Joel Bille on crafting the voice of Fe

“The music has to follow the player, not the other way around.”

Today we’re honored to be joined by Joel Bille, the composer and sound designer for the exploration-based 3D platformer, Fe. If you haven’t heard of Fe, it’s an absolutely gorgeous game both visually and sonically – check out the Electronic Arts trailer:

In our interview, Joel discusses his process of crafting the sonic world of the game. The soundtrack can be heard in its entirety on Bandcamp.

Let’s start from the beginning – when did you start composing, and who were your influences?

Composing has been my main way of expressing music since I was little. I used to sit by the family piano and just play and improvise. Sometimes I accidentally played something I liked, and then that could turn into a written song. That’s sort of still my main form of playing music, even though it’s in a more refined way now. So in a way, I’ve been composing or at least writing songs for most of my life, even if it was very primitive in the beginning. I did my first game soundtrack about 8 years ago now. I was, and still am, very influenced by the old Nintendo folks, Koji Kondo, Masato Nakamura, and the likes, even though it doesn’t always shine through in the music. I love the way they work with melodies as main elements of the music.

Considering there’s no traditional dialogue in the game, how do other sonic elements communicate the narrative of the world of Fe?

The fact that there’s no traditional dialogue leaves the sound design and music in a special place. Music probably becomes the strongest parameter for setting the mood, even more so than in other games. But it’s also more exposed, and easily gets over-explicit. You can’t use music to have the player feel anything special unless it’s called for – otherwise it just gets corny and the player might lose their connection with the game. There’s a close relationship between the design and music to get the magic to actually happen.

The ambiences are also very important in setting the mood. The crickets, wind, creaking trees… it all builds up an atmosphere. Not only a sonic one, but an emotional one. It says something about the state of the forest – is it a lively and lush one, or a haunting one? Ambiences alone can make a lot of difference there.

The voice acting that’s actually in the game, consisting entirely of animal cries, is very important too. It’s in a way dialogue, but in a language incomprehensible to humans (and all other animals). It’s very important in setting the character of the various animals, and fills the same function as ordinary dialogue, only much, much, vaguer… But the different expressions the animals have are as close as we get to classic dialogue and the functions it usually serves.

I’m immediately drawn to the gorgeous string performances, but there are also some really inventive synthesized sounds in the music and sound design. How did you go about crafting these sounds?

There’s one recurring sound, most noticeable in ”De Tysta” on the soundtrack and in the background of the “Fe” track, but it’s there in pretty much all tracks. It’s actually sculpted from something as weird and simple as the built-in sampler in Logic, EXS24, with its ”no instrument” default sound, which is basically a sine tone, mapped and pitched over the keyboard. I’ve edited the ADSR into something more interesting and dynamic with a slow attack and lingering release, added distortion to reach some overtones, and had a series of LFOs controlling a filter that blends in and out for dynamic variation. There’s also various chorus, compressor, and EQ effects applied. That sound turned out to be very versatile – depending on how you play it, it can be very gritty and ominous to really gentle and welcoming. If you play more dense intervals, there’s lots of intermodulation going on and it gets really gritty. But if you play more wide and open chords, it sounds very warm and soft. The even more rough and bassy synth sounds are created with a Moog sub 37, which I also used in the sound design for some of the Silent Ones’ more noisy ‘tech’ sounds.

I’ve also recorded lots of bells and chimes and created a couple of sample libraries from that. The most frequently used is a set of bells that used to hang in my grandmother’s house. They were sorted and pitched pretty dramatically to fill a few octaves. The extreme pitches sound a bit weird when listened to closely, but dubbed with some other similar sounds and integrated in a mix, I think they sound great. These also reappear in many parts of the sound design.

I also want to mention the “Fracture” plugin from Glitchmachines, which is really useful for creating glitchy sounds. That one’s used all over the Seeker’s sound design, as well as in the Autechre-inspired Hunted-theme, also heard on the end of “De Tysta.”

Fe places a heavy emphasis on discovery and exploration. How do you feel the music and sound design complement these themes?

