive and be represented in two listener perspectives is no small task.

Above all, I am beyond proud in regards to the extreme work and dedication every single audio team member has vested into this title. Each designer has left their indisputable mark, and I am so grateful for everyone’s work. There were many moments during production where our task felt absolutely insurmountable, the amount of content we had left to create versus the timeline we were on was simply too overbearing. The fact that we, together, were able to push through and make real our belief that this game could be something great is cathartic to say the least!

GK: Hmm, this is a difficult one for me. Since I’m more of a facilitator of sorts, delivering needed functionality, fast workflows, and DSP plugins. I don’t create sounds; I merely help my amazing team to do more of their already great work by shaving off seconds, minutes, or hours of wait time, or perhaps eliminating tedious tasks with automation or changes to the sound engine and making it possible to actualize their amazing sounds and designs.

I take great pride in increasing their efficiency and overall workflow by changing a few lines of code.

I take great pride in increasing their efficiency and overall workflow by changing a few lines of code. Just as an example of one of those features that both affects the workflow speed (from sound design to in-game) and soundscape is how I’ve added the functionality of dynamically creating RTPCs (Real-Time Parameter Controls) at runtime.

We call it Node Properties; any bus, ActorMixer, or sound source in Wwise is a node, and we can change a selection of its properties (some that only we have). Without any pre-setup, the designer can just jump into their SoundDef, play their sound/music, and, for instance, modify pitch/Qfactor/volume on their selected node.

GL: I think for me, the biggest impact I had on the game’s sound didn’t come from a singular sound or a particular design realm per se, but rather my informal role of audio scrum master during the production. A specific instance where this was clear was side-stories. At the end of the production, we had one week of sound design per side story, starting from scratch, which was very tight given the sheer amount of content within each side-story. That included: ambiences, bosses, abilities, puzzles, events, interactables, and destruction.

If it were any other team facing this David-versus-Goliath moment, they would have failed, spectacularly so!

Working from Philip’s roadmap, I delegated tasks to everyone in our audio strike teams, weighing each person’s sound design strengths so that we utilized everyone to their best abilities and minimized potential pitfalls, since there was zero time for iterations. Everyone had to essentially nail it on the first try. Our mantra, as mentioned, was: “First pass only pass.”

Making sure that everyone knew what to do, where and when, and in what order was key. I was a point of contact when questions arose. That helped streamline the entire sprint and make sure that there weren’t too many chefs in the same pot, which would have slowed us down and been devastating to the soundscape of each side story.

If it were any other team facing this David-versus-Goliath moment, they would have failed, spectacularly so! But all of us were “locked in” and knew what needed to be done. Our team had a work ethic that would put even the most senior developers in the dust, and for that, I hold our team in the highest regard.

FJ: Honestly, I don’t think I can mention a singular sound that I’m most proud of making for this game. There’s just sooo much content all of us in the audio team put together to ship Split Fiction in the state we did. That we managed to get sound in for every level was something we actually didn’t think we would be able to pull off as production was coming to an end. So, thinking back on it, that fact also makes it kind of difficult to even remember all the sounds I have made.

We could have ended up in a situation where we compromised a lot on our vision for the audio experience of this game and shipped something we would not have been as happy and proud of. But that’s not the case.

what I am most proud of is how we all collectively supported each other without a doubt, kept our spirits up high, and had fun even when things were super hectic

So, what I am most proud of is how we all collectively supported each other without a doubt, kept our spirits up high, and had fun even when things were super hectic, and that we managed to ace it in the end. I think, as Gustav (GL) mentioned, if it would have been any other team stepping into our shoes, they would have failed. I don’t want to be cheesy, but we have some sort of psychological kinship in our team that I have never experienced at any previous workplace, and that is such a strength for us. Perhaps then my personal achievement would be my contributions to this collective effort, focused on professional work and team health.

GG: I’d say it was to pull off creating the energy the game asked for. In mine and Jonatan Järpehag‘s (External Co-Composer) case, it was the music, of course, from the moody, smooth tones of fantasy to the wide atmospheres and high-action of the sci-fi. But really, the real high-point is the entire team effort to convey the vast dynamic and variation of the game, from the chirping birds, waterfalls, and soft orchestral music of “Hopes of Spring” to the non-stop action of certain segments of “Final Dawn.”

It’s not just a lot of work but also about co-creation and coordination to keep the audio alive and fresh.

A game that constantly creates “newness” in terms of all aspects was a beast to pull off, with almost unreasonable amounts of content, special cases, transitions, and dynamic mixes. It’s not just a lot of work but also about co-creation and coordination to keep the audio alive and fresh.

PE: I am really fucking proud that we managed to make it so that all of Split Fiction had sound in the end. This was a very difficult project to work on. The time before we shipped the game was incredibly busy. At some point, I truly thought that some of the side stories would consist mostly of music, since we had to cut as many corners as possible.

So I guess one of my own biggest achievements is to have made some sort of plan that, in the end, worked out. But to be honest, it was not the plan. It was due to our fantastic team that we managed to finish it!

I am most thankful for and most proud of my awesome colleagues and the efforts they put in.

I am also very proud to have been an audio director for a sci-fi and fantasy game and to have made those genres sound like I wanted them to sound for this story. I am most thankful for and most proud of my awesome colleagues and the efforts they put in. Everyone has pushed and worked so hard and everyone really wanted to make this game sound awesome. Everyone grew so much during this project and it would not have sounded like it does without every single one of them. They are the best and I thoroughly enjoyed working with them all on this crazy project!
 

A big thanks to Philip Eriksson, Joakim Enigk Sjöberg, Göran Kristiansson, Gustav Landerholm, Filip Järnmark, and Gustaf Grefberg for giving us a behind-the-scenes look at the sound of Split Fiction and to Jennifer Walden for the interview!

Please share this: