Creating the Soundscape of 'It Takes Two'

[Music] welcome to this presentation my name is anza fimonjo and i'm a senior sound designer at haze lights and i am joaquim any quebec technical sound designer at hastelike and today we're both here to talk about how the hazelite audio team developed the soundscape of it takes two and we will consider both the technical and the creative challenges that were brought by the split screen but also that were brought by the creative vision for the game before we get started let's first acknowledge and give due credit to the members of the hazelite audio team so everyone on this list has been essential to everything we're going to talk about today myself and joakim are here as representatives for the team but this project has truly been a team effort this presentation is divided into two parts in the first part we will talk about some challenges and found solutions while designing the split screen sound of it takes two

this includes managing spatialization and maintaining mixed clarity through carefully thought out sound design strategies the second part will focus on two selected sound design highlights from the game soundscape where we will demonstrate our inspiration our creative vision as well as our sound design approaches for those two highlights so let's dig in first of all what is it takes two it takes two is an action adventure platformer split screen co-op game that means there are two main characters cody and mei played by two separate players and the screen will always be split in two regardless of the players playing in the same room or online cody and may are two human characters who are transformed into dolls at the very beginning of the game and this is basically the story of how they navigate this new fantasy slash magical world which is based on the real world in order to get back into their real bodies so overall you could describe it takes two and its creative vision as somewhere

between toy story and honey i shrunk the kids where all the usual everyday objects become huge in comparison to the character's tiny doll size in all those objects they come to life they gain a personality and they play a role in the story either helping the characters or hindering them towards their goal so if you haven't played the game i want to give you a few examples of how this vision is realized by showing you a few images of some of the objects or the characters you might meet throughout the story you have a vacuum bus and all its vacuum hose tentacles you have a talking hammerhead guiding you through the level and is also used as a smashing ability a group of against gangster squirrels sending you on a mission to kill giant wasps the giant wasp queen who is actually an imposter in a wasp costume an overly protective baboon determined to trap you into an outer space realm

a train track turning into an infernal railcar ride a parallel kaleidoscope dimension giant evil microphone snakes and so much more of that so in short this game presented many challenges but also many opportunities for audio and among those opportunities and challenges are the following concerning the split screen how to manage ambiences and other world sounds from both sides how to maintain mixed clarity as well as emphasis on the right things in relation to what is happening at each given moment how to enable two spaces to coexist and be represented at the same time or one space to be represented from two different perspectives and about the creative vision how to convey this sense of bigness of everything in relation to the character's size how to maintain both playfulness and realistic immersion to support the tone

of the game and of course how to live up to the latest animation sound design standards such as pixar but in an interactive world the solutions and the strategies that we have developed to answer these challenges are both creative and technical and they would not have been possible without a close collaboration between those two aspects that is why we're both here today so first to introduce our split screen sound sound design strategies let's start by clearing up a few terms so the first concept of understanding of as we learn about this game is that of the listener which is a single world bound point of reference from which we observe sounds in a three-dimensional space and looking at this picture right here in front of us it's not really that hard to imagine a more conventional single player game where that listener will likely be attached to the character or the camera and serve as that singular point of hearing but it takes two is not a conventional game and part of our split screen setup means that we always have two listener

presents one for each character so instead of adhering to the idea of a lone listener representing one absolute truth of hearing we have to accurately represent both player perspectives as they are summed through one shared output and speaking about that output it's worth mentioning that it takes two is delivered for a channel based format of 5.1 surround so we opted to not go for an object-based approach and what we just said here is that this is the governing principle of audio it takes two it is the fundamental law and for the people tasked with filling this fantastic world with sound it was of utmost importance that one learned how to live within these restrictions that they were that this imposed but more importantly how to leverage the untold opportunities for us to tell new stories through sound and take the listening experience to places that the aforementioned conventional game couldn't really do so next up on concepts is that of 2d

