Behind the Captivating Sound of ‘Clair Obscur: Expedition 33’ – Huge Interview with the Sandfall Interactive Sound Team!

Since its release in April, the turn-based role-playing game Clair Obscur: Expedition 33 — developed by Sandfall Interactive and published by Kepler Interactive — has won the devotion of gamers and earned near-perfect scores from game critics. Set in a French “Belle Époque”-esque dark fantasy world, the game transports players to a nightmarish yet stunning land where expeditioners seek to defeat the mysterious Paintress before she erases humanity.

Here, the sound team — Raphaël Joffres (Audio Lead & Music Designer), Maxance Playez (Principal Sound Designer), Olivier Penchenier (Technical & Cinematic Sound Designer), Alexandre De Mester (Junior Sound Designer), Théo Clavel (Junior Sound Designer), Charlie Briere (Cinematic Sound Design Intern), and Victor Desrat (Music Design Intern) — talk about everything audio, like their inspiration for sound, using dubstep and drum ‘n’ bass sounds for attacks, creating coherent and flexible language for the Gestrals, creating processing chains to link characters together, designing a variety of unique, fantasy-based locations, pouring tons of hours and effort into the cinematics, building the interactive music system, breath system, and dynamic mixing systems, analyzing the game signal in real time directly in VSTs via OBS, and much, much more! You don’t want to miss this deep dive!

Clair Obscur: Expedition 33 – Launch Trailer

What were some sonic inspirations for Clair Obscur: Expedition 33? What was the overall direction for the game’s sound? Were there sounds or types of sounds you wanted to avoid?

Maxance Playez (MP): For Clair Obscur: Expedition 33, one of my biggest sonic inspirations was the anime Fire Force. I was particularly influenced by the bold sound design approach of Yasumasa Koyama, under the sound direction of Jin Aketagawa. In Fire Force, attacks are represented with heavily saturated, distorted bass sounds that feel almost industrial or sci-fi. That unexpected, stylized approach inspired me to explore similar territory, using distorted, musical basses and wobbles, especially in counters and some of Lune’s skills.

Electronic music genres like dubstep and drum ‘n’ bass also had a strong influence on my work.

Electronic music genres like dubstep and drum ‘n’ bass also had a strong influence on my work. From reese basses to wobbles, I incorporated these elements directly into the sound design. Even in the mixing stage, I borrowed from those genres, using heavy compression, saturation, and dense layering for character skills and powerful counters.

Lastly, I wanted to blend the aesthetic of classic JRPG sound design with something more modern and Western. For example, buffs and debuffs often use lighter, more traditional JRPG-style sounds, which creates a contrast with the more aggressive and textured combat audio.

We wanted each attack feel impactful and satisfying, with a strong artistic identity across the entire team.

From the beginning of the project, we defined several key pillars for the game’s overall sound direction. The first was environmental ambiences. Exploration is a quiet, contemplative moment without HUD or UI, so immersion relies heavily on sound. We aimed to make each environment feel alive and support the strength of the game’s visual direction.

The second pillar was the character attacks. We wanted each attack feel impactful and satisfying, with a strong artistic identity across the entire team.

The third pillar was the voice design of the Nevrons. Each biome features creatures with distinct textures and vocal identities.

The fourth pillar was the cinematics, which play a central role in the story. We made sure that all the sound design work we had done, from ambiences to combat to voices, was properly reflected and elevated in those key moments.
 

Clair Obscur: Expedition 33 | Behind the Music

There is a ton of beautiful music in Clair Obscur: Expedition 33. Can you talk about your collaboration with composer Lorien Testard?

Raphaël Joffres (RJ): When I joined the project, about two years before the release, Lorien already had composed a big part of the OST. So it felt a bit like having a mountain of tracks to put in the game!

The first challenge was technical, as all the music systems needed to be properly designed and developed before we could put any new assets in the game. So I spent the first month or two building the basis for an interactive music system. Unreal Engine’s MetaSounds are really powerful, but you have to create everything from the ground up!

all the music systems needed to be properly designed and developed before we could put any new assets in the game

Once I had the toolbox I needed, I began to discuss with Lorien and Guillaume Broche, the creative director. They worked very closely together on the OST, and I wanted to know what overall music design we would need. We agreed on a pretty simple structure, as the game is mainly divided between exploration, combat, and cinematic. The idea was to have a music design as rationalized as possible, to fast-track the implementation process, and make it robust and production-ready. I worked with Florian Torres, a programmer from the team, to develop a Music Manager, a kind of master system that would handle all the conflicts and transitions between tracks, and I created a short set of MetaSounds presets for each music scenario. The ground rules were there; the systems were there, so I could finally start scaling the mountain track by track!

We quickly developed a good process with Lorien and Guillaume. For each level, we would talk about the music flow, meaning which track starts when, how it’s evolving, when it’s transitioning to another track, and how battle tracks are spread across the enemies. I even created some kind of flow chart at first, to help the QA team test the music, but I quickly stopped after two levels, as it was too time-consuming. Then, Lorien would export all the stems for each track of the level, so I could rework their mix for the game and edit all the assets I needed. Once all the tracks were implemented, I always made a capture of the flow to get Lorien’s and Guillaume’s thoughts and refine if needed. Level by level, the 8 hours of music slowly but surely made their way into the game!

I worked with Florian Torres, a programmer from the team, to develop a Music Manager […] that would handle all the conflicts and transitions between tracks, and I created a short set of MetaSounds presets for each music scenario.

One thing we constantly had to consider was the space we gave to the voice. Alice’s voice is really beautiful, powerful, and gives a very distinct identity to Clair Obscur‘s soundtrack! But her voice could also take up a lot of room in the mix, not only in the music itself, but also in the overall sound of the game. We really wanted to avoid any kind of ear fatigue or boredom due to the voice. I made it less upfront in the in-game mixes of the tracks, especially during the battles, and tried to spread it across the exploration tracks, sometimes creating variations where the voice is more ambient, almost ethereal, or even remove the voice from some part of the tracks, so it alternates between singing and instrumental. This was a pretty tight balance to find.

In the end, it was quite a challenge to put that much music into the project, and I really thought there were too many tracks at first! But as I was implementing everything, discovering the range of the scope, the diversity of levels we had, and the diversity of Lorien’s music, it turned out everything had its place, and Lorien even continued to produce new tracks until the end of the game, which were really needed! (a special mention for “We Lost, “the battle theme of Simon, which was composed and implemented between the gold patch and the Day 1 patch…)

What went into the sounds for the Continent locations in the game? There are quite a few, so did you have a favorite location to design? What was the most challenging location to design?

Alexandre De Mester (ADM): Shaping Expedition 33‘s ambient soundscape was an appealing challenge. We tried to give each environment its proper identity while keeping an overall consistency with the game’s artistic direction. There were so many varied locations: cities, forests, underwater areas, caves, seashores… It really felt like a huge playground!

Lumière is the first location that you can freely explore. To design the crowd full of citizens, we recorded some of the Sandfall dev team in our studio. We basically asked them to have random conversations about life (just like any resident would have with their neighbor). These talks were in English, but we had to have them pronounced with a very nice French accent, of course! (a special mention to the Life Enjoyers…). We also recorded the same group of developers for the Gestral Village’s Market & Fighting Arena. No more English, they had to switch to our Gestral language!

To design the crowd full of citizens, we recorded some of the Sandfall dev team in our studio.

For the Continent’s vegetation, we had a lot of fun breaking old VHS tapes to gather a considerable amount of tape. We manipulated these tapes with various movement intensities through different microphones to gather as many textures and proximities as we could. These recordings helped a lot in the designing process of Spring Meadows, Ancient Sanctuary, Falling Leaves, Yellow Harvest, and all the Forest Locations. It is always nice to record your own sounds rather than picking from libraries.

We had to find the right sound for Flying Waters. You are wandering in an underwater world, but you are not swimming, and this nuance is something that can be translated through ambient sounds, so we iterated a lot on this one. Watery convolution reverbs were vital for the sounds of this location.

finding the right sound for the Paint was important since it is the game’s essence. We went for a mix of liquid, medium rumbling textures to get the main design.

Talking about watery textures, finding the right sound for the Paint was important since it is the game’s essence. We went for a mix of liquid, medium rumbling textures to get the main design. And after that, we could make a range of loops and one-shot sounds that can often be heard in abstract locations such as the Painting Workshops, Renoir’s Drafts, Sacred River, or Old Lumière.

There are two segments in the game that I really enjoy. The first one is the spot right before the Lampmaster Cave in the Stonewave Cliffs. I really like the rain and the thunder there. I feel it fits with the mood of the scene; you can tell that something is around the corner!

The other one is Visages. We used a lot of vocals from various sources to design the three emotions of the island: Anger, Sadness, and Joy. The final ascent before reaching the Axon is what I love the most. You cross the three emotions with your grappling hook, and the gigantic Masks are either screaming, crying, or laughing at you. Each transition is marked with a big, dark clock ringing; it is so cool! Visages embodies a very good balance and blend between the Nevrons’ voice designs and ambiences. It is probably one of my favorite zones, with the Camp and the landing in The Abyss.
 

What was the audio direction of the Gestrals?

Théo Clavel (TC): The Gestrals’ voices were such an interesting challenge. Without a lot of time and too many resources, we needed to create a coherent and flexible language that can be used for cinematic, in-game dialogue, and battle.

At the beginning of the research, I did multiple tests by recording my voice, which wasn’t very conclusive. It was closer to Animal Crossing‘s voices. After many iterations, we decided to do a small casting with all of Sandfall’s developers and told them to do some funny gibberish with a deep, slow voice. We wanted a lot of variety in voice tone and, above all, some funny sentences.

Then I wrote a document with all the words we experimented with, like “Noco Cafe Toro Fiso Poke Nalu,” and we used that as a word bible to help “voice actors” have some kind of consistent vocabulary. Afterwards, we chose 6 of the best devs from our casting to record all the generic sentences for each Gestral archetype (Tall, Big, Small, etc.). Recording developers allowed us to quickly record new lines for all of our needs. For each type, we recorded different emotions with several sentence lengths, like Short, Medium, and Long, for Sad, Joy, Angry, and Neutral to cover as many cases as possible. And of course, there are easter eggs in some of the phrases like “Karo Parfé,” which means a very nice petanque move.

Then I made a post process with a wooden convolution reverb, to emphasize the wooden nature of the Gestrals, and a little bit of vibrato that smoothly changes pitch to accentuate their goofy and funny tone.

Monoco is a special kind of Gestral. Can you explain his distinctiveness in terms of sounds?

RJ: Monoco is one of a kind! It’s the only Gestral who speaks English. He’s very old, has seen a lot of the world, but is very strong and likes to fight all the time, as a proper Gestral. Guillaume (Broche) wanted to give him a deep and charming voice, with snappy dialogues, and some pretty fun back and forth with Verso. But even if Monoco has a human voice and almost a human personality, he’s still a wooden creature with a mask (which was very useful for the cinematic, as we didn’t have to worry about lip sync, for once!)

Olivier Penchenier (OP): Monoco was a particularly exciting character to sound. The idea of imagining a more massive figure, equipped with complex paraphernalia to play with in sound design, is really galvanizing. In fact, I originally created Noco’s foleys in a cinematic — the little Gestral you discover after finding Maelle — and then reused some of these sounds for Monoco.

I used a MetaSound that triggers a wide variety of sounds simultaneously, with lots of randomization.

As for the footsteps, I used exactly the same recordings as for the other main characters, but made them heavier and slower to match his slower animation. I added a low layer on each step to reinforce its heaviness, and performed a slight time-stretch to accentuate this impression of slowness and mass.

I used a MetaSound that triggers a wide variety of sounds simultaneously, with lots of randomization. I separated the sound elements into two main families: the frame (everything attached to his body) and the backpack, which is played with a delay of about 0.4 seconds to simulate his more floating, independent movement.

In the game, each type of foley is structured into several movement categories: walk, run, sprint, and a more generic sound bank for other movements.

The frame sounds feature a variety of textures: metallic squeaks, rope stress, wood squeak, and a main layer made entirely of wood with a slight creak. This main layer is played at every footstep, while the others are triggered more randomly, to reinforce the impression of a unique movement each time.

The backpack layer, meanwhile, contains a main layer of shaking wood (for transported branches) and a layer of chains played randomly, to create variation.

In the game, each type of foley is structured into several movement categories: walk, run, sprint, and a more generic sound bank for other movements. This bank is organized according to two lengths (Short, Long) and three levels of intensity (Slow, Medium, Heavy), enabling me to adapt the foley to any animation.

In all, Monoco has just under 500 sounds for its foleys. I’m really pleased with the result; it really felt like the equipment is alive!

RJ: The final particularity of Monoco comes from his skills. Unlike the other Expeditioners, Monoco doesn’t have many skills of his own, but he has the power to transform into Nevrons and mimic their attacks.

Reusing the same assets as a playable character that’s closer to the camera and viewed from behind posed a lot of technical questions, especially as the Nevrons’ attacks were designed in mono, while Expeditioners’ attacks were in stereo.

The challenge there was more technical than a creative one. Monoco became fully playable pretty late in the production, and most of the Nevrons’ attacks were already implemented and working in the game. But they were tailored for a certain perspective and camera distance, to have a certain place in the mix. Reusing the same assets as a playable character that’s closer to the camera and viewed from behind posed a lot of technical questions, especially as the Nevrons’ attacks were designed in mono, while Expeditioners’ attacks were in stereo. But it was one of these situations where “wait and see” seemed the right thing to do.

So we waited until Monoco was fully testable in the game, and Maxance tested everything. And it turns out, it was working perfectly, so he didn’t have much to do except rework a few timings in some attack sequences that were modified especially for Monoco. Good surprises like that don’t happen very often in development!

Can you talk about the voice design of special characters, such as Esquie?

TC: For Esquie, it was quickly decided he would be voiced by Maxence Cazorla, the mocap actor behind Gustave and Verso. We asked him to try something between Baymax (from Big Hero 6) and Chien-Po (from the original Mulan), but much slower and more naive (or, like Guillaume said, “a bit high”). Maxence sent us some recordings he made on his phone, and I started to do some post-processing tests right away.

we recorded everything directly on site at Sandfall studio, which enabled us to test the post-processing and integration of the vocals into the engine very quickly

Once we’d found the right acting direction, we recorded everything directly on site at Sandfall studio, which enabled us to test the post-processing and integration of the vocals into the engine very quickly after the first recordings.

The voice of Esquie is time-stretched and low-pitched for a deep voice that’s even slower. The voice is also sent into another track with aquatic textures, bubbles, and whale recordings that modulate with an audio follower to vary according to the intensity of the voice, all with a vibrato that will slowly vary the pitch and will highlight the roundness of the character, to make him sound like a big and friendly talking balloon full of water.

OP: As another example, one day we were trying to figure out what to do for the voice of the Orphans — Nevrons who have a sort of straw basket for a head. And in the studio, we have big wicker baskets to hide cables in. I put a basket on my head as a joke, did some vocals, and the muffled aspect of the basket worked strangely well. So we kept it in the game as is!

What were your favorite enemy combatants to sound design, and what went into their sounds? What was the most challenging combatant to sound design, and what went into their sounds?

MP: For Nevrons’ sound design, we often used a biome-based approach. The idea was to treat all the creatures within a given zone as part of a shared sonic space, with subtle audio elements that tied them together. In biomes like Goblu or Spring Meadow, certain textures or processing styles appear across different Nevrons, helping to establish a distinct sonic identity for each biome.

The idea was to treat all the creatures within a given zone as part of a shared sonic space, with subtle audio elements that tied them together.

While each creature had its own characteristic sounds, we always looked for small sonic links between them to maintain cohesion. Sometimes we used the same processing style or similar effects, but applied to different source materials, which helped us create variation while still feeling unified.

Convolution reverbs played a big role in achieving that sense of unity. By treating Nevrons’ voices with similar impulse responses, we made them feel like they were really existing in the same acoustic world.

Sometimes we used the same processing style or similar effects, but applied to different source materials

At the same time, we aimed to optimize production by making certain sounds reusable across enemies. Many Nevrons share the same whooshes, landing sounds, or jump impacts, for example. The voice design, however, was always unique for each creature, and that’s where most of their personality came through. It was a great balance: reuse where we could, and focus on originality where it mattered most.

TC: My favorite enemy to sound design was the Lampmaster, the last boss of the first Act. The process was the same as for all the Nevrons. We started by taking as many visual references as possible: the texture, the size, the objects it wears, the shape it has, the animations it makes, all based on the work of the art team, to define what we needed to highlight with the sound.

The idea was to have as many characteristics as possible in the voice of the Nevrons, because it’s the sound you hear the most in battle and exploration.

All enemies are divided into three distinct stems: the voice, the attack sounds, and the foley. The idea was to have as many characteristics as possible in the voice of the Nevrons, because it’s the sound you hear the most in battle and exploration. Then with all of these textures we could create different types of screams, varying in length and intensity to create attack, pain, idle, spot, dash, or death screams.

The Lampmaster is characterised by a ball of limbs and hands, like an assembly of several bodies. So I wanted to emphasise its misshapen and disjointed aspect. That’s why we recorded all sorts of sounds with our mouths and cracking hand bones with the development team, which was a lot of fun and gave a very organic feeling.

For most of the Nevrons, I used SnapHeap from Kilohearts, which allowed me to vary several audio parameters with an audio follower like pitch formant, chorus, or any other kind of filter. We also used a lot of convolution reverbs with textures to give a more metallic sound to it, and some reverse reverb as well.

Recording bone cracks for The Lampmaster and Nevron sound design

The Lampmaster is made of several layers: in the lower midrange, there is throat and gutural design sound, in the midrange, there are several screams that characterize the Nevron, and in the high end, there are clicks and cracking bones. All of that makes a lot of variation for a very organic result. For the attacks and foley part, I highlighted the metal with chains, rusty swords, and the squeaking of lanterns, which make a good contrast with the voice design while underlining its key traits.

The most challenging enemies to design were the end bosses, like Renoir and The Curator, as their attacks are much more complex. So we had to find a good texture for the “void” attacks, because unlike fire or ice, we don’t know what void sounds like. It’s a mix of watery organic elements to evoke the paint, with a sort of lightning sharpness. I also wanted a metallic noise that tended towards gold, and I was wondering: “What would dark luminous paint sound like?”
 

What went into the sounds of the different character attack/dodge/parry/skill sounds? What were your favorite weapon sounds to design, and what went into them? What about your favorite skills to sound design?

MP: For the design of character attacks and skills, I started by establishing a strong, unified direction across all characters, while still giving each of them a distinct sonic identity. The common foundation was built from a mix of realistic weapon sounds, organic, paint-like textures (mostly liquid), and magical or light-based elements. This shared base helped create consistency while still allowing room for personal touches.
Each character then received a custom layer of identity:

• Sciel’s abilities include sharp, dry textures inspired by cards.

• Lune features wobble-heavy low-mid bass elements that enhance her mystical vibe.

• Maelle, who wields a rapier, has a more elegant, refined sound — almost like a whip.

• Gustave and Verso both use heavier, weighty sword sounds to match their brute-force style.

To bring cohesion, I relied heavily on processing. Every sound went through aggressive chains including multiband saturation, OTT compression, hard clipping, and more. This approach helped standardize the frequency content of the attacks, creating immediate cohesion while still preserving the identity of each character. I also made a lot of volume and low-cut automations directly on the master to create more dynamics, especially by adding short intervals of silence right before impacts.

I also made a lot of volume and low-cut automations directly on the master to create more dynamics, especially by adding short intervals of silence right before impacts.

My favorite sounds to design were definitely the counters and especially the Gradient Counters, because that’s where the more stylized, almost extra-diegetic direction took over. Those were the most satisfying to work on, both technically and artistically, because they allowed for bold textures and impactful sound moments.

Maelle was probably my favorite character to design because she essentially set the tone for the rest of the skill direction. Along with Gustave, they were the first two characters I worked on, and they shaped the audio identity for the rest of the cast.
 

What went into the UI sounds in the game?

OP: For the menu’s sound design, I had two main intentions.

The first was to create a very dark atmosphere, directly in line with the “obscur” side of Clair Obscur. The menus are deliberately plunged into shadow, with an almost heavy feel. This aesthetic was very much inspired by games like Elden Ring or Souls-like games in general (reference fully assumed).

To reinforce the coherence of the whole, I used a common convolution reverb on all interface sounds. The idea was to create the illusion that everything takes place in the same sonic location

The second, more subtle intention was to evoke painting, but without making it a central element. It’s not an omnipresent sound motif, but rather a discreet touch, a visual echo integrated into the sound identity without ever dominating it. What I wanted to bring out was the atmosphere, rather than the materiality.

To reinforce the coherence of the whole, I used a common convolution reverb on all interface sounds. The idea was to create the illusion that everything takes place in the same sonic location — each interaction resonates in the same space, as if the UI itself had a dedicated “room” in the game universe.

Finally, I worked to establish a clear sound base for the interactions. Each action (validate, return, navigate, hover, etc.) has its own sound, but all share the same acoustic and gestural identity. The aim was to create a coherent language that would give meaning and feedback to the action without ever distracting attention. It’s a balance between functional legibility and artistic direction.
 

Did you do any custom recording for Clair Obscur: Expedition 33? Any helpful sound tools and/or indie libraries?

MP: The sound design for the main character’s skills was very much process-focused, centered on transforming and processing existing sources.

Kilohearts plugins were some of our key tools, especially for the Nevrons’ voice design. Their modular system gave us lots of control and flexibility to shape highly detailed and dynamic textures.

We also used Soothe 2 by Oeksound extensively throughout the project. It was essential for taming harsh resonances and helping to balance the overall density of the mix without flattening it. It became one of our go-to tools for keeping things tight and clean.

I was able to reuse some of my own recordings, such as recordings made at my father’s house of a large wooden garage door, which served as the basis for certain layers for the manor house door

OP: We didn’t record a lot of sound specifically for the project, but I was able to reuse some of my own recordings, such as recordings made at my father’s house of a large wooden garage door, which served as the basis for certain layers for the manor house door in a cinematic at the end of Old Lumière. There are also door sounds recorded at Sandfall, which, funnily enough, is also a manor!

As for sound libraries, we used Penguin Grenade’s Essential Magic a lot. The tonal sounds and beds are useful for creating texture. Of course, the BOOM Library was also an essential resource. I used Creature Foley for footsteps, especially for enemies.  

we’ve always been careful to build an artistic sound direction. For this, the processing work was fundamental.

For the more magical elements, Magic Arcane worked very well for bass and midrange, while Magic Wisps provided transients and tonal elements. We also drew heavily on Brute Force, which brings that raw, electronic, almost “out of reality” feel — an aspect we wanted to emphasize throughout the game.

Soundly was also an essential tool. With its subscription, you have access to an extremely broad library, covering a ton of needs without having to buy a complete bank for a single sound. It’s a perfect complement to other libraries, especially for finding atypical or specific sounds.

Even though we use numerous banks, we’ve always been careful to build an artistic sound direction. For this, the processing work was fundamental. Each of us had our favorite toys, but we shared a common base of tools: the Kilohearts suite and MeldaProduction plugins. For my part, I used Soundtoys and Valhalla Supermassive often, which are super-creative tools!
 

Can you talk about the sound work on the cinematics?

OP: The cinematics were a huge undertaking on Clair Obscur. In total, there are around 4.5 hours of them in 5.1 in the game, which is quite a lot. To start working as soon as possible without losing time due to changes in timing or editing, we had to be cunning, so we cut the work into successive passes, tackling first everything that could be done without depending on VFX or final editing.

The very first step was the voice pass. During the mocap sessions, the actors also recorded their dialogues, giving us placeholder voices perfectly synchronized with the animations.

The very first step was the voice pass. During the mocap sessions, the actors also recorded their dialogues, giving us placeholder voices perfectly synchronized with the animations. Then all I had to do was replace these voices with the final takes, precisely matching the files. In most cases, it was a simple matter of drag-and-drop. When there was a discrepancy in duration between the placeholder version and the final voice, I made adjustments directly in Reaper to correct lip-sync.

Once all the voices had been replaced, we did a QA pass to check that no placeholder was dragging and that everything was in tune.

Then, we moved on to a first ambience pass. The aim was to avoid having completely silent cutscenes and to establish a base volume for the pre-mix of the voices.

Alexandre went to each level where there were cinematics, launched the game, and recorded a minute of ambience in Reaper, which has a WASAPI loopback mode that lets you recover Windows sound directly in a Reaper track.

To achieve this, we wanted the ambiences to be as faithful as possible to the atmosphere of the game, so Alexandre went to each level where there were cinematics, launched the game, and recorded a minute of ambience in Reaper, which has a WASAPI loopback mode that lets you recover Windows sound directly in a Reaper track. The result was an ambience pad faithful to the location, with RFX included. Once the pads were ready, I inserted them into each corresponding cinematic. These ambiences were then reworked for the final 5.1 export.

Then, the cinematics team focused on the Flying Waters and Ancient Sanctuary maps, making 13 cutscenes (about 15 minutes). It was at this point that I began structuring our workflow around the cutscenes. I started by making a Reaper template in 5.1 with the help of Raphael, who had some experience in this area. We had one Reaper session per cinematic. To ensure sound consistency and work efficiency, I needed direct access to all the game’s assets. As all our sounds were exported to a Google Drive, I synchronized the “export” folder with Soundly. This way, every time I opened Soundly, the bank was up to date, and I could dip directly into it while I was working on the cutscenes. Flying Waters and Ancient Sanctuary also served as references for the sound design aesthetic of the rest of the game, so I designed them entirely: ambiences, foley, footsteps, and SFX.

As all our sounds were exported to a Google Drive, I synchronized the “export” folder with Soundly. This way, every time I opened Soundly, the bank was up to date

To ensure that I could work on a cutscene that wasn’t going to undergo any changes in timing or editing, we set up a process: once a cutscene was considered “timelocked” by the cinematic team, they would send a record to Raphaël and Guillaume, who would validate whether or not there were any other modifications to be made to the cutscene, and if it was validated then I could get on with the sound design. This process remained in place until the end of the production.

When we realized we might not have time to do all the foley and footsteps for each cutscene, we prioritized a first SFX pass on important cinematics. It was during this phase that I was able to have fun with the sound identity of important scenes or elements in the game, to create some strong signatures that would become part of Clair Obscur‘s world.

This was the most creative and narrative part of the game: creating sound identifiers for key moments and characters.

We even created a small soundbank dedicated to Renoir’s cane, which was very useful for the later cutscenes and trailers where he made an appearance.

It was also during this phase that Charlie joined me on the SFX of the cinematics. One of the first sounds we created together was Renoir’s walking stick. In just a few hours, he found the perfect delay, so much so that Guillaume and the cinematic team adjusted the timing of some of Renoir’s cutscenes so the cuts fell right in sync with the delay. We mixed our ideas: the transient and delay came from Charlie, and the lows came from me. We even created a small soundbank dedicated to Renoir’s cane, which was very useful for the later cutscenes and trailers where he made an appearance.

For sound design, we made extensive use of the Kilohearts and Soundtoys plugins. Crystallizer was my favorite, especially with the “Back in Time” preset, which I often modified to add stereo or length to certain sounds.

As Charlie, Victor, and I were working in the same Reaper sessions shared on Drive, we divided the maps between us to avoid conflicts, as there are often several cutscenes per map, so we could work simultaneously without getting in each other’s way. To keep up to date, we used a dedicated Discord channel, where everyone noted which cutscene or map they were working on.

Charlie Briere (CB): For the voice mix, Victor and I started with an initial volume balancing pass across all cinematics, following a loudness guideline to ensure a dynamic range that’s both comfortable for the player and faithful to the acting performance. After this first balancing stage, we moved on to a second pass focused on voice processing, using compression, de-essing, EQ, etc., and added backing tracks such as onomatopoeias, laughter, crying, and other character sounds. Those tracks were recorded separately from the voice lines, so we had to do some editing to make everything fit together.

The localization system was designed to swap audio clips depending on the selected language, and if one version was even slightly shorter, the voice line would get cut off during playback.

An additional challenge came from the multilingual setup. The English voice clips had to be exactly the same duration as the French ones. The localization system was designed to swap audio clips depending on the selected language, and if one version was even slightly shorter, the voice line would get cut off during playback. That meant we had to be extremely precise about synchronization between the two versions, not just in terms of length but also rhythm and pacing.

For our first pass of voice editing, we used Reaper to align the waveforms of the English and French clips, making edits to match the timing as closely as possible by trimming or adjusting the audio as needed. It was quite tedious work, but the extra care we gave to the French version, which was the only other language we had to handle, helped to make it good and qualitative.

We also experimented with integrating a separate reverb track, but quickly realized that Unreal’s sequencer was too unpredictable to maintain perfect and consistent sync with the dry voice lines. These lines are imported as single audio clips since the subtitle system relies on them, and if lip sync adjustments were needed, the process became a nightmare. We ended up switching to real-time, in-engine reverbs, which were much quicker to implement and less time-consuming, even if it meant sacrificing spatial depth and automation control.

I built the foley using a mix of in-game libraries and external banks, and pre-mixed everything to ensure a coherent overall sound, which made the final mixdown smoother.

Once the voices were mixed and the cinematic SFX were integrated, I did a full foley pass on all the cutscenes that didn’t yet have one, around 80% of them. I built the foley using a mix of in-game libraries and external banks, and pre-mixed everything to ensure a coherent overall sound, which made the final mixdown smoother.

Both SFX and voice work were handled in order of cinematic priority. We tackled the most important ones first (Tier 1), followed by Tier 2 and then Tier 3. However, we chose to integrate the foley chronologically, from Act 1 to Act 3, to make the final mix pass and Olivier’s adjustments more manageable.

During the foley pass, I also took the opportunity to refine some of the earlier sound design, adding missing or overlooked details, layering additional elements to anticipate the final mix and tweaks at the end of production.
OP: This work was done as a team, in a constant back-and-forth: each person completed or refined what the other had laid down, according to the needs of the scene. It was this collaborative approach that enabled us to make rapid progress while paying particular attention to every detail.

Victor Desrat (VD): On my side, once the work on the voice was done, I started editing the music into the cinematic. Lorien Testard had composed an enormous amount of music for the game, and most of the cinematics already had a rough edit done by Guillaume or William (from the cinematic team), reusing some of Lorien’s tracks. I then had to rework this rough edit, or to start from scratch, following Guillaume’s artistic guidelines.

The main goal was to make musical transitions cleaner and smoother by refining the timing of the edit point and adding reverb and SFX to mask them

It’s a process that requires a certain balance and reflection, because you have to appropriate the music, but not too much, and then manipulate it while remaining in tune with the vision of the composer, the music designer, and also the artistic director. The main goal was to make musical transitions cleaner and smoother by refining the timing of the edit point and adding reverb and SFX to mask them (there are a lot of stingers and risers in there!) — all of that to make the music strongly tied to the image.

My favorite part of the job was sublimating important emotions and actions by playing with the stems, adding or removing some tracks of the music, and creating a new custom version of the music that the player could already hear in the gameplay.

I kept some standard cutscenes open in Reaper tabs: some quiet scenes, and some busy ones. I then opened all the cutscenes of a map to be mixed in other tabs, and compared stems one after the other, adjusting volumes in relation to my standard cutscenes.

OP: About the mix, all the compression, saturation, and clipping had already been done during sound editing. I had my own work habits and effects chains, which often boiled down to light compression, clipping of spikes, and SFX resonance management with plugins like Venn Audio’s Free Clip, BOOM’s Uberloud and Oeksound’s Soothe2. As I usually clip, compress, and manage frequencies during my work on the SFX, there was no major work to be done during the mix, as most of the work was done beforehand.

The other part of the mix was the readjustment of levels, which was done very quickly. I call it that because the aim was to ensure that the levels of each stem were consistent between all the cutscenes. To get organized, I kept some standard cutscenes open in Reaper tabs: some quiet scenes, and some busy ones. I then opened all the cutscenes of a map to be mixed in other tabs, and compared stems one after the other, adjusting volumes in relation to my standard cutscenes.

As we had finished the foley, footsteps, ambiences, and SFX rather late, I had to do this mix and readjust cutscene levels in five days, which gave me an average of 10 to 20 minutes of work per cutscene, which was quite fast!
 

What were your biggest creative challenges in designing the sound of Clair Obscur: Expedition 33?

MP: One of the biggest challenges for me on Expedition 33 was figuring out how to help players anticipate and parry enemy attacks using sound. The parry system is central to the gameplay, so we needed each attack to be readable even without full visual focus.

One of the biggest challenges for me on ‘Expedition 33’ was figuring out how to help players anticipate and parry enemy attacks using sound.

We went through numerous iterations. Early whooshes were functional but lacked clarity. What worked best in the end was layering a sharp, high-pitched sound on top of every whoosh in the game. That cue cuts through the mix and acts as a clear audio signal to alert the player that an attack is coming. Whether it’s a sword, a projectile, or a spell, that high tone is always there.

We also added contextual audio hints, like Nevron screams or charging sounds, to create a kind of pre-cue just before certain attacks. Balancing artistic style with gameplay clarity was a real challenge, but it was a necessary one to make the parry system feel right.

TC: The biggest creative challenge of the project was to be free artistically, while keeping in mind that the sound had to help counter the attacks. We had to think about how to intelligently integrate a small sound just before the damage to help the players.

Most of the time, it’s a small metallic sound that stands out slightly in the mix or, for certain bosses, a designed sound that accompanies the artistic direction of their attacks.

The metal water bottle used in the Bénisseur sound design

Also, the biggest creative challenge was to find new ideas to create original and unique designs for each enemy. Sometimes, just a random object helped to define an entire design, such as a creaking chair for the Glissando, a metal water bottle for the Bénisseur, a kazoo for Pétank, or some funny voices recording.

What were your biggest technical challenges in creating the sound of the game?

RJ: Creating the interactive music system was quite challenging, but also super interesting. Florian and I designed it to be as robust, flexible, and integrated as possible, so new music could be triggered or affected in a lot of different ways, from a trigger box in the levels, to cinematic events, scripted game actions — even dialogue sequences could affect the music!