The fact that we were making a game focused on exploration was absolutely central to the music, but also partly to the sound design. Fe isn’t trying to be a cinematic game – we’ve always wanted the experience to feel like the player’s own, and thus we didn’t want to interfere too much with cool music cues. There are a few places in the game where we use more advanced music triggers to have movie-like musical transitions and effects, but they are few and well called for, and hopefully go by so smoothly that you don’t really notice them happening.

Generally, we’ve avoided the “sad music for sad scenes and happy for happy scenes” approach to using music. Instead, we tried to focus on how the music can affect the player’s relationship to the different areas in the game, rather than composing for specific situations. The forest’s theme for instance, heard in its plain form around the hub (or “Skogen” in the soundtrack), is meant to create a connection with that place, a feeling of safety and trust. It shifts slightly on different visits, but it’s always welcoming and embracing. The purpose is not mainly to have the player feel something particular while hearing the song, but rather to build on the emotional associations to the place, and thus the player’s relationship to the world of the game. In this sense, it’s very different from a game more focused on progression and cinematic narrative.

In general, I think it’s important that the music doesn’t try too much. We want the player to take their time and study their environment, and thus the music has to say that it’s okay to do that. So overall, the music has very little direction or drive. There are a few exceptions with passages in the game that demand more specific actions from the player (climbing the giant, the wolf-fight, the ending, etc). But in almost all of those cases, the player has the option to leave and go do something else, and in that case the music follows and goes down in intensity, to make sure it doesn’t poke the player in a direction they don’t want to go. The music has to follow the player, not the other way around.

You went on quite a few adventures recording ambiences and other sounds for Fe. Could you speak on how sound design decisions you made revolving ambiences and other subtle details contribute to the game’s overall aesthetic?

As a sound designer you tend to gather sounds over time by going on recording sessions and just bringing a recorder wherever you go, slowly building up a library. Some of the sounds mentioned in the EA post, like the Transylvanian insects, where recorded long before Fe was even thought of. Other recordings from trips such as our visit to Thailand and the Swedish island with the wind sounds were made with Fe specifically in mind during the development process.

Early in development, we talked about the general direction for the sound design. There was one idea to go with some sort of ‘low-poly’ audio, with synthesized bird sounds and a general digital feel to match the graphics. But I felt pretty strongly that I wanted to do the opposite – have more realistic sounding environments, filling out what the graphics leave to imagination. By being that serious about the ambiences and sound in general, I could contribute to making the world feel lush and complex. All ambiences in the game are 3D sounds with a location; every insect you hear has an actual location in a bush somewhere, as do frogs in ponds (they stop if you go too close), winds around cliffs, streams, waterfalls, etc. The idea is to make the player feel like they’re actually there, or more cynically put, it creates a stronger illusion. No matter how you see it, I believe this kind of approach makes the player feel more connected to the world, which is what Fe is all about. One more apparent element is how everything in the forest, from tiny flowers to big animals, answer to the player’s cries. There are around 10 different plants, without any actual gameplay function, that still answer with a little sound whenever they hear the player’s voice. Most plants have around 6 – 10 variations of bell-like tones, some using the bells mentioned earlier, tuned to the music of the game. In a way, they become the sound of the forest. And by using sounds similar to those in the music, they tie the music and sound design together, subtly enforcing the game’s theme of “everything is connected.”

What was working with voice artists Kristina Issa and Kellen Goff like in designing the game’s various creature sounds?

The voice recording process was very interesting. It was pretty similar to how you’d work with an ‘ordinary’ voice recording with spoken lines. There were three people in the studio – me as the sound designer, the voice actor (Kellen was on Skype, since there’s an ocean in between us), and the director, Hugo Bille. We often spent some time before the recording session finding the general voice of the animal. We gave the voice actor some material to get started – screenshots, animations, and our ideas of what we were looking for. We could say to Kellen, “think husky and boar, but blend in some goat.” We were looking for something that didn’t sound like anything you’d heard before, that was at the same time flexible and able to form a variety of expressions. Once we got the general voice of an animal, we’d start recording the actual phrases. Most animals have around 15 phrases, tied to various behaviors. Those could be something like ‘greet player,’ ‘escape,’ ‘surprised,’ etc. Each of those have around 10 variations.

It proved real difficult to express a variety of emotions with growls and snarls instead of words, while also remaining in character and keeping the voice of the animal. I do believe we really managed to get there, mostly thanks to our amazing voice actors, Kristina Issa and Kellen Goff. I actually do some of the animal sounds myself, for the record.

You can read more about the voice acting here.

What’s your favorite track that you composed for the game?

It’s really hard to say, but if I need to pick one… I’ll go with “Nästet part 2.” Both that song and the track “Fe” are actually difficult to talk about without spoiling the game. There are a lot of musical things happening in there that reflect various events and relationships of the game. It probably goes unnoticed by most, but my hope is that it subtly shines through, even if the effect is mostly on a subconscious level.

How were the sound design and composition for Fe different from your previous projects? How did you grow from working on the game?

The biggest difference is probably how Fe is so focused around audio. In most contexts, sound design is something that you notice only if it’s bad and annoys you. If it’s good, it melts in and go by unnoticed. With Fe, especially through the singing mechanics, I had the unique opportunity to take center stage, which is very unusual for a sound designer and very different from my previous projects. The music of Fe is also very different from my previous work with Zoink. It’s more honest and emotional, and it has a core content beyond its aesthetics: trying to touch something subtle yet sublime. That puts the music in a really special place, and I had to find a way to create the connections I mentioned earlier. To make it more understandable for myself, I would often bring that ‘sublime’ idea down to something concrete, maybe an experience from my own life that speaks to the feeling, and from there take it back to something more subtle.

The development of Fe went on for more than a couple of years, and I’ve developed a lot as a sound designer over the course of the project. I’ve just had loads and loads of experience and time to develop my skills, and got to learn what works and what doesn’t. I’ve also noticed how sound can have an actual impact on the general experience, as I talked about earlier, and not just be something in the background.

I’ve also become more and more interested in having strong core values in artistic projects. Sometime games can have a quite shallow idea from the start – a graphical concept, an idea for a gameplay, etc. But, while working with Fe, I’ve started to think that projects that have a strong core, not just games, will come off a lot stronger as works of art. And it helps a lot to have a core to go back to whenever you lose direction. So I’m all into that with whatever project that comes next – to take some time with the director or team to understand what we’re doing and why, and to find some kind of mantra. It’s not unlike how marketing works with finding a way to talk about the game, but with a more internal focus.

Anything else you’d like to add?

I just want to mention the interactive voice system, since I think it’s probably the biggest accomplishment of Fe’s sound design. If you’ve played the game, you know that you can sing or talk dynamically, either by using the analogue R2 button on the PS4, or tilting the controller on the Switch. In the case of the default Fe-voice, it starts as a mumbling, and gradually and nearly seamlessly makes its way up to a clean singing tone.

Each voice consists of 4 to 7 looping phrases of around 8 seconds, assigned to a specific part of the input scale. Some loops are also set to only play on the way down or up, just to make things more complicated. So, each sound is shifting in volume and pitch according to input, and then crossfades into the next loop when you get to that part of the input scale. To make that crossfade sound smooth, it’s crucial that the pitches of the two sounds are exactly the same at the time of the fade. That’s where the tricky part comes in. My brother and creative director Hugo Bille was able to write a tool that through various pitching mechanisms and offsets enabled us to find that sweet spot where the first loop has the exact same pitch as the one we’re crossfading into.

On top of all the technical stuff, there was another real challenge in recording loops that was both expressive and preferably somewhat musical, while sounding good with the dynamic system. And trying to explain this system to the voice actors… I don’t think I’ve ever heard anything similar in a game before (though I believe the motor sounds for racing games would do something similar) and I’m really proud that we actually made that work.

Fe is available on the Nintendo Switch, Xbox One, PS4, and Origin for PC and Mac.

March 19, 2018

Harrison Shimazu

Harrison Shimazu is a composer, content strategist, and writer who’s passionate about democratizing music creation and education. He leads the Splice blog and produces vocaloid music as Namaboku.