Terrifying film music mimics alarming acoustic feature of human screams
The Journal of the Acoustical Society of America (JASA)
Authors: Caitlyn Trevor, Luc H. Arnal, and Sascha Frühholz
In this episode, we interview Caitlyn Trevor from the University of Zurich about horror film music mimics human screams.
Read more from The Journal of the Acoustical Society of America (JASA).
Learn more about Acoustical Society of America Publications.
Music Credit: Min 2019 by minwbu from Pixabay. https://pixabay.com/?utm_source=link-attribution&utm_medium=referral&utm_campaign=music&utm_content=1022
Kat Setzer 00:06
Welcome to Across Acoustics, the official podcast of the Acoustical Society of America’s Publications office. On this podcast, we will highlight research from our four publications, The Journal of the Acoustical Society of America, also known as JASA, JASA Express Letters, Proceedings of Meetings on Acoustics, also known as POMA, and Acoustics Today. I'm your host, Kat Setzer, editorial associate for the ASA.
Today, we have a special Halloween themed episode of Across Acoustics. Joining me is Caitlyn Trevor from the University of Zurich. We will be discussing her article Terrifying Film Music Mimics Alarming Acoustic Features of Human Screams, which appeared in the June 2020 issue of JASA Express Letters. Thanks for taking the time to speak with us today, Caitlyn, how are you?
I’m good. Thanks for having me. How are you?
I'm good. Um, so first, just tell us a bit about your background.
Well, I have a music background. So I play the cello and I did some composing; I used to really want to be a film composer. That was my goal when I was younger. And then I was in school for music, and I learned about music cognition research in my music theory courses and that really fascinated me. So then I went to grad school to do that kind of work, to get a PhD in music theory, and do music cognition research. And yeah, now I am here as a postdoc in Switzerland, and I research music and emotion. So you can see if I was really into film music, which is so good at inducing emotions and portraying emotions, that then that's where my research ended up going.
What made you decide to study horror film music and the correlation to human screams?
Well, as I mentioned, yeah, I'm really into film music. And I especially like horror films, I find them really interesting. And horror film music didn't seem to be getting that much attention in the work that I was doing. I mean fear was studied but didn't seem to be studied super in depth. And I wanted to really look into the subject and see if there was more to be said, and particularly if there were more categories than just scary music. I think there's music that causes terror, and there's music that causes anxiety, and they're very different. So yeah, I saw kind of a gap in that area that I was excited to fill and so I went for that, that type of research.
That's so interesting. So the premise of your work in this study has to do with the concept of mimicry of music? Can you explain this concept a bit?
Sure, yeah, there are a lot of theories out there on how music causes emotion and how it induces emotion and portrays emotion. But one of them that I find interesting is that music sometimes might mimic aspects of vocal communication, in order to portray or signal emotions to the listener. So there's one example: there's this kind of half step motion and music, it sounds like *hums notes*, this really small fall. And some theorists say this kind of mimics a sigh, and it's used for sad music. So it's kind of portraying this longing, sadness and sort of mimicking the sound of a sigh, which is a vocal cue. So yeah, this sort of music mimicry is one area of music and emotion that you can study, essentially.
Okay, that's very interesting. So what acoustic features characterize the sound of the human scream?
So a human scream is very, very loud. It's very rough sounding, which is sort of like this scratchy kind of raw, harsh sound quality, and it's also really high-pitched. And so two of those things are not super unique to screams. Like when we laugh, it's very loud and high pitched, and when we maybe sing, it can be really loud. But that harsh, rough quality is really unique to a human scream.
Oh, that is really interesting. Okay. So what are ethological signals?
So essentially, they are signals that are multimodal, so there's a visual component and also an audio component. And they are between species. So they're a signal that you create at a member of your own species (this is across the animal kingdom) that is meant to provoke an intended response. So in humans, you know, some of them might be laughter. You know, when we laugh, we smile, there's an audio component, and it's a very clear signal to our fellow human being that we’re like happy or enjoying whatever's going on. Crying is an ethological signal; you have tears, you have the sound of crying. So these are ethological signals. And it’s something that we sometimes might mimic with music. Actually, when I was talking about that part before a little bit, a sigh could also be kind of an ethological signal as well, because it conveys some clearer emotions, like disappointment, or boredom, or sadness. And I guess it also has a visual component: maybe your shoulders kind of go down, your face looks a little sad as you sigh. So it's something I'm interested in because ethological signals are something that I think music mimics occasionally.
Yeah. Okay, that makes a lot of sense. Um, so going back to roughness and such, how would you expect those acoustic features to be mimicked in music then?
Well, it depends. So in this paper, in particular, I was looking at kind of what we call a timbral cue. So timbre is the quality of a sound. So if you can think of, if you hear a horn playing the same note, and a violin playing the same note, they don't sound the same even though it's the same pitch technically, they sound different and that's because they have different timbres, so they have a different tonal quality. And so sometimes when you're looking at if vocal cues or ethological signals are in music, you can look at timbre cues like roughness, that's a timbre kind of cue, but you can also look at other things, like loudness is one you can look at. You can look at the tempo of the music you can look at the articulation. So for example, once I was kind of looking at if funny, or not funny, but humorous music mimics laughter and that has a really, when you think of how laughter sounds, it has this punctuated sound. This “ha ha ha” is a really sharp punctuation that's much different than that slower sigh *hums note* that's not like punctuated, right? So that's, that's an articulation thing you can look at. Yeah, those are a few examples.
Okay, that's so cool. So then how do you test whether horror music mimics human screams?
So I looked at a timbre cue, which was roughness. Roughness can be measured in a lot of ways. First of all, roughness, again, is that sort of grating sound quality I mentioned before, this really harsh timbre that screams have that no other signal has. And it can be measured in a few ways. But the way I measured it was using the modulation power spectrum, which is a pretty complicated acoustic analysis that essentially kind of gives you the timbral fingerprint of a sound. And there's a really specific area of that spectrum that is, that if there's power in that area of the spectrum, that communicates that that sound has roughness. So I was looking at if scary music has that same kind of roughness fingerprint as human screams. And then I also did a behavioral study as well, which I can detail more if you would like.
Awesome! Sounds great. Well, let's first talk about that first study. You mentioned that you used the modulation power spectrum for your acoustical analysis. Can you explain what that is and why it was useful for your analysis?
Um sure I can, I can go into it a little bit more, although for me, the math is still a little difficult. But essentially, it's a fast Fourier transform done twice. So when you do one, one time, the fast Fourier transform takes a sound... And all sounds have a huge range of frequencies active at the same time, even if you sing one pitch like *hums note*, if you actually look at the sound using a fast Fourier transform, you can see that there's a lot of frequencies happening in there. It's not just one active frequency, there's many. So what the fast Fourier transform is, it sorts sounds into all these frequency bins, essentially. So it tells you how much power is in each frequency level. So you can see, across time – it looks really cool actually, especially live – you can see the pitches changing as maybe you use it to look at a bird call or something like that. And so when you do modulation power spectrum, you actually do it again. So it's sort of this double fast Fourier transform, which then you get that kind of fingerprint instead of over time. It's not over time anymore, and it's showing you the temporal modulations and the spectral modulations. Yeah, I wish I could distill it even more, but I think the best way to think about it is that it's sort of this timbral fingerprint, essentially.
So in the second study, you tested whether screams and scream-like music are perceived as sharing similar emotional qualities. What did that involve?
So, there are a few ways you can measure how people perceive emotions, and one of them that's really common is to use these dimensional scales. So there's two, there can be three, but we use two of them. One of them is called valence. So it has a positive and negative side to it, it's a scale. And valence, essentially, if it's negative valence, that means really negative emotion. So it can be like really, really sad or really angry, that's super negative valence. Then the middle is sort of neutral. And then positive valence is for positive emotions, like very happy, or joyful, or excited in a good way. And then we have arousal, another dimension, which can also be called I think activation. And this one is about how intense the emotion is. So really high arousal is really intense emotion. Low arousal is very, very calm, not a very active emotion. So when you use these two scales, you can theoretically kind of map out all the emotions using the two. And so when we were testing, I had this scream-like music from horror films, and then the human screams, the sound of them. And I was having people listen to those and rate those two dimensions. There was also, to counterbalance that, there was non-scream-like music and also non-scream-like vocalizations: these sort of held sounds like *hums note* instead of a scream.
Okay, that's so cool. Um, so what did you end up finding out about horror music and its mimicry of human screams? Did the results support your hypothesis?
Yes, the results were really supportive of the hypothesis, which was nice to see. The scream-like music is very rough as well. Comparably in a way to human screams, it's a little bit less; which I was surprised actually, because when I listen to scary music, to me, some of those examples sound incredibly rough and grating. But I think there must be a difference. Yeah, it's not quite in the same region, in some ways, as the human scream. And you can, I mean, when you listen to scary music, it doesn't sound, you don't mistake it for human scream literally almost ever. Sometimes it's awfully close, but you can tell it's not a voice. And I think that's the difference we're seeing in the data. And in terms of perception, we see something similar also. People rate scream-like music as not quite as negative or highly arousing as a human scream. So it seems that the vocal signal is still having that priority spot of extremely alarming, which makes sense, because it's the one that we really, really need to listen to and react to.
So then, do you have any future research related to this topic planned?
Yeah, this study is part of a bigger project where I'm kind of studying scary music very, I don't know the right word…not wholesomely… comprehensively, there we go. Yeah, comprehensively. So one of those projects is I am creating a database of scary film music. One half will be all anxious music, and one half is panic-inducing music. So I'm listening to, well I already did this part, I was listening to tons of horror movie soundtracks, and like taking excerpts that really well reflect these emotions. And then we had people listen to them and rate them in a similar way to what I described before, so that we can sort it into the best excerpts. And these can be used then for research either as stimuli to scare participants while they do some other tasks like decision making, or I don't know marketing, or, you know, who knows, there's a lot of ways emotional stimuli are used. And also, I think it could be useful to look at the acoustics of this to really kind of, I guess, define how music communicates these different emotions, in terms of musicality, and also in terms of these timbral features. So yeah, that's one part. And, yeah, I hope that I can use the database then to look into other features that are present in scary music.
That is so fascinating and fun. Well, thanks again for speaking with us today, Caitlyn. It's been so much fun learning how music for films is composed to mimic how humans express emotions. I definitely need to go watch some scary movies to see this concept in action. And yeah, thank you!
Thanks so much for having me! Really fun to talk about these things.
Thank you for tuning into Across Acoustics. If you'd like to hear more interviews from authors about their research, hit subscribe and find us on your preferred podcast platform.