Unreal 4:
Bump in the Night
Project Snapshot
Bump in the Night is a stand-alone Unreal 4 project demonstrating the best practices for audio as a horror device. This project is the culmination and artifact of my Master's thesis research into the audio techniques utilized by horror films and games to control tension and create pacing through peaks and troughs in player tension. The artifact is a first-person horror game about a character moving into their new home that contains no visual scares and creates horror and tension only through audio.
Engine: Unreal 4
​
Development Time: 150 hours
Models and Textures by: Mike Porter (thesis advisor, SMU)
​
All Other Development: Sierra Clark
Gameplay Synopsis
Bump in the Night is a simple first-person horror experience starting with the player moving into their empty new home. They proceed through the space turning on electronics and placing furniture, decorating and filling the house and slowly building a familiar and comforting residential soundscape.
​
Things begin to take a turn when the player finds an abandoned doll in the guest room closet. Tension builds as the soundscape around them falls suddenly silent. Uneasy, but still within the realm of ordinary, the player proceeds upstairs to continue decorating the remaining bedrooms and study. Along the way they hear movement around the house including doors closing across the house, footsteps, and muffled voices.
​
The unnerved player exits the study to put on music downstairs only to find themselves followed by a set of footsteps that is not their own. When they turn on the gramophone, they're relieved to find the soundscape returns to normal. This respite ends abruptly when the player goes to place the washer and dryer in the garage. As they enter the garage the record scratches and the song changes to an unfamiliar track. When the player tries to check what happened they find themselves locked in the garage. They retrieve a spare key to readmit themselves to the house, and the music snaps back to the expected track instantly.
​
The telephone begins to ring in the master bedroom. Eager for help, the player races upstairs to answer only for the door to slam in their face, locked. They hear muffled voices as the child's bedroom door opens beside them. When they follow the voices inside, the player discovers they seem to emanate from the mysterious doll they'd placed on the child's bed. As they approach to investigate, the door shuts and locks beside them.
​
The doll starts screaming. Trapped, the player is helpless as chaos ramps up across the house. Doors slam, telephones ring, footsteps charge up to the room the player is locked inside of, and the door is nearly pounded down by unseen forces. Throughout, the doll continues shrieking, rotating to face the helpless player as they search the room for any way to escape. Finally the door is opened for them. As they exit the bedroom, they see and hear the bedlam of every door in the house flapping and slamming randomly. As they flee towards the stairs to escape, the player loses control as the character slips, the voice of the doll sounding just behind them as they fall down the stairs. A grisly crunch hits loudly as the screen snaps to black.
Level: Unremarkable House Interior
Bump in the Night takes place in a single two-story, three-bedroom house. The first floor includes a foyer, living room, kitchen, dining room, guest bedroom, bathroom, and music room. The garage connects to the kitchen. The upper floor includes a master bedroom with attached bathroom, a child's bedroom, a bathroom, and a study.
​
The space was intentionally designed to be as visually neutral as possible. As the focus of the research was on the effectiveness of audio horror, it was critical that the level itself not provide visual elements that made the player too comfortable or too uncomfortable. Deliberate choices in colors, lighting, room size, and layout were made to create an intentionally unremarkable house. All models of furniture were provided by my thesis advisor, Mike Porter.
Neutral colors and flat lighting were used to create an environment that would not influence the player towards comfort or discomfort. An uninfluential environment was needed to test the efficacy of the audio alone.
Best Practices for Horror in Audio:
Act by Act
The overarching body of my research was investigating the techniques used by effective cornerstones of the horror genre in film and games and identifying which were used cohesively across multiple projects to accomplish the same intended tension effect in the audience/player. My research involved critically reviewing pieces spanning from 1947 to present and identifying common themes in their audio techniques, describing them, identifying at what point in the tension curve those techniques were most effective, and demonstrating them in my final artifact. All in all, I identified 10 distinct techniques and the best practices for their application. They're listed by order of their best timing in a horror project.
Act 1: Introduction
Establishing Normal
The first technique I identified, used in the first act of successful audio horror experiences, I defined as "establishing normal." Audiences and players will not respond to later horror techniques unless those techniques are out of line with what they already expect from the environment. This is very in line with the standard first act of a horror movie, where for example the family moves into a haunted house and spends the first few days there living in peace. The later scares with the ghost wouldn't be as unnerving to the audience or characters if hauntings were an everyday occurrence for them.
​
In my artifact, this technique is exemplified by the period the player spends decorating the house before getting the doll. They build up tolerance to an expected ambiance of air conditioning, appliance electrical hum, and birdsong out the windows.
Common household white noise sources such as AC, the refrigerator, and birdsong lure the player into a sense of security.
Act 2: Rising Action
Strategic Silence
As the player reenters the guest bedroom from the closet, carrying the doll, the house and birdsong fall silent.
Transformation of Expectation
The next technique I identified was the most subtle of all the techniques. Once a baseline of ambiance has been established, players subconsciously latch onto it and come to expect the audio events that happen throughout it. The most subtle modification possible of this expected soundscape that builds tension is simply silencing it. This has the unique impact of creating extremely subtle anticipation as they await something they subconsciously expect - only for it to never come. This pairs with an instinctive prey response: if all the other animals and insects fall silent, it's because a predator is nearby and they don't want to be noticed either. It subconsciously hushes the audience and puts them in a more cautious, self-conscious state.
​
In my artifact, this was demonstrated as soon as the player exits the closet with the doll. The air conditioning falls silent and even the birds out the window stop singing.
The next technique I identified as the transformation of expectation. While strategic silence involved removing expected elements entirely, transformation of expectation is all about replacing the expected with the unexpected. This can be done subtly to create a sense of unsure dread while the audience questions whether they actually heard what they thought they heard or if they're "just hearing things." It can also be done very obviously, typically to create a flinch response similar to a "jump-scare," particularly if the sound is loud and/or brief.
​
In the artifact, the subtler variation of this technique is demonstrated by the refrigerator the player turns on. It initially has an expected electrical hum, but this sound is later replaced by the sound of human breathing. If the player draws closer, the microwave beeps and it abruptly returns to normal. An example of the more obvious version is the replacing of the musical record with an unexpected song, and the replacement of the doll's expected "I love you" recording with screaming.
Strategic Volume
The next technique I defined as strategic volume. The volume of a sound is one of the many ways the human brain immediately tries to ascribe importance and meaning to sound. Louder sounds are interpreted as both more important and closer than sounds that are muffled or quiet. A sound that changes in volume can therefore be construed (accurately or not) as either getting closer and possibly more dangerous or more distant and less crucial. Additionally, humans have a natural cringe response towards sounds that are perceived as loud in an instinctive effort to protect our hearing long-term.
​
In the project, I leaned into modulating volume with steps running up to the door of the room the player is in. The implementation of certain distressing sounds, such as the doll screaming, also causes ducking on the other sounds. This effectively makes them seem much louder without needing to actually play them loud enough to damage my players' hearing.
As the player is locked in the garage, the record they've been listening to is replaced with St. James Infirmary Blues.
The doll's scream at the climax of the game ducks other audio sources, tricking the player into feeling the sound is even louder and more distressing.
Audio Jump-Scares
This technique is fairly self-explanatory. Any unexpected sound that is loud or sudden enough can cause a flinch response. While people typically associate the idea of a jump-scare with something visual popping up in front of the player when it's least expected, it's also crucial to notice that these visual jump-scares are almost always accompanied by an equally unexpected sound. As a testament to their effectiveness, it's actually far more common to find audio that triggers a jump-scare response without a visual jump-scare element than it is to find a visual jump-scare without any audio jump-scare at all.
​
In the project, an unexpected audio jump-scare plays when the player places the piano: a loud, discordant chord of notes causes the player to jump.
When the player places the piano they're startled by a sudden discordant chord.
Act 3: Climax
In addition to the new techniques described below, all preceding techniques are now fair game.
While the player is locked in the bedroom, the record on the gramophone is changed. Its pitch begins to distort rapidly, as though the turntable is spinning at changing speeds.
Instinctive Terror in
Sound Sources
Strategic Pitch Manipulation
The technique of strategic pitch manipulation is another play on making an expected sound suddenly seem more foreign and unsettling. In particular, I identified that if the pitch continuously modulates it can achieve the greatest effect. Another trick for increasing efficacy is to specifically pitch-manipulate organic voices; particularly human. Every human innately knows what other humans are supposed to sound like, therefore the modulation of the human voice causes the most instinctive unease and distress.
​
In my artifact, I use this technique on both the doll's scream and the music being sung outside the room during the period while the player is locked in the kid's bedroom.
This technique is very simple. There are certain sounds or kinds of sounds that are inherently scarier than others - even without much or any manipulation. These are sounds that activate subconscious survival instincts. The first obvious example is that screaming humans or animals will always trigger an uncomfortable response. Particularly and extremely disturbing are sounds of genuine fear or pain. There are poorly-understood characteristics to genuine screams that cause an autonomic, physical response in the listener. Fortunately the rare, uniquely talented vocal actor is able to perform a scream organically enough to outwit the subconscious and trigger the same result, allowing these types of sounds to be collected ethically. Infrasound is another type of instinctively terrifying sound - sounds so low they're felt more than seen. While not yet fully understood, it's believed these terribly low sounds cause discomfort and distress due to their counterparts in nature: earthquakes and large predators such as tigers and bears all produce infrasound.
​
In my project I was able to demonstrate both of these techniques simultaneously with the doll's scream. Mixed from a combination of human vocal screams (performed by a voice actor) and modified crocodile sounds, the shriek at once contains looping and overlapping human screams and real-world infrasound.
While the player is locked in the child's bedroom they hear a cacophony of stressful sounds including screaming, birds fighting, distorted infrasound, and more roaring out of the study, master bedroom, and bathroom.
Helplessness
The player is locked in the child's bedroom with the screaming doll, unable to interact or leave until a certain amount of time has passed.
While this technique falls more to the level designer than the audio designer alone, there's little that can make a sound inherently more terrifying than helplessness to change or resolve a sound that is currently scaring the player. If the sound designer is leaning into the other techniques, the player is already on edge and distressed with the sonic environment. The easiest way to magnify this discomfort is for the level designer to deny the player the ability to either escape or resolve any of the sounds. The feeling of powerlessness and lack of control is a potent multiplier for all kinds of horror, but particularly potent if the user feels they are unable to escape a terrifying audio experience.
​
In my project I demonstrated this by locking the player in the child's bedroom with the screaming doll. For the full duration of their time in the room, they're unable to interact with anything and are simply subject to a more and more distressing audio experience until they're released.
Conclusion & Defense
At the end of the project, I successfully identified and documented each of these techniques into a comprehensive thesis. This was a combination of a paper, a presentation, and the artifact. After collecting player feedback via surveys, I was able to confirm my theory that the use of these techniques alone could craft a traditional pacing structure. I successfully presented my findings and artifacts to the Design panel at SMU, and was awarded my Masters of Interactive Technology upon graduation.
Post-Mortem
What Went Well?
What Went Wrong?
What I Learned?
- Solo development of an Unreal 4 project gave me meaningful insight into areas of the engine I hadn't previously explored
​
- Gaining experience with blueprinting both Unreal's built-in sound support and Wwise events.
​
- Directing a pipeline of art assets from concept to final including providing actionable asset request spreadsheets and reference materials.
​
- Studying and successfully defending a unique and typically under-represented facet of design highlighted audio as a key component of level design to my professors and peers.
​
- Using film in addition to games as a source allowed the inclusion of a much longer timeline of references - and reinforced that often times film techniques can strengthen games and vice versa.
​
- Critically analyzing the mechanics of what makes a piece of horror art successful versus unsuccessful sharpened my analytical eye for studying efficacy in games.
- Better time management and prioritization would have left more polish time for the climax as the keystone moment of the piece.
​
- Without expert audio faculty at SMU, I had no professors to turn to for support or resources on the topic.
​
- The defense presentation was somewhat of an uphill battle, as I needed to define and explain audio terminology as I was also trying to use that terminology to describe techniques.
​
- A few of the players I sent the game to did not read the ReadMe, played with the game muted, and were unable to provide data for the survey.
​
- At the last minute I was asked to add an appendix to my document describing the effects of various level design decisions on acoustics, which was not immediately relevant to the research. I feel this request came from a lack of understanding about what my thesis really was about.
- Techniques for putting advanced ideas into layman's language for non-expert audiences.
​
- Feedback dialogue techniques between art and design to ensure a cohesive vision and project.
​
- The pros and cons of both Wwise audio and Unreal default audio, including a deeper understanding of when which might be preferable to the other.
​
- Critical analysis techniques for both films and games to extract what about them is working the best, then distilling them down into universal practices that can be applied to any project.
​
- Survey techniques for assessing how a player is responding to a game, including how to effectively frame survey questions to avoid biased answers.
​
- Blueprinting best practices and techniques to create reusable systems rather than creating bespoke blueprints for every individual gameplay object.