Exclusive Interview with the Audio Team of “Medal of Honor”

Medal of Honor has been one of the most important franchises of warfare video games. Most of their titles have really good sound work, and Medal of Honor (2010) is not an exception of it.

Below is an interview I had with the audio team of the game, talking about the different challenges they had and the techniques/processes applied on recording, designing and implementing the sound of this adventure.

Danger Close Games Audio Team:

• Audio Directors: Erik Kraber and Paul Lackey
• Sound Design Leads: Tyler Parsons and Jeff Wilson
• Dialogue Lead: Joshua Nelson
• Sound Programming: Eduardo Trama

Desinging Sound: With all those great Medal of honor titles already in the market, what was your approach on this title to make it as good as the previous ones but also new and different?

Erik Kraber: We come away from every project with a laundry list of things we wish we could have done better, but were unable due to time, resources, or technology. It is often hard to get past the frustration of knowing what “could have been” on the previous project, but it just fuels us to keep pushing it further each game. For this latest Medal of Honor we had quite a challenge, because it was an entirely new setting and decade for us to create within. So our massive library of sounds that we had recorded and built of every World War II weapon, vehicle and foley suddenly became dated.  MOH 2010 was about reinvention on all fronts, and audio was a big part of that. Fortunately, we had a very talented senior team of sound artists and programmers who have all worked on multiple MOHs before, so the franchise audio aesthetic sensibility and quality bar is now part of our DNA.

Our focus was to take our learning s from the previous games and go deeper on all the things we try to do on each MOH – focus on authenticity and the personal story of the soldier. We want to give the player the feeling of truly being in that world, living that experience. While not every sound in the game is necessarily realistic, they are intentionally not hyped to the point of feeling like a Hollywood blockbuster. Guns need to represent their real-life counterparts accurately but still provide distinct character and interact with the world in believable ways.  The voice acting needs to feel more like a documentary than a feature film and represent, as accurately as possible, the foreign languages of the indigenous people. The music needs to be current, exciting and fresh, but true to MOH’s orchestral roots and its focus on scoring the humanity of the soldiers throughout the story and not just support action.

DS: How was your collaboration inside the audio team and also the relationship with other development crews?

EK: Within the audio team, communication was easy, as many of us have been working together for years. We also collaborate with many external contractors for music, dialogue, and sfx and with other studios, especially DICE, who handled the development of the multiplayer portion of our game. Fortunately, as part of a big company like EA, we have the opportunity to share and learn from all the other great sound people in the numerous studios across the world.  Paul Lackey organized a Gun Summit, where we had EA audio teams who were working on weapons-based games come to Los Angeles for a week-long weapons recording session and knowledge sharing session. The result was a great exchange of ideas and approaches to sound recording, design and technology and some of the most comprehensive weapon field recording sessions ever done.

Tyler Parsons: For the single-player portion of the game, Paul, Jeff, and I each owned individual levels as well as global sounds (and focused on our own shares through the bulk of production), but we also held group play-through sessions as we got closer to final in which the entire audio team would gather and really scrutinize each level in turn. This made sure that everyone got “ears on” everyone else’s work and had the chance to offer feedback to improve the game.

Our sound designers interact with every other discipline in the game (design, background art, animation, visual effects, et al). We had a very strong relationship with the level designers – we would have frequent meetings and spotting sessions to discuss the latest changes in the levels, how we could best use sound to tell the story and enhance the experience. Implementing sounds throughout the game required us to constantly work with design scripting, the VFX and animation tools, and so forth (as well as collaborate with software engineers to draw up new audio features).  Luckily, the other disciplines were staffed by rock stars who appreciate the impact that sound can have on the player’s experience. We got excellent support.

DS: Could you tell us about your approach and designing the sounds of the weapons?  There’s a wide variety of guns and firing methods.  How this affected your decisions on the sound?

Paul Lackey: When I started authoring our weapons my initial goal was simply to be authentic and to strive for the highest fidelity in a game. I wanted guns that leapt out of the mix but with no distortion or heavy limiting so they could pop without hogging up too much space or fatiguing the ears. Fidelity came with outstanding source recordings and having a lot of options. The pop required setting the overall game volume pretty low to give the required dynamic range. However, once a few of the guns went in and were meeting my initial fidelity goals… I really wanted to reinforce that the primary way players interact in the world is through their weapons.  So gun shots needed to reflect off walls, echo thru canyons, distort thru air and be delayed by distance. Envelopes were added to provide subtle air distortion effects. Band pass filters that shifted based on distance were also used to tweak tonal qualities across the battlefield.  We sort of cheated on the acoustics and interior spaces by authoring the effects per zone rather than using the CPU to ray trace for collision. We had the tech but were already pushing the CPU so left it out. Hopefully next game. Additionally, the heft, power, and reach of particular weapons was conveyed by how much they interacted with the space. I couldn’t always make big guns louder but I could make them seem like their sound traveled further by boosting their tails and slap.  So how the weapons excited the air was used to tell something about the weapon in the player’s hand or the one firing at them. Finally, and I think a really important point, was that the weapons allowed players to reach out and “touch” the world. Of course we already had lots of bullet impacts to be heard when in close proximity to the player, but early on I found myself pulled out of the game when I shot distant rock faces but felt like my bullets passed through the geo because I heard nothing. So I worked on some very distant bullet impacts that smack the rocks, echo, and arrive via speed of sound well after you see the puff of dust. These are probably my favorite sounds in the game because they sound super cool and really do a good job of making me feel like I can interact with the world pretty much anywhere I can see.

DS: And what about the ambiences and warfare backgrounds?

Jeff Wilson: We designed our ambient backgrounds as realistically as possible, utilizing many audio recordings we captured ourselves. Wind, dirt debris and sand were components that were often mixed near field, giving the player the tactile sense of the rugged Afghani terrain. Our story often dictated that the player was a certain distance away from an ongoing battle, so we then chose distant explosions, weapon fire and aircraft to fill out the rest of the soundscape. We developed the tech to ‘anchor’ a quad ambience to the world, so we would mix the distant battle assets either stereo front or stereo rear, and then set the quad ambience file facing a certain direction in game.  This way the ambience would stay fixed to the world as the player moved in the 3D space, keeping the sound of the distant battle eminating from a specific direction.

TP: Backgrounds involving combat usually featured very distant battle elements to set the scene (i.e. distant engagements near Bagram airfield or in the Shahikot valley) with the bulk of near and mid combat sounds like weaponfire, explosions, bullet whizbys and impacts generated by actual NPC (non-player-character) events using sounds we’d previously created. To make certain battles more intense, I would create multiple versions of their areas’ backgrounds with extra layers of mayhem and then either swap to them (at an opportune moment) or layer them on top of the existing backgrounds if streaming bandwidth permitted.  I also created positional sweeteners like extra whizzes, impacts, and debris and scripted them to play at random points around the battlefield.

There are a couple of places in the game in which there’s a ferocious fight going on but the NPCs are of an inexpensive type that doesn’t generate any sounds in sync – weaponfire, foley, or dialogue.  Those moments had to be covered completely with chaotic backgrounds combined with positional sweeteners going off around the player – not unlike for a film, but with no way to sync it to the action. I think those moments ended up working pretty well despite the fact that they were completely smoke-and-mirrors,

DS: How was your workflow and process on the cinematics?

EK: For the pre-rendered cinematics, we handled them the same way post is handled on most animated films. We recorded the actors interacting with each other either on a stage or in a studio and assembled the dialogue into a cohesive radio play. That was delivered to the art department to use as their guide track for the mocap actors and animatiors. There were certain situations where rewrites or re-recordings had to be done to accommodate changes to story or technical issues, but mostly we were able to get the feel right on the first pass. Once the final animations were done they rendered a final movie file for us to work on in post. The sound editing work was divided between me and Earbash, whom we love to work with and have for many years. We use Nuendo here at EALA, and it has excellent mixing automation tools to handle complex multichannel mixes.  All of the mixes were done in 5.1. Our biggest challenge was trying to create an experience that never pauses. If you noticed, we have no loading screens in the game. To have that seamlessness, we split the music out as a separate track from the beginning and ends of the cinematics, so we could pre-cue the music for the cinematic before the end of the level and have the music continue on after the movie into the beginning of the next level.  Getting the mixing just right at the 11^th hour was really tough, but we are really happy with the results.

TP: Many cinematics were handled in-engine as opposed to cutting to pre-rendered movies. I would do a video capture of these sequences when they were as close to complete as possible, then create a one-shot bed of sound covering everything needed for a sequence. Depending on the way the scene was being handled (for example, whether or not the player has control of his movement or view, whether the player’s actions can change the length of the sequence, etc.), I would bounce specific elements of the cinematic separately so they could best be fit into the scene: positional elements in their correct places regardless of player movement, elements that need to persist playing as loops, and so on.

DS: All games, specially warfare stuff like Medal of Honor have constantly big amount of loud sounds.  How you dealt with dynamic range?  What challenges you found in the mix?

PL: We have a very rich mixing environment so I could modify groups of sounds based on lots of simultaneous game events. Much like ducking for dialogue but used when the player shot, sprinted, was hit by bullets…etc. This allowed key sounds to stand out without having to make them overly loud. This was ideal for sounds that occurred a lot, but for big moments… The headroom was there to let things punch up. One of the more challenging things we went after was seamless audio transitions between our levels. On paper this was a one line task, but to accomplish required lots of crazy logic to trigger and set volumes for music that bridged the gaps between levels and cut scenes, set cut scene volumes, cleanly handled fading out game play audio while pulling in assets for the next level …all while the levels that initiated the transitions got unloaded. This turned into a fairly big headache during our final days but was a bit of polish no one on the audio team could live without. Seriously a dedicated team.

TP: Some of the space for dynamics was provided by the script – the story features lulls as well as peaks, so we weren’t tasked with creating a neverending battle that’s always cranked up to 11.  The mix system we created for Medal of Honor allows extensive control of dozens of different buses and was used to make detailed, deliberate mix changes on the fly based on nearly any game event.  It let us easily trigger global or level-specific changes to carve out space for punchier explosions or weaponfire – or suppress or filter what would normally be loud in order for quiet details to peek through.

DS: How you deal with perspectives and space from the recording and also in the design and implementation processes?  What’s the approach on getting spatial realism in the game?

PL: Our approach to perspective was to record from all the distances we needed and then layered them in game. I didn’t rely too much on distance based low pass because distant sounds already sounded distant in the recordings.  I actually applied a good amount of high pass based on distance to keep things from getting muddy, and then as described above for weapons…implemented envelopes, delays, and filters to simulate air distortion and reflections. The other thing we did was to record our weapons in two locations. One shoot was in the mountains and was very clinical about capturing the authentic voice of each weapon plus we got nice long mountainous tails. The second shoot took place in an urban setting and was focused on capturing reflections from inside buildings, courtyards, alleys, and on rooftops. A trick I learned a few years back from a colleague at DICE was to add the same environmental gun tails to every weapon in a space. This really serves as a sonic anchor by putting each gunshot into the same location.  I would vary pitch a good deal on these but at their core…every gun shared sets of interior and exterior tails.

JW: While recording weapons, we set up our microphones at multiple distances to capture the weapon fire with a built in worldized effect. These assets were then edited into three distance models:  Near, Mid, and Distant. Then in game, we dynamically blend between these three perspectives to create a realistic sounding weapon at any distance.

TP: Many sounds (particularly weaponfire, but also a lot of foley and hard effects) were recorded from multiple perspectives to allow us freedom in creating our final game assets. Character foley usually had a close mic and a distant mic; vehicles were recorded from onboard and third-person perspectives as the gameplay required; weaponfire had as many as 72 channels of audio being recorded per gun, miked from close, medium, medium distant, and extremely distant locations. We also worldized dialogue assets in various canyon locations to give some of our distant dialogue a realistic sonic character.

Beyond the recording phase, we added spatial realism using our mix system. To handle environmental acoustics, we set up a series of plug-in chains whose parameters can be updated by any game event (such as entering a specific location).  Combined with the ability to modulate bussing to and from these effects, this let us adjust reverbs, delays, filters, and so on to match any space (or condition) the player was in.

DS: In an interview for Sound & Picture magazine, Tyler Parsons talked about using worldizing techniques in the dialogue.  Did you use that technique for other sounds as well?  How was the reverb used to reinforce the environments?

PL: We did not run anything besides dialogue through speakers , but we did do a lot of Foley and dirt/debris recordings in some quiet desert washes where we were able to capture the space of the place in our assets.  The results are so much better than close micing in a studio and then trying to simulate outdoors. Reverb was used in game for interior spaces.  Two slap delays were dedicated for the exteriors.

TP: We also worldized “Purple Haze” and a few other songs while we were out there, but somehow they didn’t end up in the game. We did of course record some sounds (weaponfire being a major one) in outdoor spaces like canyons or mock urban environments that gave us the sound of the space as well as the direct sound.

Several of our plug-in chains were set up with particular canyon or hillside spaces in mind, containing combinations of reverbs and delays that would create the aural characteristics of those areas.  We also created plug-in chain templates and mix setting templates for the different interiors in the game (caves, warehouses, huts, vehicles, various rooms, etc.).

DS: And what about the implementation process?  What were the challenges you found when integrating sound in the game?

TP: Despite our tools and overall pipeline being pretty sturdy by this point, we still had to regularly repair or re-implement a handful of sounds that would get accidentally disconnected by other departments’ last-minute changes (especially as things got busier toward the finaling phase).

Eduardo Trama: On the technical side, we had three major systems to balance between the three platforms – PS3, Xbox 360 and PC. They were memory limitations, CPU usage and streaming bandwidth. As we started adding/integrating sounds into the levels we found issues within these 3 major areas and worked on tuning each sound per platform to accommodate their varying restrictions. In some cases it was as easy as changing the compression rate, whereas, in other cases we had add extra technical features to reduce memory, CPU and/or stream usage. The challenge was that those system resources were shared with everything else in the game (e.g. physics, AI, graphics, animation etc) so we had to be clever and vigilant about keeping everything balanced and sounding right within these restrictions.

DS: What were your goals on blending music and sfx together in order to reinforce story, emotion, and other elements?

EK: Music has always been a very special part of the Medal of Honor franchise, with incredible scores written by composers Michael Giacchino and Christopher Lennertz. Moving to modern day we realized that the classic orchestral score was not going to be the right approach but we wanted to retain the heartfelt emotional quality that reflected more than just “action”. For this game we found Ramin Djawadi, an extremely talented composer who had many film and television credits – including Prison Break and Iron Man – but had never worked on a game. He had a great sensibility about how the score could fit emotionally between all the density of gunfire and explosions. We decided early on that we didn’t want the score to be a constant presence, as it had been in some of the previous titles (and with scores that good, can you blame us? J), but rather an element that comes in at the right moments to support action or stir emotion. We felt that with more areas where you just heard the detailed work of the combat soundscape, it would feel more real, more documentary-like, and then when the music came in it had more weight because of its absence. While at the end we had more music covering the game than we had originally thought we would (Ramin’s score turned out amazing) we still have much more of a range of music presence than before.

Because the score was a combination of electronic, solo ethnic and orchestral instruments, we had many layers of elements to work with. During the orchestra recording sessions on previous MOH games, we would have certain instrument groups not play on multiple passes ( e.g. have the entire orchestra play once, then have the strings not play, then have the woodwinds not play, then the brass, etc). This gave us several submixes with specific groups of instruments allowing us to build the intensity of the score with both off-line editing and in-game dynamic switching. But even with those submixes we were still limited to playing entire groups of instruments and not individual instruments. With Medal of Honor 2010, we were able to pick just tiny individual instruments or voices and build textural pieces that would subtly underscore the scene and foreshadow the full piece later in the level. The result was a score that enters the back door rather than the front.  The score creeps up on you more.  At the end of the day, while it is not as thematic as previous scores, I feel it has been the best at really tying in to the on-screen action and story by melting into the rest of the soundscape in a way that feels natural and cohesive.

Lastly, we wanted to go against the grain of the action in a number of spots in the game. While the default expectation is to score the action, Ramin and I looked for moments when you would score the emotion of the story, rather than the combat. An example of this is during the end of the Belly of the Beast level. With you and your fellow soldiers pinned down at a defend point, Tech SGT Ybarra, who has been desperately calling in for some support for some time, realizes that the enemy force has overwhelmed them and that they aren’t going to survive the attack. He radios back in to call off the air support, wanting to make sure that more American forces don’t get killed trying to help them. Within the hail of bullets from both sides, rather than score the intense action of the moment, the music scores the feeling of accepting death and acknowledges their quiet heroism. The SFX start to slip away and the slow string piece rises as the enemy forces push in to surround you and your men. This approach gave the game heart and connected the player with the tragedy and triumph on a more personal level.

DS: How challenging was the dialogue production?  How it was done?

Joshua Nelson: We were intent on crafting engaging, realistic and memorable game moments for Medal of Honor, so our team, including Writer Mike McCarthy, worked closely with military consultants throughout production. Our Tier One contacts shared how they communicate with one another and how they react to different situations in the field. When we shared game scenarios with them, they offered advice on how they would handle those scenarios. Mike did a lot of research and took much care with getting phrases and lingo correct for the dialogue. I think this helped tremendously with the feel, authenticity and coolness factor of our MoH characters because we didn’t want them to feel like “Hollywood.” That said, there are some great catch phrases that characters, like Voodoo, have in our game. Many of them come directly from Tier One Operators who tell us that’s just what they say when in the field.

Over the course of production we recorded much temp placeholder dialogue to test different combat scenarios and storytelling moments. We did multiple iterations of temp recording for each level and noted things like intensity and whether a character is winded from running. During each iteration, we worked to more accurately convey scene context and character realism through the dialogue. Later during final recording with the voice actors, it helped to have specific notes about intensity and contextual requirements, because it is easy to overlook whether a particular “Yes Sir” in a script of 9000 lines needs to be delivered calmly, stealthily, or over the noise of combat.

We recorded about 6000 total lines of Arabic, Pashto and Chechen dialogue for the foreign characters in the single player campaign. This included randomly triggered contextual dialogue for combat, and also many short conversations between 2 or more persons for use with enemy patrols and camp positions. We worked closely with translators who coached our actors in the recording studio.  They also served to change common English idioms to their Pashto, Arabic or Chechen equivalent. Literal translations sometimes don’t make sense, so they told us when something else would be said. Getting the foreign dialogue into the game is a fun process because the in-game characters begin to come alive, and co-workers in the EA office start shouting memorable enemy catch phrases at me when I pass them in the hallway.

In most levels of the single player campaign, the player receives communications through a radio com, which meant we would be radio processing a large number of the dialogue assets. I worked with Paul and Erik to get the right combinations of static, fuzz and background noise for characters. Erik wanted the radio transmission’s processing to reflect the sender’s proximity to and level of involvement with the player position and current situation. Squad members near the player had a clear channel, while characters who were not immediately near the player or who might be in trouble had more interference and background elements. A character in a dire situation might send a transmission that makes the player think “wow…that character could really use my help right now; maybe I can get to him later in the level,” so we looked for ways to reinforce this with the processing.

Another effect that we worked for was a radio doubling of character voices in close proximity to the player. In single player, when one of the squad members is nearby, the player can hear the character’s voice in stereo or surround sound space, absent of radio processing but effected by the local environment. The volume of the character’s voice would vary by distance to the player. The player will also hear the line through the radio com at a constant volume, processed and with a short delay. The combination of both sources together produced a nice effect, especially when the distance between the player and character varied over duration of a conversation. Since we did the radio processing on the sound assets themselves instead of dynamically processing them during gameplay, we actually had two WAV files for every dialogue line that had a radio component. This was a lot of work to edit, process, and get working in-game, but the result was worth it.