versus 3d so a concept of audio for games that we must have a firm grasp on is that of spatial audio so that is to say a sound render to convey a point of emission within that three-dimensional space and commonly we would address those sounds as being 3d sounds and as a counterpoint to this spatial audio we have sounds that are non-spatialized it's non-spatialized which means that they do not appear to occupy space in the world so much as the occupy space on the screen through which the world is then viewed and we would refer to such sounds as being 2d sounds so a direct contrast to those that live within the actual game world and tackling the sound design of the takes 2 required a very conscious approach to these two different archetypes of sounds indeed so considering all that was just said about spatialized sounds one of our

very first brainstorm sessions in the very beginning of the project was about how we were going to manage spatialized and non-specialized sounds so both 2d and 3d for split screen there simply weren't all that many good sounding split screen games out there that followed the similar type of gameplay and creative vision as us from which we could get inspiration so our research basically told us we had to come up with our own way of doing things and follow our own intuition about how the game should sound so to introduce our split screen strategies i will start by talking about the ambience content in it takes two because ambiences and ambient world sounds form the bed of what players hear when entering the game it was very important to us that we start by clarifying how those were going to behave from early on so that we could then build from that foundation so and it takes two each player always has their own ambience independent from

the other players representing where they are in the world all ambiances in the game are usually made from a combination of a quad wav file playing the role of bass surround ambience so that is a one four channel looping sound that includes as many details of the ambience as possible we consider the quads in it takes two to be 2d especially considering that they are static in the sense that all four channels of the quad will always be routed to the same four channels to the speakers regardless of the player's movements the reason for that is because of the split screen so just imagine if we did allow the quads to be spatialized that means if the player on the left side is moving around let's say they turn to the left side that means that the whole ambience bed shifts through the surround axis which means some elements would go from the left side to the right side basically on the other player's side this could easily become confusing for the player on the right side we simply

thought that it would feel weird for both players to have their ambience moving that way in the context of the split screen that is and the soundscape of it takes two simply didn't benefit from this added layer of complexity so then we have also randomly generated spot sounds and those are helpful to add a bit of variety and randomness to the quads and help to world eyes the ambience elements so randomly generated spot sounds or as the name says spontaneous isolated ambience elements spawned at random 3d locations within the player's environment and those we consider to be 3d and what they include of course depends on the environment itself but common examples might be wood creeks random animals and insects dust wind gusts brussels things like that on top of those random spot sounds we have also manually placed spot sounds and those are of the similar nature so

also 3d in the world but they are placed manually in a level to match the visuals with a bit more precision so for instance extra wind gusts where there might be vfx to support the visual feedback same for leaves rustling or water lapping around the water areas and finally there are objects from the world emitting their own sounds so those are also 3d and can include for instance critters or animals passing by it could be machines or fans or moving platforms basically any objects any object that belongs to the world and is emitting their own sets of sounds so all of this put together creates a consistently rich atmosphere without any dead spots but that was only for one player so then comes the question of how to convey all of this and maintain the clarity with two players so before getting really into our technical solutions let's first listen to a video that will show how these described ambience layers

are combined into the game and what the sounding result is so in this video you will first hear the left ambience so the quad on may side and then the right quad will be added on kodi's side and then we'll one by one add the other ambience layers just as described so the randomly generated spot sounds you can listen in this case mostly for frogs crickets and flies which will be highlighted by the debug on the screen and then some manually placed spot sounds we have some water lapping around may some leaves rustling around cody and finally the critters are added so in this case we have dragonflies on may's side and some ladybugs on cody's side and the end result is a full mix of ambient sounds and after the video we'll start digging into how we made this work in split screen along with how our strategies for these ambient sounds are applied to all other types of sounds as well

is oh [Applause] [Applause]

uh [Applause] so as we might be going to see the gameplay of it it takes two is incredibly varied and diverse the most common case though will have the screen split vertically right down the middle with both characters views getting equal screen space there and this very rigid scene view really defined our design choices all throughout the project and before we had even begun putting sounds into the engine we started having long discussions on how to manage our screen space and this process of learning on how to build for our viewport was ongoing for the full project and it became a concept that i like to refer to as sonic real estate for it takes two sonic real estate meant that we at all

times adjust audio content to fit within the given visual frame by using carefully placed spatial and non-spatial sounds in that each occupy compartmentalized spaces on the screen so in much simpler terms it means that every sound is placed based on its visual position on the screen and again for the sake of clarity it takes two uses a channel based format 5.1 surround not any object-based audio so now let's take a look at a video that exemplifies sonic real estate by showing how the soundscapes of the individual player frames contributes to the final shared output so we'll watch a video now and at specific points throughout we will transition from hearing the full mix of the game to hearing isolated layers of ambience and reverb which will additionally also contain some voice over and now we really encourage you to pay close attention to the sounds that you hear and try to think about how they are

placed on the screen [Music] uh [Music] [Music] [Music]

is that water what is this machine uh who cares water pillars are cool so the first method developed to help us establish that groundwork of spatial placement is found in a basic technique aimed to place sounds on the horizontal plane this approach called speaker panning is one of the most fundamental concepts of audio audio processing even in linear media but in it takes two it became a very impactful tool so the mindset we employed when we used it was that speaker panning served to impose kind of a soft ownership of sounds towards either player

based on which player interacted with that moment at any at that object i should say at any given moment and this relationship was not at all static instead we always reacted to how object player relationships change during gameplay and then pan those sounds towards the side of that player's viewport so in essence maze world is a little bit more to the left and code is a little bit more to the right and while this basic technique was instrumental in outlining the fundamentals of the mix it was really only applicable to our non-spatialized sounds so again our 2d sounds taming spatialized sounds in a way that would allow us to control the sonic real estate like we wanted to would prove to be a much bigger task and we'll return to that topic after ansofi has delved a little bit deeper into what world dicing it takes to meant so just to summarize what we've spoken about so far what we've managed at this point is to build a foundation for a

split screen audio system we have established a degree of ownership of the sounds through speaker planning by sending each player's respective sounds slightly more towards their own sides we've also set a few ground rules regarding sonic real estate real estate so which sounds should be 2d or 3d and how much of the screen space they should be allowed to take in relation to gameplay and we have carefully designed ambiences and other world sounds with as many details as possible in order to reinforce each player's environment and their perspective in it with split screen audio due to the the nature of the medium and the fact that two environments need to be portrayed at the same time we felt like we needed to take every possible step towards greater immersion and towards selling the world in order to make it feel natural and believable which ultimately we think helps to minimize the distraction that the split screen might cause and so due to the split screen itself we needed to do that with some degree of

subtlety and so that meant not by trying to add more stuff which might quickly start to compromise the clarity of the soundscape but rather to make sure that the sounds that we do choose to include contribute to bringing the entire world to life we do that through worldizing strategies for days too our world housing strategies were these using convincing reverb that meant convolutional reverb with imposed responses that we recorded and designed ourselves and 100 of the game sounds in-game sounds are going through the environment reverb we also have real time environment-based delay reflections and so thanks to our raycast and environment type systems which allow the game to know in what type of environment the player is currently in so is it for example a small and large indoors outdoors and so on and sounds that are sent to the delay include vo foley weapons gadgets and vehicles

and we also have sound designed baked reflection tales that are added to sounds like weapons explosions and other large sounds and even footsteps sometimes so to illustrate these worldizing strategies and how they contribute to the soundscape we listen to a video showing how the world sounds with and without them so you'll first hear the awesome weapons and explosion sounds designed by philip eriksen where we will initially there will be initially no worldizing elements so there will be no reverb no delay and no reflection tales and then you will hear the same weapons and explosions with the added baked reflection tails and finally with the reverb and the delay added as well the video will also then show how the vo foley and other environment environment sounds may be affected by the environment type the reverb and the delay in real time

you know i wonder where that annoying book even came from he said rose bought him well he doesn't sound like a book she would buy more like a pretentious self-help book you'd buy you know

self-help books are very important to some people okay forget about the book let's get to rose get please inside gentle it smells moldy yeah humid down here did you see that cody i saw something what what was it i couldn't tell oh maybe a rat no it looked huge all right thank you and sophie so like i previously said our speaker panning solution and our ambient system it allowed us to bring this split screen mix and balance to a good level but we still experienced spatialization issues that were compromising the immersion and this was a result of our two listeners the problem was twofold first problem was that we suffered from

an unwanted increase in volume when both players observed the same spatialized sound source simultaneously and you see that visualized here the problem was in the way that weiss inherently combined channels when a single sound source was being picked up by both listeners at the same time the result was that for situations where both players experienced a sound from an identical or near identical point of reference to one another we would double the signal ratio per channel which translates to about six decibels of increased output so in layman's terms two players listening to the same spatialized sound would make it louder compared to it being heard by only one player this unwanted behavior was a massive thorn in the side first it would make it virtually impossible to mix this game and we had recognized its unpredictable implication on the mix early on in our design and our first response to balance the volume output was to implement and

continuously update a parameter based on listener proximities to the sound source so slightly reducing output volume when both listeners were close effectively counter adding the added in intensity from the double signal but this was very much nothing more than an ugly fix to an even uglier problem and we always knew that we had to go deeper to get a result that we would be happy with the second problem was that attempting to use one output device to represent one sound source but from the perspective of two listeners always resound results in a third perspective that is the sum of both that speaks the truth of truth of none and this some perspective meant that not only was the sound of this explosion for instance louder than it should have been because it sums the signal from both players but also that spatial positioning was not always truthful due to it only considering the closest player

so for example if mei is standing closer to that explosion but is looking slightly to the left the explosion will be perceived as coming from the right even though cody might be looking straight at it and we all already knew that the missing piece of this puzzle here was a result of how spatialized sounds were considered in our two listener setup energy distribution between the output channels was a some result of both listeners which means that the represented listener position was always a compromise of both we see that in this graph right here one listener could have a sound source dead ahead but if the other listener was closer to that sound and fully turned away then that sound would still appear to have the presence in the rear speakers of the surround setup so we decided that we needed to rewrite the rule set of how spatial panning worked in our game and make fundamental changes to how we rendered output based on our two

listeners this took a bit of shifting of a mindset so instead of pursuing the idea of always fully representing both listeners we instead decided that we wanted to create a soundscape where a spatialized sound was represented from the perspective of the most attentive and relevant out of the two listeners by constantly choosing just one out of our two perspectives and omitting the other one then any one sound source would always be rendered in a way that made sense and this new rule of spatialized audio in multiple listeners is what we decided to implement in an approach that we've simply called spatial panic so we'll now take a look at the video to exemplify said spatial panning in action so we have a sequence here and we'll listen to a series of explosions as they are happening around the players who will observe them from two different perspectives we will run the exact same sequence twice

first with just the default panning and then once more with our spatial panning engaged and to help you out here also visible on screen will be a volume meter showing the overall audio output so again now try to listen closely to the difference in how you might perceive the locations of the explosions as they're happening are on the screen and more so how the addition of the spatial panning affects it [Applause] [Applause]

[Applause] so what you just saw and hopefully heard was how spatial padding serves to improve accuracy in the mix its purpose is to take attenuation so that is how a signal reduces in power over distance take that and decouple it from the panning so it works in such a way that whenever more than one listener is actively mixing the same sound source we begin to look at and compare those listeners individual relation to that sound the listener the closest listener is chosen as a reference point for the attenuation and then the listener with the closest forward angle to the

position of the sound source will be our observer for panning together these two structures are combined into our one final listener output so in this slide right here we see an example of a situation that without spatial panning produced very questionable results with regards to the idea of sonic real estate one sound source two listeners one listener looking straight at the source one listener with its back fully turned but the combination of the two would sometimes produce as much signal in the rear speakers of a surround setup as it would in the speakers in the front even though it can only be seen by the player looking directly at it and who of course would expect this sound to be heard as if it's playing from the front but spatial panning solves this for us by in effect linking to what we hear to what is on screen and in doing so allowing those big sounds that really should fill the space

provided by the surround channels to do so without being muddied by unwanted and untamed spatial sources to make this whole procedure as transparent as possible we introduced interpolation to help smoothen out any drastic changes in listener sound relationships and a challenge in this was to make the panning update fast enough to be responsive to gameplay but still smoothly enough to not make for any jarring jumps in panning or amplitude that's not something that we could allow but in the end of all spatial panning was a major benefactor to the soundscape of it takes two allowing the spatial sounds to be rendered in a coherent and world-based way even when in range of both listeners and at the same time help maintaining that all important clarity in the mix and the final output mix and just pack everything together nicely so to summarize what we've added to our split screen strategies over the last

few slides on top of the solutions we've already spoken about like our speaker planning solution our sonic real estate management and our carefully designed ambiences we've we have added roralizing strategies such as convolution reverb real-time delay reflections weight reflection tails and of course spatial panning that compensates for those spatial discrepancies but when it comes to developing audio for split screen for all of those solutions to really live up to their full potential it was important to be clever about our actual sound design strategies and to choose what were the elements that were the most important and that needed to be represented as such and i like to call those the elements of subjective importance meaning that we choose based on our own judgment so subjective which sounds are the most important in relation to all other game sounds into what is happening on screen and in gameplay and those elements are then prioritized in both the design the sound design and the mix so that they come through

clearly and so that the player is getting the information the feedback and the story that they need from their actions and from the narrative development and this is where we relied on our pixar inspiration and the idea of bigness and exaggeration as a basis for our creative vision bigness was very important in it takes two because in order for that contrast of the characters being tiny in a big world to come through objects that were meant to be large in comparison to the to the players needed to sound even larger so in the end the elements that we decided were the most important were designed to be bigger larger and more powerful than any other sound playing at the same time and by focusing in such a way on one important element at a time it helps in keeping to a minimum the chaos that can come from busy sequences such as combat scenarios we also made an effort whenever possible to keep those important elements within the screen and by that i mean within the

left center right speakers rather than having them spatialize across the full surround axis so that they really would remain in focus and in the frame of the screen space in that way we believe that this important sound would be perceived as more cinematic and would be allowed to take more space in the mix in other words just imagine a sound made for a movie stuck to the screen so for sounding events that remained within the screen visually we preferred to simply leave them 2d as opposed to 3d and finally this bigness in the sound design was complemented with the use of hdr in wise to help us maintain this clarity at all times and just to make sure everybody understands what i'm talking about hdr refers to high dynamic range and in game audio that means we continuously look at what is playing and prioritize some voices over others based on set conditions that are pretty predetermined in the audio engine and for us that means wise

and this is meant to create the illusion that the dynamic range heard in the game is just as wide as the one heard in real life even if that is not true so for instance by lowering the volume of a category of voices such as ambient sounds while other types of more powerful sounds are playing such as explosions so that the explosions is perceived as being much louder than the rest and to translate this into our game uh you could say that those elements of subjective importance that i was talking about will always be given a higher hdr value than for example ambiences or other less important sounds so that the important elements are always heard clearly so if you want to learn more about hdr and especially about how it worked in wise and for this project i recommend you simply visit audio kinetics help page about it so now the video we're about to watch shows a few examples of how we approach the sound design and the mix for those elements of subjective importance

you can listen carefully to the sounds that are highlighted in the text as being the most important ones and notice how everything else in the soundscape seems to disappear and that is the result of our conscious choice so that these elements become the true focus and so that the confusion due to the split screen is minimized we basically use sound for narrative purposes we tell the player what is important what to look at and what to pay attention to in relation to the gameplay [Music] i can see the keys

the plate's not looking too good [Music] [Music] oh [Music] uh

wow i'm starting to like this place so before moving on to the second part of this presentation let's summarize what we've said so far about the split screen sound of it takes two two listeners sharing the same screen meant that we needed to represent both players perspective while maintaining a clear mix at all times we did so by using a speaker panning system that pans sounds coming from each player's environment slightly more towards their respective size by compensating for specialization discrepancies through spatial planning by designing detailed and rich ambiences and environment sounds that represent the world in as convincing a way as possible by relying on realizing strategies to make the world more immersive and reinforce the sense of being there by using hdr and by considering the elements of subjective importance when sound designing and mixing now that we have described our main

sound design strategies when it comes to the split screen we will enter the second part of this presentation and this part will further our demonstration of the sound design for it takes two by presenting two selected sound design highlights which have also required a very tight collaboration between technical solution and creative content the first highlight that we chose to share with you is called the time control ability so across one level called the cuckoo clock where all puzzle mechanics are time based kodi is given the ability to control selected objects in time and this means that when he activates his ability he is able to scroll through a timeline and move the selected object back and forth within this timeline and can also pause it and basically have time stand still for that object so it's 100 based on player input so here's a short video first showing what this ability does in the game and then we'll talk about how we made it work

[Music] [Music]

so to make this ability work we considered various complex solutions including granular synthesizers and other custom tools to be able to read through a sound file in both directions but we eventually put aside all those grand aspirations when we realized that the best sounding solution was not to develop a system that allows to manipulate and play a sound file in all sorts of ways but to actually sound design both the forward and the backward sounds as two separate sounds and have full control over what ends up being the sounding result of scrolling through that timeline as opposed to designing just one forward sound and hoping it sounds good when we manipulate it in all directions so in order to be able to do that what it came down to was actually much simpler than what we originally thought would be needed for such an ability so all we needed to know from the game was the scrolling direction so is the

player scrolling forward or backward and the scrolling progress so where in the timeline is the player currently scrolling and by sending those values in wise as rtpcs was parameters it was then fairly easy to structure our sounds based on both the direction and the progress and so the layers to build a time controllable interaction in their simplest form usually looks something like this we have one forward and one backward sound that are the exact same duration as the timeline and those are two distinct sounds each designed separately to fit the picture as well as possible we also have one forward and one backward tail sounds and those will play at the end of the timeline in both directions and are also designed as two separate sounds to match the visuals and finally the looping sounds used when scrolling is paused within the timeline and the object appears to be frozen in time so there can be many looping sounds basically as many as needed for instance there can be different sounds based on

the direction so was the player scrolling forward or backward prior to pausing and there can be different sounds based on progress for instance to have more intense or busy sounds playing if the timeline was paused in the middle of a destruction sequence as opposed to during a quieter moment in addition to the sound specific to one time controllable object cody is also manipulating a clock and has his own time control ability sounds that will always play no matter which object the player is interacting with the ability sounds follow the exact same logic as the object sounds so they have forward backward looping sounds and tails so as shown in this picture the object sounds usually come from the actual object being time controlled and the ability sounds come from cody who is manipulating the clock and this helps to bring more depth to the overall time control ability sounds as well as consistency throughout the level and through the various interactions because it gives the ability something that is recognizable and it becomes quickly associated with that action of scrolling through time

so to understand this a little bit better we'll watch this video isolating the various layers as i just explained and then we'll move on to our second sound design highlight so [Music]

so the second highlight that we wanted to bring you guys is called the singing ability so in the music level where all gameplay mechanics are driven by the background music mace singing is used to solve puzzles and progress through the

game the challenges here were that the ability is completely interactive just like the time control ability it's based on player input and that it must always be compatible musically with the currently playing background music track so this is a perfect time for us to really acknowledge the composition from gustav and christopher are two composers for the game responsible for all the fantastic music that you hear throughout and also samya rajabi who is the vocalist who performed maze singing for this ability so let's start this all by watching a short video showing you how this ability works within the game and then we'll delve deeper into how we handle it from a technical standpoint oh [Music]

oh [Music] [Applause] [Music] so obviously we knew that we had to do something spectacular with regards to this ability the requirements that i outlined previously and the whole way that this ability interacted with the actual music made us sit down and build glorious plans on how to make the singing always feel dynamic and musical so we decided

that we had to build a sampler we started this by drawing up a sheet that outlined the kind of audio content we needed to make the sampler come alive the conclusion that we came to was that for each scale of each music track for each four sub-levels we would then sample each note of the scale with three different intensities ranging from low to high so that the like the velocity setting on an actual sampler this would allow us to have some control over the energy in the singing based on the gameplay circumstances and to keep the vocals from ever feeling static or robotic we would also record three different variations of attack and release per note and then lastly to follow the key of the current music may would be able need to be able to go between sampled notes whilst singing so to realize this in a way that felt natural we would include additional layers of glissando covering each

interval of going from every note to every note in the scale and if you're trying to do math in your head right now let me just end it by saying that we ended up in source material that would have ended up in the thousands of files and furthermore at this point in time we still hadn't sorted the question of who would actually perform the singing to be recorded for these assets and after philip and i spent two days mocking up a track just to try and see what it would take to see our ground plans through to success it was very clear that we had given ourselves a mammoth of a task and so we did something that we do not always do in the haste light audio team we asked ourselves are we being reasonable right now and so we took it back a step a big finding in our prototyping is how clear it became to us that for this feature to be interesting keeping the singing in the correct key would really only get us halfway there so in our plans to record the different type of source we thought we needed we

hadn't really thought about the importance of allowing the recording singer to perform to the vocal track in a natural and a musical way and so we quickly formulated a new plan instead of building a sampler we would have each music track of the game be accompanied by an actual vocal track implementing the singing in this way means that we could allow our vocalist to perform to the music in the same way that you would approach approach any conventional song and just like that the musicality of it all was back now to make sure that our systems could stay synced with the performance of the singer we came up with a method where we our composers would i should say for every music track create a midi track that would represent the lead melody performed by the vocals the midi data then was parsed as wave markers and ingrained into the source files of the singing readily available to be read by the game engine

so the result now is that every track has a full-fledged vocalist track with all the musical bells and whistles and whenever the singing ability was activated by the player this track would be introduced into the mix playing alongside the background music and by using the data parsed into the markers from the midi track we could design around specific parts of every vocal track to make sure that whenever singing stopped whenever the player stopped using the ability the singing would fade out nicely based on the current passage that was playing so instead of an incredibly complex sampler-like system that likely would have taken us weeks if not months to deliver on we built a really really nice volume slider so a tricky aspect of our original plan had always been ensuring that whenever the singing stopped it would fade out in a natural way since that happens at a moment's notice whenever the player decides to stop the ability and short notes here were especially hard to keep

from sounding bad when stopping but our marker programmed fader really came through in solving that issue for us so now in the following video here you will see the effect of these markers as we progress through the music of the singing and pay attention to the top of the screen you will see gray bars that shows the markers and each of them denotes if the vocal track is currently allowed to stop when the player is no longer singing and if so how quickly it should fade out [Music]

and finally here are a few summary points and lessons learned to conclude this presentation the keywords surrounding the story of creating the soundscape of it takes two include deterministic we left as little as possible to chance and as much as possible to remain in our control so for example by designing sounds made to fit the screen space as opposed to aiming for a fully spatialized mix and just hope for the best we wanted to make sure both players would always experience the game as nicely as possible and that meant taking control over a few things so for example by designing important sounds in 2d meaning leaving them in the left center right speakers instead of specializing them and risking them ending up sounding from the wrong side of the screen the same goes for leaving our quad ambience is static while a different approach might have been beneficial for a different type of game this

deterministic strategy has been essential in creating the sound for it takes two and then the simpler the better the design of both the sounds and the systems needed to start simple and to be built upon based on pure necessity as opposed to designing complex systems in complex environments just because we can so what matters is really the result our experience in it takes two has shown us that the most complex solutions are not always the best sounding result and both the time control ability and the singing ability are excellent examples another example is choosing to deliver this game in stereo in 5.1 as opposed to object base and this was based entirely on what we believed would do the greater service to the split screen soundscape and the clarity of the mix and then prioritization in the split screen in a split screen game if everything is audible at the same time it just becomes messy so maintaining clarity is all about prioritizing and choosing what is important

and finally teamwork in our team we rely on each other's strengths and skill sets to raise the bar to the highest level our collaboration between the technical and the creative designers has been truly essential to bring this project to success and finally it's important to say that all our solutions worked very nicely for this specific game but split screen audio is truly a pandora's box and a different project might require entirely different solutions but that with those takeaways we are confident that we are ready for the next challenge thank you very much for attending this presentation and we hope you enjoyed it you just stay right where you are san francisco see you soon [Music]