Marty O’Donnell is a man on the bleeding edge, being responsible for the music and audio vision for the stratospheric Halo trilogy from Bungie studios. I wanted to try and find out what this leader in the field of music and sound for games had to say about the subject and get some input for my then to be completed thesis. Around August of this year I embarked on a campaign of multiple emails to multiple addresses to somehow get in contact with Marty for an interview. After what seemed like a longer period than it probably was, finally in October a scant week before I had to hand in my thesis I spoke to the man himself via phone. Even though I approached him essentially out of the blue, Marty held no qualms about chewing my ear off for close to 90 minutes, to my own delight of course, and it is clear that music for videogames is a subject about which he is fiercely passionate.
In this, the second of the series, I ask Marty about his views on using a ‘granular’ live assembled music approach for videogame music, and he explains both why he doesn’t like that approach (for the time being) and what his own approach looks like.
Ben: I guess what’s most deceptive is, because Halo’s levels are designed so that there are multiple ways through it, but you go from point A to point B, you end up hearing all the music in the same order and maybe that contributes to the sense that you hear the same thing every time through. But I never got bored of the music in Halo either, so it’s obviously doing its job.
Marty: Halo 2 was a while back for me now, but I think the first level is the space station, right? If you got to the last encounter in the space station just before the end cinematic, pay attention to when the music actually starts, what the music does, how the music changes based on how you perform and then how the music actually seamlessly goes into the cinematic music, which is a more linear piece because I’m scoring something that’s cinematic… If you were to play through that sequence several times, try to go really fast, try to go slow, try to see what happens when the music in the level and the music in the cinematic go together, you’ll actually get several different kinds of recordings out of that.
Now it’s not like an entirely different experience, that’s not the point… it is still like, ‘that is the piece of music that plays there’, but it’s not a linear piece because different sections don’t happen in the same order and the way the music leads into the cinematic… it doesn’t [just] cross-fade it actually plays simultaneously with the cinematic music but it plays in sync with it in a way that changes depending on how you get into it.
B: So what would be your reaction to someone that wanted to do away with that sort of authorial control you have by mixing it up, and says ‘well lets set some parameters, and if the player health gets to this, do this’ (obviously that’s a simple way of doing it) do you think you do lose something if you take that step back?
M: Well you know it’s interesting because, as someone who’s worked with computer generated music or computer controlled music… and as you know in the music business and creating music, we use digital for everything… I remember back when there were music programs that were… basically lets generate music algorithmically. And I remember hearing a lot of that stuff and playing with all those things and… because I think I’m probably more of a traditionalist when it comes to what music I like, and what music I think actually evokes emotions and what music actually speaks to me, I think the composer still makes better choices than just a combination of random events.
Even when I was using algorithmic music programs, out of an hour’s worth of stuff that was generated I would find maybe 30 seconds of it that was actually interesting. So that’s my problem with that. I think that you might get some fun stuff, but I don’t think you get stuff that really speaks. Most of that stuff doesn’t end up telling a story musically, and I still think that the power of music is a storytelling power, and I might be wrong about this, but I still think that a composer tells stories better than computers do.”
However, if something is repetitive, even if it’s really, really great and you keep hearing it over and over again, and it plays back the same way every time – that is to me where boredom set in or it just starts getting annoying. Which is why when I hear games that use linear music that always plays back identically and starts to get repetitive, and loops... when I notice where the loop is I get really annoyed. That’s something I’m trying to eliminate. I really don’t want people to hear where the loops are.
B: Michael Chion wrote in the 90’s about audiovisual relationships in film and he talked about how if you strip the music off a piece of image, and you just place a selection of random other songs… some pieces of music work better with others. He said that in a few pieces… you will get a few moments of almost serendipitous synchronization between the audio and the video, so I guess where I see the potential for the live generated, granular, building the music up from individual notes, is the potential to pick out those points of synchronization and specifically hit them with the moment that you really want.
M: I would say… I have not been one of the guys who advocates the tiniest granular approach. There are some music engines and music for games approaches that’s very, very granular and right down to the individual sample level. ‘This will interact in this way’ and ‘this will interact this way’, it’s all possibly midi controlled, note generated everything.
There’s a couple reasons I don’t like that, number 1, I lose some of the fidelity that I like, I lose some of the live performance that I think is still essential… A midi flute performance just does not compare to a live flute player performing a melody. I don’t like walking away from something that has a giant history of success, so I’ve never really been an advocate of the high granular approach to writing music for games.
That doesn’t mean it’s the wrong way, it’s just that it’s my preference. The biggest problem I have with scoring a game is that if there is a sequence or an encounter or a moment or whatever it is… the most important moment is how it begins and then how it ends. Because it’s a game and there’s a human being interacting with it, what I don’t have any perfect knowledge of is exactly how long the entire experience is going to last.
So what I decided early on was, I can control when it begins, I can control when something ends, musically, what I need to do is keep the middle section malleable. And if I can do that without people knowing that it’s being malleable – so if they have a 2 minute experience some place then they get a 2 minute piece of music and they’re happy, but if somebody else play it for 5 minutes they get a 5 minute piece of music and they’re still happy because the beginning, middle and the end all correspond with what they wanted the experience to be, then I feel successful.
So it’s all about variations on the beginnings, on the ends, and being malleable with the middles. I keep trying to advance the way that stuff is manipulated musically and it’s sort of an interesting puzzle for me. I really enjoy it, I like doing it, I like composing music that I can kind of dissemble and say “what are different ways of telling the same musical story but making the middle something that is sorta indeterminate.” I don’t know if that makes sense but that’s the system.
In the next post, Marty talks about how close he’s coming with his own musical approach to realising a high level of ‘granularity’ in the music. I ask whether you can get away with more if you’re aiming for an ‘electronic’ sound, as opposed to orchestral and Marty talks about Rez & the potential for more ‘synaesthesia’ in games.