We had to create a very modular and bulletproof system, both on the music manager side, developed in Blueprint, and using MetaSounds presets, which relied on a lot of common functions and layers.

We had to create a very modular and bulletproof system, both on the music manager side, developed in Blueprint, and using MetaSounds presets, which relied on a lot of common functions and layers. But in the end, implementing new music was super fast, and the use of presets allowed me to quickly update the behaviour of a whole category of music all at once. So spending months to build good tools paid off in the end!

ADM: The Breath system for the main characters was quite a technical challenge since these are sounds that can recur often in a short time. So we had to be careful about the way we would implement it. We went for a single MetaSounds Source that is called on the character’s spawn and that can only be stopped by a cinematic, a battle, or a location change.

The idea of this system is based on a value that simulates the heart rate of the character. The longer you run, the higher your heart rate will get.

The idea of this system is based on a value that simulates the heart rate of the character. The longer you run, the higher your heart rate will get. If the player decides to slow down or stop, the value progressively reduces, like a human heart would behave. The intensity of the breathing is tied to this value.

We split our logic into three parts (Low, Medium, and High). We can smoothly switch among those thanks to our heart rate value. This required various performances from the voice actors. We needed distinct breathing intensity recordings to keep a coherent in-game result. We then kept this single-source method to handle Esquie’s swimming and flying movements in the World Map.
 

What was your approach to mixing Clair Obscur? What were some of your biggest challenges in mixing this game?

OP: The biggest challenge in mixing Expedition 33 was clearly the combat phases. We wanted something brutal and aggressive, but without overwhelming the player or obscuring the music. The balance between legibility, impact, and musical mix was therefore at the heart of our concerns.

Dynamic mixing was one of the first things I did when I arrived on the project

Dynamic mixing was one of the first things I did when I arrived on the project around August/September 2023, and it was perfect timing as I’d just finished Alex Rivere’s recently published book Game Audio Mixing! I’ve done dynamic mixing systems early in the production, as soon as the gameflow of the game is in place, because then you can make them evolve throughout the project and have new ideas to add, and so on.

In battle, I started with classic sidechain. For each hit (whether received, given, or during a parade), a slight drop is applied to the music and SFX to leave room for immediate feedback. On the QTE (quick time event) there are several sidechains. The first (rather basic) one lowers the SFX when the QTE is played. There’s another more subtle one — some spells can chain QTEs together quickly, so when a “QTE appear” arrives, the “miss,” “normal,” and “perfect” sounds are lowered slightly to give the “appear” feedback more impact.

To help with parries, I’ve added one that lowers the music when an enemy makes an attack, adding a bit of legibility during the dodging or parrying phase.

To help with parries, I’ve added one that lowers the music when an enemy makes an attack, adding a bit of legibility during the dodging or parrying phase.

There are sometimes important lines of dialogue during battles, as in the final battle against Renoir, so there’s also a sidechain that lowers the music and SFX.

We also relied heavily on Unreal’s Control Bus Mixes (via the Audio Modulation system).

when a main character or enemy launches an attack, a CBM (control bus mix) is activated to lower the music and idle sounds of the Nevrons, to better highlight the main action

As the game is highly structured and sequenced, it made sense to create sound “states.” For example, when a main character or enemy launches an attack, a CBM (control bus mix) is activated to lower the music and idle sounds of the Nevrons, to better highlight the main action. Counters are also managed with CBMs triggered from attack sequences — during a counter, almost all battle sounds and music are filtered and attenuated.

We had tested a temporary pitch-down of the music during counters, but the effect was a little ridiculous, so we removed it – except for SFX, where the pitch is always applied. It’s very discreet, but when it falls on a Nevrons scream, it can really add relief.

Another important aspect of the combat mix is everything that was done upstream in Reaper on the assets directly. For example, Maxance manually adjusted the EQ on the main character skills according to the exact moment of the hit in Reaper, to reduce the low frequencies and lower the main sound a little at the right moment, thus adding impact to the hits. For my part, I did a major bass clean-up on all the enemy and UI sounds. I processed the files one by one, shortening or even deleting certain bass tails that were too long.

I still remember the aural relief after cleaning up the combat IU sounds; it aired out the mix, and since then, I’ve been paying a lot more attention to the impact of bass on aural fatigue.

I still remember the aural relief after cleaning up the combat IU sounds; it aired out the mix, and since then, I’ve been paying a lot more attention to the impact of bass on aural fatigue. It’s amazing how much it can weigh down a mix without you realizing it straight away.

For final mix sessions in 5.1, I’d found a super-efficient method when I wanted to take measurements of the game in real time. OBS lets you add VSTs to your audio chain, and it can retrieve a 5.1 input. So we’d set up in our mixing room with a screen dedicated to Voxengo Span and MLoudnessAnalyzer and we could analyze the game signal in real time directly in our VSTs — super handy!

And in the same vein, as I said earlier for recording ambiences, in Reaper the WASAPI loopback also allows you to retrieve the sound from Windows and have the sound enter a track, so you can then use VSTs like Span or Youlean. If you do this, be very careful of the latency this method can cause, and remember to remove the routing to the master of the track that will receive the Windows sound!

Clair Obscur uses Unreal Engine 5. Was this a good fit for the sound team? How did you take advantage of the UE5 audio pipeline?

