‘Back 4 Blood’ Composer Martin Stig Andersen on Crafting a Handmade, Game Engine-Generated Score

Acclaimed Composer Martin Stig Andersen (LIMBO, Inside, Control and many more) scored the latest release of Turtle Rock Studios: Back 4 Blood. Andersen's unique approach features a palette of handmade instruments, out-of-the-box recording techniques, and an in-game modular approach to scoring the game (as opposed to a more traditional score with pre-composed tracks). Here, he drives into details of his inventive process from creation to completion:

---

Interview by Jennifer Walden, photos courtesy of Turtle Rock Studios, Inc., WB Games Inc., Martin Stig Andersen

---

Please share:

• More

Back 4 Blood — developed by Turtle Rock Studios and published by WB Games, Inc — is much like its spiritual predecessors the Left 4 Dead games in that Back 4 Blood is a multi-player online FPS zombie-survival game, where four players cooperate to kill hordes of the undead known as Ridden.

In terms of sound, one thing that makes Back 4 Blood stand out is the score. Composer Martin Stig Andersen (who teamed up with co-composer Nathan Whitehead on the game’s soundtrack) chose a palette of innovative handmade instruments as the basis for his tracks. He worked with various sound artists — like Katrine Amsler, Fathma Fahmi, Daniel Porcelli, Daniel Martin Nielsen, and Nicolas Becker — to create, curate, source, record, and process a collection of one-of-a-kind instruments and sounds that, when combined in-game, form the score.

Here, Andersen delves into his process of discovery for the ‘voice’ of this soundtrack, how they created, recorded, and processed the instruments, how they used a modular approach to build the score in-game so that it’s tailored to each player’s unique experience, and so much more!

Back 4 Blood – Launch Trailer

Going into Back 4 Blood, what were some of your initial ideas for the sound of this score?

Martin Stig Andersen: Right from the beginning, the developer had this idea that everything should sound DIY; that it should be composed from everyday objects you could find in the game’s world, like objects you could find in an abandoned barn.

Composer Martin Stig Andersen

Likewise, they wanted to explore Hillbilly instruments, so I acquired a huge zinc washtub and built a washtub bass from it. This became my main instrument for the project, and I ended up composing entire sections of the score using only the washtub. Nathan Whitehead, who also composed for the game, bought a shovel guitar which he used in much of his music, including the main theme for the game.

For my contribution which is focused on the in-round music, we wanted the music to be very soundscap-ish and interactive. Since the game is very dynamic with intensity changing moment by moment, we chose a rather granular approach to the music — meaning that it consists of short musical segments and layers that are stitched together by the game on the fly.

We also tried to encode as much gameplay-relevant information into the music as possible in order to help and empower the player. In some ways, the music is actually driven and almost composed from gameplay information.

You mentioned a unique and varied palette of instruments and sounds in your score, like that zinc washtub. I hear a lot of metallic, resonant sounds and tones in there. Can you talk about your process of finding/creating these instruments and sounds?

MSA: I worked a lot with a sound artist here in Copenhagen, Katrine Amsler, who has also helped me on many of my previous projects. Katrine usually builds her own instruments, often string or spring-based.

Sound artist Katrine Amsler bowing the zinc washtub

To achieve the metallic sound, we amplified the instruments using the washtub. I had created this setup where I could play sounds into the washtub by attaching an audio transducer to it and then recording the sound with microphones and contact microphones.

I also had a mechanical record player that I would attach on the side. I would put the needle down on the surface of the washtub so it would pick up and amplify the vibrations. Additionally, Katrine would put some springs on the tub that would rattle like hell. All in all, we ended up with this crazy mechanically distorted sound.

…I could play sounds into the washtub by attaching an audio transducer to it and then recording the sound with microphones and contact microphones.

For the Hag, I wanted to record a one-of-a-kind instrument called “Bronte” in Paris. It’s a metallic sound sculpture that can produce an almost female-chorus-like sound. But unfortunately, I couldn’t go and record it because of the pandemic. We tried really hard to make it happen but it was impossible. My friend Nicolas Becker — who did the Sound of Metal film — had already captured recordings with this instrument so I licensed those from him, and then we did all kinds of treatments to those recordings. Fathma Fahmi, who also helped me a lot on this project, would process the sounds using her voice, for example.

Henrik Knarborg Larsen playing a gong submerged in water.

The Titan boss in Back 4 Blood is almost swimming underground, so the developer had this idea that it should sound a bit under-waterish. I had a remote recording session with percussionist Henrik Knarborg Larsen where he would play tam-tams and gongs submerged in water. After processing and composing with the sounds, they were post-processed through water by means of filling the washtub with water, projecting sound into the tub via the audio transducer, and picking up the sound using a hydrophone.

Daniel Martin Nielsen discovered that it sounded cool when he put the transducer into his oven.

While helping me with this process, Daniel Martin Nielsen discovered that it sounded cool when he put the transducer into his oven. Later on, we used that trick for creating some of the source materials for the Breaker boss in Back 4 Blood.

We hunkered down in a scout cabin with Katrine Amsler and played string and spring instruments into the oven so all of the baking plates would rattle. The sound is somewhere between a shipwreck and a distortion pedal.

Going back to this granular approach to music and the game engine piecing together bits of music segments based on the player’s action, how did this impact your approach to scoring Back 4 Blood?

MSA: To compose something highly adaptable, I felt it was important to avoid things that seem predictable. For example, a four-bar period or even just 4/4 will make the listener anticipate the next period or first beat, and any sudden changes will potentially disrupt that musical flow. This may be OK when happening every once in a while but when it potentially happens every couple of seconds it doesn’t really work.

…a four-bar period or even just 4/4 will make the listener anticipate the next period or first beat, and any sudden changes will potentially disrupt that musical flow.

The same goes for melodies and chord progressions; they require more time than a game like Back 4 Blood will allow for. So instead, I’ve been focusing on texture and randomized time signatures, including odd signatures such as 17/8th. In that way, we could really make the music change moment by moment because the listener doesn’t get those expectations as to what is going to happen in the music. So we could make musical shifts real quick.

…the developer had already decided that for any sound effect with a tonal quality they’d aim for the key of D.

Before I got involved with the game, the developer had already decided that for any sound effect with a tonal quality they’d aim for the key of D. They asked if I wanted to change that but I thought it was really cool that they had even considered this. In the end, it really helped blur the line between music and sound design.

I’d say that I’ve composed in D-ish — meaning that everything happens somewhere between a half note below and above D. So overall, it became much more about finding the right sound or texture for the score rather than melody or chord progression.

For the length of these musical segments, were you going for something shorter than 30 seconds? Less than 60 seconds? Did you have a standard length for what these elements should be?

MSA: It depended on the function of the music, but for the general in-round music the segments are mostly pretty short. We were working in Audiokinetic’s Wwise, and I was using both the interactive music editor and the sound design editor as a compositional tool.

When it comes to the rhythmic parts, those segments were rarely longer than one bar. They were often around three to six seconds including pre-entries and tails — sometimes even shorter. This allowed for a lot of randomization and fairly quick changes in tension, etc. without the need for applying crossfades.

While the shorter, rhythmic sounds are mostly arranged in the interactive music editor in Wwise, the longer, often more ambient sounds sit in the sound design editor.

Stingers tied to enemy and player actions are also a major part of the music.

While the shorter, rhythmic sounds are mostly arranged in the interactive music editor in Wwise, the longer, often more ambient sounds sit in the sound design editor. For example, there’s a bunch of swells that are continuously being triggered from the music editor via events or MIDI. Different swells are triggered according to overall intensity, and there’s also stuff like Shepard tones that will come to the fore when things go crazy.

During in-round, we occasionally have scenario and boss music which is more linear and involves much longer segments. The scenario music tracks, composed by Nathan [Whitehead], are musical set-pieces often occurring towards the end of acts to enhance climaxes. They offered a really nice variation to the general in-round music as well; it’s a place where the tonality is changing and we could hear some of Nathan’s theme music.

…we occasionally have scenario and boss music which is more linear and involves much longer segments.

We didn’t have music for the special Ridden because — as compared to Left 4 Dead — they appear all over the place. Instead, each of them has a custom set of stingers that will play when they make a specific attack or when you hit a weak spot, for example. The amplitude of any stingers relevant to the player’s action is monitored in Wwise and used to affect the general music by means of switching, filtering, modulation, etc., so triggering any of those stingers is like throwing gasoline on a fire.

Compared to the general music, the Boss music involves longer musical segments, say 10-20 seconds each.

Is the game engine able to change the tempo of the music segments? Or is the tempo set and can’t be changed?

MSA: The tempo is set. There are different tempi but in this game, I really wanted to keep the original playback speed of the assets.

…I really wanted to keep the original playback speed of the assets.

Some of the bosses have different tempi, but I made sure that they were compatible. So for example, the huge Ogre is 65 BPM while the smaller Breaker is 97.5 BPM. So there’s an interesting ⅔ relationship going on when they coincide in the game.

How did you organize these musical segments? Did you sort them by “ambient textures,” “percussion pieces,” “guitar-esk pieces,” and so on? Or was it broken down by areas of gameplay, like these segments will work for the parking lot and these will work for the warehouse?

MSA: In the beginning, the intention was to have a music suite for each different Act. But we discovered that it was more interesting to shuffle them around a bit, to get more variation within an Act. So we have a suite playing during each round and we picked the one that we felt would work best for that specific round.

Each suite has two main parts: explore and horde. It’s a bit misleading to call it “explore” though because in this game you’re virtually always in combat. So exploration music ranges from ambient (when no enemies are nearby) to intense combat.

Each suite has two main parts: explore and horde. It’s a bit misleading to call it “explore” though because in this game you’re virtually always in combat.

Horde music also covers the whole spectrum but is more intense. The change between the two lets the player know when a horde is triggered and when they beat the horde.

Explore and horde music each have a rhythmic and a textural layer. Adding to that, each special horde type also has a unique textural layer that will override the suite’s generic layer when active. The horde type may change while a horde is going on, say between generic, Reeker, and Snitcher horde, and the textural layer will change accordingly. The horde-specific textural layers will combine with the rhythmic part of any of the suites which adds variation in the music.

>

Hear about Martin Stig Andersen’s critically acclaimed approach to the sound for another game title – Playdead’s ‘Inside’

The horde-specific textural layers will combine with the rhythmic part of any of the suites which adds variation in the music.

Then we have the aforementioned scenario music and custom music for the four boss-type special Ridden. The player will occasionally be facing more than one of the bosses at once, maybe two or three at a time, so the music has to be able to play in any combination. For instance, if the Breaker is active while the bigger Ogre spawns in we’ll change the main music from Breaker to Ogre while the Breaker’s music is translated into stingers that will play on top of the Ogre music.

The Hag’s music is based solely on stingers and texture and blends well with any of the other bosses. Triggering a horde will also change the boss music itself, and the relevant horde-type texture will also blend into the boss music. In this way, the musical assets are continuously combined in new ways and that helps to keep the music fresh.

Then there’s all the music between rounds, cut scenes, etc. that Nathan composed. One of my favorites is the safe room music which has three different versions that will play corresponding to how well the player did in the last round. So if you went to a new round and beat it right away, you hear a very soothing version of safe room music. But if you did really bad and you died the first time around, when you come back you hear a more gritty version.

In a different interview, you mentioned that the music isn’t just supporting the action and the emotion, it also provides audio cues that can help the player survive. Can you elaborate on that?

MSA: Whenever there is a special Ridden around, doing different attacks, you will hear musical stingers. Each Ridden has a set of stingers, including a handful for each of its attacks, for example. You could have one attacking you from behind and one flying through the air — all these moves that they make cause stingers to play.

…the stingers aren’t playing on top of the music but rather they become the music.

It’s the same for the bosses. You can hear when the Ogre is about to throw a tumor and things like that. I like the idea that the stingers aren’t playing on top of the music but rather they become the music.

Accordingly, if nothing happens in the game then nothing is happening in the music. You may run away from the fight and do some looting, and all you’ll hear is some distant ambient textures.

The fun part is creating something that is solely tied to gameplay and then at the same time sounds musical as well.

The fun part is creating something that is solely tied to gameplay and then at the same time sounds musical as well. In other words, the score is essentially a musical structuring of gameplay information relevant to the player.

After hearing the music come together in-game, what surprised you the most about the result of this approach?

MSA: How different the music sounds depending on who’s playing the game — even between players playing the game together.

Throughout the development, the music team — Nathan, Silvano Matthews (music supervisor), and Alex Pappas (audio programmer), and I — had weekly playtests together. I would sometimes hear a teammate being excited about the intensity of the music while everything sounded lowkey on my end.

There are some chord progressions and Shepard tones that you’ll hear only if you do an equivalent of one headshot per second over a period of time.

After the release, I spent some time watching Twitch just to hear how the music works in practice, and admittedly it often sounds much better than when I play it. There are some chord progressions and Shepard tones that you’ll hear only if you do an equivalent of one headshot per second over a period of time. I’ve been mimicking this in Wwise by spam clicking stingers in the soundcaster, but it’s really cool to actually hear it in-game.

I like the way the responsiveness of the music rewards the player for doing well. I still check in on Twitch now and then. There are gamers playing on Nightmare, who have played well beyond 100 hours, and yet haven’t touched the music slider. I take that as a compliment given there are around 80 minutes of music in the game, virtually playing wall to wall.

A big thanks to composer Martin Stig Andersen for giving us a behind-the-scenes look at the sound of Back 4 Blood and to Jennifer Walden for the interview!

Please share this: