Spectral Analysis: Interview with Saki Kaskamanidis
Many thanks to Brad Dyck for contributing this interview. You can follow Brad on Twitter @Brad_Dyck
BD: How did you get involved in game audio?
SK: It all started back in 1995. I was in a band and our keyboard/synth player, Jeff Van Dyck, was working for EA. He was working on the soundtrack for NHL 96 and needed a guitarist to lay down all the riffs, so he hired me. Soon after that, Jeff heard some of my own compositions and thought that they would suit a different game EA was developing at the time called Need For Speed. Eventually I got that gig and for the next 4 years or so I was one of EA’s full time composers, mainly focusing on the Need For Speed and NHL franchises. At some point around the millennium, EA as a company was doing really well and the trend towards licensing music from well-known artists began. It was natural for us to feel that our careers were in jeopardy. I remember being warned “You know, Saki, are you thinking about doing anything else because your job is in jeopardy.” So I started getting into sound design. Interestingly enough, nobody lost their job. You either found something else to do, like sound designing, implementing or leading a project, or you left the company because you wanted to keep composing.
BD: What were you using to compose at EA?
SK: Logic. My first DAW was Logic 3.0, now it’s Logic Pro X, owned by Apple but back then it was under Emagic, which was a German company. I also remember using Samplecell which was maybe one of the first soft samplers to come out. It came with its own NuBus cards that you would plug into your Mac’s nubus slots, I feel so old talking about this shit right now (laughs). Back then it was all Macs all the time. Also around this time, and fortunately for me, audio streaming had been recently introduced to gaming. Jeff would tell me how he’d compose for games prior to that – I think he started in ‘93/94 and they were doing step sequencing. Using a lot of MIDI, Synths, really (and I mean really) small amounts of memory and step sequencing to program music. To me, if that was the case when I started, I wouldn’t have been able to deal with that at all (laughs). I didn’t come from that school. In many ways, I felt like an outsider because, at the core, I was just a guitarist. They hired me to compose rock music, which was my forte at the time, so I composed some pretty cheesy stuff back then. Eventually I had to expand my horizons and explore electronica, which I embraced. It was a little shaky at first, trying to learn how to use a synthesizer and all. As cheesy as some of my early rock songs were, my first attempts at electronica spawned some embarrassing stuff, but I got better and better at it.
BD: I know your resume is pretty long for composing, do you still do that right now?
SK: I do, yeah, but not full time. I just do it in my own time. When I was composing full time, I confess I didn’t know all the tricks of the trade. I didn’t know how others were able to produce such high volumes within tight deadlines. You know, you can’t be inspired all the time – although some people are, and lucky them (laughs)! But, if you’re not, you need a different approach, which I never really utilized. I strived for real inspiration all the time and as a result, I ended up burning myself out. Fortunately this all timed out around the time when licensed music was taking off so I got out of it and ended up composing on my own time. The stuff I’ve been doing recently is my personal music which I upload to my Soundcloud site for the public to enjoy and I also submit some of those compositions to libraries (like Pump Audio) where they get licensed for TV shows like Gene Simmons Family Jewels, Jersey Shore and Hard Core Pawn to name a few.
BD: Were you tempted at all to do any composing for Sleeping Dogs?
SK: I actually composed a few tracks for the Zodiac Tournament DLC pack but, other than that, no. I had a lot of other responsibilities on my plate. Because I was starting out with a new company as its audio director and working on a new IP, I had to get all the systems in place for everything: driving, ambiences, fighting, gunplay, dialogue, etc. etc. Plus put together an audio team, which peaked to around 12 staff members. So there was plenty to do. Besides, every composer has his strengths and weaknesses and Sleeping Dogs needed something unique, like a score that fused Asian with Western sounds together which wasn’t my forte. In hindsight, I probably would have pulled it off, but considering everything else I had to do, we had to go with hiring a full time composer to get the job done.
BD: Yeah, I thought it might be compelling for a composer to do a Hong Kong inspired soundtrack but I could also see it being daunting to do that type of style.
SK: The way I thought about it was that my role on the project was audio director and I had to give opportunity to someone else. Someone who would focus on the music and nothing else. I didn’t have the bandwidth to compose and deal with all of the other responsibilities. I know others who probably could’ve done it… but my brain doesn’t function that way. In the end, I assigned all the music sourcing to my right hand man, Rashid Hille. I said “Rashid, you handle the music, you’re going to deal with it.” Obviously I was in the loop on all high level decision making but otherwise I was really confident in Rashid. His previous project was working on the Need for Speed series at EA, and during his final years there, he was doing a lot of music sourcing and supervision so he had a good grasp on that stuff.
Sleeping Dogs was the most complicated project that I’ve ever worked on. I thought hockey was hard (from a play by play and crowds perspective). Developing any open world game is challenging. So that was certainly a huge learning experience. It was like working on, Need for Speed, UFC, Dead Space, and Tomb Raider all in one project.
BD: …and cockfighting
SK: Cockfighting, that was funny! Open world games are a huge technical challenge because you get into a lot of memory and streaming management. I was surprised it all worked out in the end. I was bracing myself for a really, really grueling final phase. But, when the project got cancelled three years in, I was partially relieved.
BD: Did you have a certain deadline that you were already working towards and Activision just realized that you weren’t going to meet it? Or you needed more time than they were willing?
SK: All in all it was just an issue of time. You can’t develop an open world game in three years and hope for it to be as good as GTA or Assasin’s Creed…especially on your first attempt. So Activision thought it wasn’t going to be good enough meaning that they realized that it wasn’t going to keep up with the ‘big boys’ (GTA and AC).
BD: They were worried about it coinciding with GTA?
SK: Yeah, they were worried about the timing as well which is a moot point now considering GTA V isn’t even out yet (at the time of this interview). Teams like Rockstar and Ubisoft have been working on open world games for 10 plus years whereas we were on our first. It was obvious the game was not going to succeed with its original release date. Before the cancellation, I was dreading the final stretch because it was shaping up to be a hellish finish. The team was bursting at the seams. To help meet the deadline, people were being thrown onto the project left, right and center. There was no room anywhere. Some people had their desks in the bathrooms! Kidding…there were a lot of spinning wheels, a lot of decisions not being made, and that’s what happens when you have too many people trying to solve a problem.
Then, all of a sudden it happened and it was a bittersweet thing. I know a lot of us were partially relieved and partially devastated because this thing that we worked on for about 3+ years was all of a sudden gone. Plus, It was a blow to the ego to have something you poured so much blood, sweat and tears into cancelled, “Oh, it’s not good enough.” You couldn’t point a finger at anyone, it was just circumstance. We needed more time, and anyone would’ve needed more time. Look at LA Noire: 7 years in development, GTA 4: 5 years in development…
I remember the 6 months after the cancellation when things were slow, True Crime was starting to become a distant memory. When we first heard about Square showing interest in reviving it, we were like, “What!? Are you fucking serious?” A lot of us were freaked out, “Oh, man, no, we don’t want to pick that thing up again!” but slowly it turned out to be such a great relationship with the folks from Square Enix. They handled things differently.
BD: Did any of the fundamentals change after it got picked up by Square, audio or otherwise?
SK: Well, for starters, the name changed from True Crime to Sleeping Dogs! But from the audio side of things we had to hire a new composer and start that whole thing from scratch. The original composer was Brian Tyler, who’s a well-established Hollywood composer that works on blockbusters. At the time, when he was composing for us, he had worked on all those XXX movies you know, those Vin Diesel movies. More recently he’s worked on Iron Man 3 and The Expendables 3. He was great to work with. I got to meet and hang with him when he came up for a visit (he’s from LA) and I remember him as a nice guy, humble and super talented. Awesome drummer. But, for legal reasons, Square wasn’t able to acquire his work from Activision. I wasn’t in the loop but it had something to do with Brian Tyler and Activision. So we ended up hiring a local composer by the name of Jeff Tymoschuk and he did as great a job.
BD: Were you able to do any of the hands on sound design for the game?
SK: Yeah, I did a fair amount of it.
BD: What sort of areas did you focus on?
SK: I focused on every aspect if you include all the DLC work. I scripted interactive music hooks, composed some of the music (for the Zodiac Tournament DLC), and was dialogue recording and directing. I also designed and implemented gun sounds, a few car engines in the early stages of the project, as well as the melee fighting. The most fun was designing and implementing the ambiences. They weren’t prebaked streams. Every element was independent and tweakable. For example, birds occupied their own ‘track’ and the time of day affected their behavior; they chirped more frequently during the afternoons and sparser at dawn or dusk, and, additionally, rain played a role in their behavior. All the daytime animals were swapped out and replaced with nighttime animals when it got dark. The crowds’ volume increased or decreased depending on the amount of people around. Like the birds, distant traffic was controlled by the time of day, so it would sound denser during peak hours and sparse at night. Stuff like that. It allowed for a very dynamic sounding system. We used Wwise as our middleware, which worked out great.
BD: I read the interview you did with Audiokinetic, I wanted to ask more about the external sources update they did. You mentioned that it assisted you with managing RAM, I was curious how exactly did it help?
SK: All of Wwise’s objects, like actor-mixers, random containers, arguments, switches and such cost memory and, if you have a lot of them – like our dialogue structure did early on – they can add up to quite a bit of memory, over 1 MB in our case.
BD: I can only imagine what your dialogue must’ve looked like.
SK: Yeah, it was nasty. In the end we recorded over 75,000 lines of dialogue but ended up using about 45,000. Thank God for external sources. The feature allowed us to store all of our assets outside of Wwise. So we were able to get rid of all the dialogue structures which ended up saving us over 1MB in memory.
BD: So it connects from somewhere outside of Wwise?
SK: Yeah, you have all your speech assets (wave files) in a separate directory outside of Wwise. In Wwise you have your ‘external source’ objects, which are triggered via events. Think of an external source object as a bus that a speech asset pipes through when called. Here’s a simple example; To trigger a line of dialogue from someone reacting to a gun pointed at them, an event will trigger when all those arguments are met. In this case the arguments could be something like “gun is pointed at character” and the “name” or “voice” of that character. All this is calculated in code so that it triggers the appropriate line of dialogue. In this case, the asset it chooses would contain arguments in the file name, like: sam_panic_03.wav. Sam is the name of the character, panic is the context and the 03 is the variation. When the file is chosen to play, it triggers an event in Wwise which contains an action to play the sample through an ‘external source’ object. An external source object looks and behaves like a sound object only it isn’t associated with any assets; it’s like a bus for the external wave file to play through. The cool thing is that you can apply all sorts of properties to the external source object, like its own attenuation curve, volume bus, effects, RTPC’s etc. etc. And you can have as many of them as you like and you can route any dialogue sample through any of them at any time. This gives you a lot of flexibility whereas before, a speech asset was locked to a speech object. If you wanted to route it through a different attenuation curve, you would have to duplicate the speech object and assign the different attenuation curve to the duplicate.
BD: What sort of sounds did you capture specifically for the game through field recording? You mentioned before that you went to Hong Kong.
SK: Yeah, we spent about 2 weeks in Hong Kong recording dialogue and ambiences; We recorded and implemented the sound of crowds, the pedestrian traffic signals which is very unique to Hong Kong and the door chimes for the 7-11s. Other ambiences like street musicians and Mah Jong interiors were implemented as well. Most of the Cantonese that you hear in the ambient world was recorded there as well. We employed Eddie Chung, who runs a professional studio and is also a member of a hip-hop group that’s fairly well known there. Eddie was a valuable contact. Great guy to work and hang with. He showed us around town to where the good restaurants and bars were, and also to some of the rougher places where we ate some things that I normally wouldn’t eat. Anyways, Eddie had a line on a lot of good voice actors and you can hear them all over the game.
In the studio, the voice actors actually re-wrote a lot of the script because the script was written by Canadians living in Vancouver trying to be Hong Kong gangsters. It didn’t translate well so Eddie and the actors fixed it. Yeah, it slowed down the sessions, but in the end it was for the better! A lot of the actors were young hip kids who had an ear for the street and were able to inject some authenticity to the dialogue. It was funny because they were yelling out crazy stuff and laughing hysterically. I’m saying, “Ok, what’s going on, what are you guys saying here?” And Eddie answers, “Aah, man, I don’t know if this’ll be allowed in the game, some of this stuff is really rude but it’s authentic, it’s authentic!” And I’m saying, “Okay, let’s keep it, let’s keep going.”
BD: You can get away with it more when it’s in a different language.
SK: Yeah, that’s the thing. I remember some of the producers being concerned about it. But because it wasn’t going to be subtitled, and the material was used as background dialogue, it was OK.
BD: Was the actor who played Wei from Hong Kong as well?
SK: No, he’s an American of Korean decent. His name’s Will Yun Lee. He’s based out of L.A. and doesn’t speak Cantonese. There was a bit of concern for that but, eventually, we realized that it fit Wei’s backstory; he spent a long time in San Fransisco before returning to HK. Will Yun Lee was the fifth voice actor for Wei. There were four actors before him. I have to give credit to PCB productions and its owner Keith Arem. He was instrumental in helping us achieve the high level of quality in the dialogue. He’s well connected and has great relationships with the actors, which goes a long way. Also, because Activision had such deep pockets, it allowed us to re-record a lot dialogue in order to get it right including replacing the main character as Will Yun Lee from his predecessor.
BD: That would be a lot of dialogue to re-record.
SK: Yeah, it was, and that wasn’t just for Will, it happened to a lot of characters. Like Jackie, same deal, there were 5 or 6 actors before we finally settled on Edison Chen.
BD: Did you attend any of the sessions where Will would have to sing the karaoke songs?
SK: Guess what? That wasn’t him singing.
BD: Really? It sounds like him.
SK: I know, the funny thing is we tried to find someone who would sound like him and it ended up being some local guy here. I was at those sessions and it was hard. We asked our casting agent, “Okay, here are the songs, give us a guy who can sing well.” And she said “No problem, I know the perfect person who’ll knock these out of the park.” And he did a great job but after a while it was really tiring for him. Some of those songs are really challenging to sing, like “Take On Me” by A-ha and “Girlz Just Wanna Have Fun” by Cindy Lauper.
BD: One thing I liked about it was that it sounded like karaoke, it didn’t sound too good like he was doing crazy vibrato or anything, it really captured how it actually sounds.
SK: Thank God for that because if it had to be perfect it would have taken a lot more work to edit his voice. This happened late in the project when my staff was reduced. A lot of the work fell on me and I had to put that stuff together. At first I was going for perfection and I realized, “Fuck, this is going to take way too long.” So then I thought, “Wait a minute, this is karaoke, who cares, just leave some of the imperfections in there.”
BD: When you were in Hong Kong, did you notice any distinctive sounds?
SK: When we were recording ambiences in Hong Kong, there was this one time when I shut my eyes and listened to the city, to hear how it sounded like compared to other cities. I realized that, other than the spoken language and perhaps some music playing in the distance, Hong Kong sounded like any other city. Of course I wasn’t close to a cross walk because if I was, the sound of the cross walk signal would’ve stood out. They’re really loud in Hong Kong and they have a peculiar sound to them, like a loud ticking that varies in frequency depending on the signs state.
An important goal that the development team tried to achieve was to recreate the density of Hong Kong. One of the memorable experiences that I took away from there was the density. Hong Kong is crowded. Definitely the most crowded city I’ve been to. It was a challenge to recreate that especially from a visual point of view because the game was only able to afford so much memory for ambient crowds. Our producers were relentless in trying to push us to achieve the desired level of density.
BD: Try to make the mix to feel more crowded?
SK: Yes. We had a lot of great production feedback. I’ve worked with many different producers in the past and none gave half as much attention to audio as the Sleeping Dogs producers. It was a breath of fresh air. Of course it wasn’t always fun and games, sometimes it would get challenging especially when I’d get countless requests to make things louder and more dense! Sometimes I’d get requests to make things sound busier when visually it looked sparse.
BD: You’ve got to put more people in.
SK: Exactly, a lot of the times it’s like, “What do you mean? There’s only 5 people here. You want it to sound louder and denser? Add more pedestrians and the ambient system will reflect that!”
BD: Did you do any recording outside to get an outside sound?
SK: Yep, Rashid and I did a couple of midnight sessions out at a park near UBC where it was really quiet. We recorded climbing over fences, getting the fence rattles, hand slaps hitting metal, stuff like that. Obviously background noise is your biggest enemy and we didn’t have the budget to rent a foley studio at the time. So how can we do this on the cheap? Well, unfortunately it’s got to be late at night and near the forest basically (laughs). So, it worked out pretty good.
BD: How many cars did you put in the game?
SK: About 20 cars were recorded and 10 or so motorcycles. The recording sessions took place in L.A. and Rashid and I attended the sessions as supervisors. We hired a well known sound recordist by the name of John Fasal. He records sounds for movies and has a lot of experience recording cars. His resume is quite impressive, having recorded sound for well over 150 movies. He’s worked on the Beverley Hills Cop series, Rocky V, Fifth Element, The Royal Tenenbaums, War of the Worlds, Life of Pi…
To record engines in the old Need for Speed days, we would wire up mics to the car at various locations like under the hood, by the exhaust and inside. We’d route the mics to a Tascam DA-88 8-track digital recorder that was on the backseat. We’d drive the car on the street, late at night. You get some great results but also a lot of issues with wind and road noise, melting mics in the engine and damaging cars from applying the brakes too much. One of the most important qualities you need when recording engines for a loop based model is to maintain a steady rpm for as long as you can. In order to achieve this, one needs to apply the brakes as you drive. One time, when I returned a rental car, I realized that all 4 hubcaps were missing. This was because they melted off from the heat of the brakes.
Once we were introduced to Dynapacks our lives became a whole lot easier. Dynapacks are like treadmills for cars only they apply load to the vehicle. So you can record an engine under load while stationary.
BD: There were a lot of great radio stations in the game, how did that come about with labels like Warp, Ninja Tune and Roadrunner?
SK: Tsunami music was the 3rd party company that helped us put together the radio stations and their selections. They were awesome partners to work with. It was Square’s idea to use Tsunami, I believe, and Tsunami’s approach was to license record labels and name each station after a label so each stations music selection contains music from that label. It’s a more cost effective approach than creating a fictional radio station and populating it with songs from different labels.
BD: It was a package deal.
BD: It had a lot of really good Hong Kong stations too. There was that one song that has a lady kind of screaming, what was that?
SK: That was Chinese opera fused with bad rock.
BD: Was that a real artist?
SK: Yeah, It was Beijing Rock by Yiming Xie. We got this library from some company that has a lot of authentic Asian music so we got to choose from a lot of great stuff and there was a lot of wacky shit like that. We used a particular part of that song in the scene where Mrs.Chu chops Dogeye’s…
BD: I remember that, I wondered if you chose specific radio songs to be in the background for certain cut scenes or missions.
SK: Yeah, some songs were tailored for specific scenes, otherwise the music was random.
BD: What sort of changes do you think will be possible in terms of game audio for the next generation?
SK: 3D audio. Have you heard the virtual barber?
SK: What did you think?
BD: It was the most realistic sounding surround that I’ve heard but it does necessitate headphones, doesn’t it?
SK: Yeah, but I know that the new consoles will include headphone jacks on their controllers. When it comes to a rich surround experience, we need to abandon the traditional 2D surround presentation formats in favor of 3D audio.
One method for recreating 3D audio that’s been around for a while is HRTF (Head Related Transfer Function). HRTF creates 3D audio imaging by utilizing parameters like volume, filter, delay and reflections in complex algorithms. I remember in the late 90s, Aureal Vortex included the tech with their speakers. They had a demo where you position yourself in front of the speakers to listen as you watched an animation of a bee buzzing around your head. It worked pretty well even without headphones (yes, it is possible to experience 3D without headphones).
Once 3D audio in gaming gets to the level of quality as demonstrated in the Virtual Barber people will look at 5.1 systems as a thing of the past. I’ve been looking around to see who’s been using 3D audio in gaming and have found that there’s quite a bit of movement going on. There’s an i-Phone game called Papa Sangre that features it as its sole gaming mechanic. It’s a game you play using your ears. It’s pretty cool. I tried it and thought its imaging isn’t as realistic as the Virtual Barber, but we have to put it into perspective that this tech is young and will only improve as long as there are people eager to improve it.
I’m sure you’ve noticed that it’s been a tough time in the industry lately. Not only are we entering a long overdue console transition, add to the mix all the new gadgets that everyone has, smart phones, tablets and the whole new model of free to play. It seems like budgets for games have polarized. You’re either going to get GTA V or Candy Crush.
I look forward to working with the new consoles which will provide more freedom and new challenges. It’ll be a sound designer’s dream compared to when I started off.
BD: Yeah, no restrictions at all.
SK: Yeah, and in ten years time we’ll be looking back at these new consoles the same way we look back at the PS2 and XBOX.
Saki’s journey in sound began with music and the guitar. Since then he’s explored music composition, sound designing, building synthesizers, playing music with people and working with game developers as an audio director, currently with United Front Games. He live’s and works in Vancouver, Canada.