RJ: I must say, like many game audio folks, we were all trained and experienced with Audiokinetic’s Wwise. We love Wwise, and have a lot of Wwise-related habits, so leaving it behind to embrace Unreal Engine’s audio pipeline was pretty challenging at first, as it’s completely different!

in Unreal, everything is spread across the data. So, understanding the structure of the audio pipeline was our first trial, especially as it’s currently transitioning between two architectures

While Wwise gathers everything in one interface and one clear architecture, in Unreal, everything is spread across the data. So, understanding the structure of the audio pipeline was our first trial, especially as it’s currently transitioning between two architectures. But once we knew our way around the engine, Unreal audio became a very powerful audio engine. The MetaSounds allowed us to create complex audio systems in full autonomy, and the level of integration of the pipeline gave us a lot of flexibility to plug what we wanted into the Blueprints. Given the limited support we could have from the programmer, as their team was pretty small, working directly in Unreal was a good fit, despite all the challenges it posed. But most of all, it’s improving fast, as Epic is pushing a lot of audio updates in each of Unreal 5’s new versions, so I’m confident it will become more and more comfortable to work with it.

Unreal’s audio system is exciting to explore. Epic offers many well-designed native tools, such as MetaSounds, of course, but also Audio Modulation, which is really ideal for real-time mixing

OP: Unreal’s audio system is exciting to explore. Epic offers many well-designed native tools, such as MetaSounds, of course, but also Audio Modulation, which is really ideal for real-time mixing, as it gives us the Control Bus Mix I mentioned above. In the utilities section, there’s Audio Insights, which lets you profile audio performance.

We’ve also made extensive use of SweejTech’s Audio Inspector, a tool that I keep constantly open. It’s a bit like a mini profiler always on hand. It’s extremely useful for quickly checking what’s playing in real time.

When you put all this together, you realize that between Unreal and other plugins external to Epic, you can have a high-performance audio engine. Admittedly, it may seem complex at first. There are a lot of systems, logic, and tools to understand. But thanks to resources like Dan Reynolds’ YouTube channel, and above all, with a lot of testing and experimentation, you can get really solid results.

The creative team on ‘Clair Obscur: Expedition 33’

What have you learned or gained from your experience of crafting the sound of Clair Obscur: Expedition 33?

MP: Expedition 33 really taught me how to think about the overall coherence of a project, and how to build a consistent audio direction over time. It was the first time I worked this deeply on a fully developed universe, and it pushed me to think of sound as a whole, with clear priorities and structure.

You never have time to do everything, so you have to make choices and define early on what matters to you in the sound design.

I also learned a lot about prioritization. You never have time to do everything, so you have to make choices and define early on what matters to you in the sound design. As I mentioned earlier in the interview, we had outlined a few key pillars from the beginning, and those helped us maintain a strong, focused vision throughout the project.

OP: I realized that what I considered important was even more important than I thought. And that what seemed secondary to me was often even less so.

It’s the famous “good enough” that I used to repeat to myself as I finished the cutscenes when the deadline was only a few days away.

You always want to be as perfect as possible, but you forget that this battle for ‘perfection’ is a war against your own ideal outcome

When your head is in a project, you’re no longer objective about your work. You always want to be as perfect as possible, but you forget that this battle for “perfection” is a war against your own ideal outcome, at a time when the changes you make are sometimes laughable, even to the ears of sound designers.

Sometimes, however, there are more important things that we leave to one side because they’re less fun, less important for our ideal, but a little more important for the common project, and that’s something I’ll keep in mind in the future!

ADM: We had complete artistic freedom on Expedition 33, which is a huge luxury. Finding technical solutions to implement our work was really motivating. We had to get things working, and I think we learned a lot since, for most of us, this was our first project.

I’ve been able to improve my creativity and I’ve discovered new ways of doing design with more personal recordings.

TC: On Expedition 33, I’ve been able to improve my creativity and I’ve discovered new ways of doing design with more personal recordings. Being part of a small team working on an ambitious project also allowed me to work on different and very interesting tasks.

CB: Working on Expedition 33 helped me let go of perfectionism. In a fast-paced cutscene audio production like Clair Obscur, you have to aim for a solid “good enough.” Consistency and immersion matter more than endless tweaks. I became much more efficient in Reaper, learned to organize my sessions to meet deadlines, and stayed creative under pressure.

Working with the team was inspiring. We all had different approaches to sound design, and seeing how Olivier handled certain cutscene SFX taught me a lot. The team’s constant, thoughtful feedback helped me improve.

I learned to be disciplined under pressure, to take feedback quickly, and to be innovative with solutions.

VD: Expedition 33 was a great challenge in terms of technical skills, creativity, and the ability to adapt to different situations, as it was my first experience in the video game industry. I learned a lot from the audio team and the rest of Sandfall Interactive while developing my skills in sound design, audio integration, and the way I communicate. I learned to be disciplined under pressure, to take feedback quickly, and to be innovative with solutions. It was a total immersion in an ambitious and unique project that takes into account how every sound detail reinforces the universe of a game.

RJ: Well, Clair Obscur was my first experience as Lead Audio! I was used to music design and supervision, but I learned so much through the team’s management. Seeing the audio direction Maxance started years ago being realized to its full potential, and receiving so much positive feedback after the game launch was really heartwarming and encouraging. There were many moments of doubt along the way, but we made it, and I saw each member of the audio team growing so much in experience, taking confidence in their skill, and doing what they’re best at.

I also learned that authenticity and originality mean more than perfect execution.

I also learned that authenticity and originality mean more than perfect execution. It’s better to create something bold and daring, even if it’s not perfect or fully calibrated, than to create something that matches all the standards. I think it’s something that really drove the team throughout the development, not only on the audio, but at every stage of the game, and it helped to make Clair Obscur something fresh and unexpected. I must admit it sometimes clashed with my “industry standards” point of view (like releasing an 11-minute long track for the reveal of the game), but I think the players felt this authenticity, and it worked out. And I’m so glad I was wrong!

A big thanks to Raphaël Joffres, Maxance Playez, Olivier Penchenier, Alexandre De Mester, Théo Clavel, Charlie Briere, and Victor Desrat for giving us a behind-the-scenes look at the sound of Clair Obscur: Expedition 33 and to Jennifer Walden for the interview!

Please share this: