Interview with Girl Talk

Photo by Douglas Stewart

In January 2010 we caught up with long-time AudioMulch user Gregg Gillis, A.K.A. Girl Talk.

Gregg performs and remixes his mash-ups live in AudioMulch, which he famously runs on a laptop covered in cling wrap to protect it from the sweat generated by his performances. To date Girl Talk has released five albums and two EPs made with Mulch, so we thought he’d probably have some interesting ideas to share with fellow AudioMulch users.

Part 1: Gregg Gillis talks about AudioMulch

We met up with Gregg in his hotel room while he was here in Melbourne to play at the Big Day Out. This intro video covers a bit of background and gives Gregg an opportunity to say what he thinks of AudioMulch. We get into the body of the interview in parts 2, 3 and 4, where we cover some of the different aspects of his music-making process.

Jump to transcript for part 1


Part 2: Gregg talks about the musical elements he manipulates when performing live

In this part of our interview Gregg loads up AudioMulch to show us what kind of parameters he manipulates during a performance, the techniques he uses to manage loops and keep control of his set, and tells us why he thinks remaining a little out of control of loop triggering helps keep things fresh. The segment also contains a patch that features in the album Feed the Animals and some insight into how Gregg works with drum sounds.

Jump to transcript for part 2


Part 3: Gregg discusses his compositional process

In this part of our interview Gregg gives us insight into his compositional process. He demonstrates an approach that ranges from rather improvisational-style processing in AudioMulch to cutting up beats by hand in a sound editor. Gregg also explains his ideas about creating subtle details that he uses to tie the disparate elements of an album together into a coherent whole.

Jump to transcript for part 3


Part 4: Gregg describes how he transitions between songs during a set

In the final part of our interview Gregg reveals how he uses AudioMulch’s network sync feature to smoothly transition between songs during a set. We learn how many AudioMulch patches he uses in a set and how he manages to juggle so many samples without overtaxing his computer.

Jump to transcript for part 4


Many thanks to Gregg for spending time with us and to his manager Jessica Linker for coordinating the interview. Big thanks to our friend Matt Laing for shooting and editing the videos.


Girl Talk AudioMulch Interview Part 1 of 4: Gregg Gillis talks about AudioMulch

K: Katharine
G: Gregg Gillis (Girl Talk)
R: Ross Bencina

K: Thanks very much Gregg for inviting us into your bed...while you’re in Melbourne.

G: No problem.

K: I didn’t really expect to be in bed with you, but as long as we don’t tell your girlfriend, I’m sure that’s fine. Okay, we’ll cut that bit out.

G: Haha. She’d be pumped.

K: Oh! sounds like she’s a very accommodating woman.

G: Yeah well actually, one of the more well known Girl Talk YouTubes filmed me playing AudioMulch on our couch while she was sleeping. It looks totally staged ‘cos it looks like I planted this woman there, like this passed out thing, and all the YouTube comments kind of reference it; "Oh how did he get that slut in there passed out drunk or whatever?", and it was just my girlfriend sleeping on a Wednesday night at like 8pm. That’s kind of always been the big YouTube hit thing, so this is almost referential to that whole thing.


I definitely think it’s not weird anymore to see someone play with a keyboard, or sampler or laptop or whatever and make a show out of it.

I like to get in the crowd as much as possible and I like to do the classic interactions with the crowd, but yeah technically the way I setup my AudioMulch template to perform live it’s more beneficial for me to be clicking the mouse every three seconds as opposed to like dancing on someone’s head or something.


I think it’s also just a comfort level thing too for me at this point. I’ve performed so many shows on AudioMulch and you know, it’s intense sometimes when you get in front of a crowd and there’s a thousand people there. If the computer skips or the program like glitches up, it loses that whole image of being a professional at what you’re doing and also kills the entire vibe of what’s going on. In the past few years I have never had an issue, knock on wood, with AudioMulch. Playing live, it’s just become comfortable and I’ve never actually had something go wrong.

I think once you put in the number of shows and it’s like oh, we did two hundred shows straight and it’s no issues. It’s like, this thing works.

Naturally I think it’s also cool because people are always like, "Oh, you run Ableton!", and I’m like, "No I don’t." So it’s also like this mysterious thing I think to people who maybe haven’t heard of it. But it’s been stable, it’s what I’m comfortable with, and I think it’s a good product.


K: What’s your singing voice like? I mean you mentioned processing instruments live. Would you extend your music into...?

G: I wish! If I could sing I would be singing every show. I just can’t sing. Occasionally I do a Nirvana cover of the song Scentless Apprentice and it’s a bit more of a like screamy sort of one where you don’t necessarily have to carry a specific tune. Maybe if I got to expert level AudioMulch and I could actually process my voice to not make it sound ridiculous...


K: Do you have any tips for AudioMulch users, particularly those who perform live with it?

G: You know, I think the way I use it is very specific. Like I said, I want to get through samples at a certain rate so that’s kind of the way I use it. But I think using some of the applications (contraptions) and importing VST’s and things into there, it can be a really fun area. It’s funny, because a lot of my friends who fool around with it use it a lot more in-depth than I do, and it’s crazy to look at their patches and see what’s going on. I just think it’s kind of like an endless world. I think it’s especially fun if you want to fool around doing experimental music to run like you know, actual audio through the computer playing an instrument or something like that. Just processing sounds and kind of making that sound palette I think is a great thing that I don’t use it for as much as I used to, but I think that’s definitely an amazing function.

I think it’s pretty straight-up once you start fooling around with it. I think it’s pretty intuitive. Like I said I got into it without really knowing any other piece of software. I think with any software it’s a matter of getting into the groove of it and kind of being comfortable and familiar.

Girl Talk AudioMulch Interview Part 2 of 4: Gregg talks about performing live

K: So we’re curious about what you do in live performance - you mentioned a little bit about that before - but which parameters in the patch do you actually change?

G: Basically, I use all LoopPlayers and Mixers. I don’t know how you refer to each of these two different mute functions, but I use the actual mute on the LoopPlayer and occasionally the mute on the Mixer. For me, the mute on the LoopPlayers, I guess based on how large of a buffer you’ve set, has a certain lag to it and I’ve grown to use that lag as like an ultimate function of it; to be able to drop multiple ones at the same time. I think in the beginning I wasn’t even necessarily thinking about that lag, but now the way I use it, it’s almost like the way you play the instrument. I’m kind of used to that specific delay on my computer. How fast it runs, how quickly that drops.

[Ed. muted/unmuted LoopPlayers will always stop/start playing at the start of the next bar]

Occasionally the mutes on the mixers are crucial if I have to actually mute something at a very specific time that might be in between bars or something like that. For more straight up percussion loops and things that are perfect, I actually use the mute functions on the individual samples. During the show, it’s a funny way to think about it, but the loops are kind of constantly going so I keep track in my head of where I’m at. So, it’s like this sample...‘cos some samples might be larger than others; one might be 32 bars. If you’re running a bunch of loops that are like two bars, and all of a sudden you want to drop an acapella sample that’s 32 bars, it’s very easy to get lost on where you were and when to start that. So occasionally, I just have audio cues within it and it’s been a new thing recently where I’ll have a loop that’s 95% silence and just a drum fill at the end. So I’ll cue that one and if I’m getting lost within the set, I have them kind of as - ‘cos sometimes it’s just flowing and it’s like the timing’s keeping up, nothing’s going even ‘cos I do this over and over and over, the timing of it every night just changes naturally because I don’t click perfectly and you know, I kind of get mixed up in it. So occasionally the set will be going and I’ll be like, "Where the hell am i?", I’m lost within the time structure. So I’ll cue the loops with the audio cues and say, oh when I hear that drum fill, then I drop the next sample. So that’s basically how it goes.

The way I assemble it, I rehearse it and go over it and definitely there’s always improvisation in terms of...I have more stuff usually loaded than I want to use. Just in case you’re at a venue and "Oh, it’s just not heavy enough. I need to add more kick or, I need more clap!" So that’s kind of the level to which it’s improvised; just on the percussion and things like this. As far as the organisation of ‘I stop this sample now and go to this sample then’, that stuff I rehearse and go over and over and over. But naturally within the set, some nights I screw up more than others so that changes where it goes.

Sometimes nights I’m fixated on one thing, I’m gonna play that with more repetition. Other nights I’m gonna just skip over that. So that’s kind of how it goes down.

K: A lot of your loops are quite short so you’ve got a small, low level of granularity...? So you’re going to’re remixing things at a really compositional level.

G: Yeah, I think that the shortness of it allows you to be in total control in terms of...I feel like sometimes if you would have a loop where the song’s building and building and building...if you were to get lost within your time and you would drop that in the middle, as opposed to having each of those parts isolated and you can kind of build it yourself.

So if there’s a song...I mean I can give you an example. This is just really basic where it’s like from the beginning of that album, like the Spencer Davis Group’s going to be hard to hear here, let me turn this up. So this is like just one sample [Plays "I’m A Man" intro rhythm loop]. And I like to have that isolated from this sample [Plays "I’m A Man" intro loop with organ riff]. This is like one sample; basically this loop. So you can actually perform that now and actually [drops organ line in and out] cue it when you want it.

On top of that it’s kind of like all the drum things I like to have pieced apart. It’s like on the album you have this Roy Orbison sample [Plays "Pretty Woman" intro snare vamp]. It’s just those drums, and that’s just a tiny, tiny snare loop. Then on top of that you have like this clicking sound [Plays filtered quaver blips loop] which is something I just processed through AudioMulch through the PulseComb or something like that and chopped it up. So that’s just one loop...and this is like a sample from Afrika Bambaataa [Plays sample] that I distorted in AudioMulch basically. Then this is a kick drum [Plays crotchet kick loop]. So I just like to have all those isolated, so like live...this is the beat with those four playing simultaneously [Plays layered sequence]. Just small things like getting rid of the Afrika Bambaataa thing changes the energy of that a lot. So a lot of times it will be just that and a bass and then when I kick into the organ, I’m gonna want to cue the Afrika Bambaataa and that’s when it comes back to that lag where I can cue up a few samples simultaneously. I’ll start this like...[Plays organ loop]...always kind of filling in.

For me it’s like as real as doing a collage live can be...and then you would have like a vocal sample on top of that.

K: So you don’t change the number of bars and change the pitch?

G: A little bit...occasionally and sometimes even just with the overall BPM, like sometimes I want this to be a little faster or slower. Then occasionally, I like to go from half-time samples - I’m trying to think of an example here - where you would just cut it in half. But yeah, sometimes just small things like...if I chose to...if I was going to transition from this part [Playing layered rhythm loop again] into something that was like half the speed, I could kind’s gonna be hard to hear the kick drum [searches through loaded loops], but you can sometimes mute things out [Mutes Bambaataa loop], slow it down, double it up, [changes Bars duration from 1 to 2 on Bambaataa LoopPlayer then unmutes, now playing at half speed] and then transition into something that was like half the tempo [Unmutes "Players Anthem" acapella] sort of thing.

I used to do more PulseComb and live processing with some of the internal features in AudioMulch live, but not anymore. It’s like the pace that I want to keep dropping samples, I can’t really do any of the processing live. Well I could, but I want to keep moving so quickly and it’s...things are isolated that much, my hands are full, so it’s tough for me to just be like tweaking the reverb on a sample, ‘cos it’s like I gotta be dropping like five more things.

R: I just wanted to ask a question about the overview of the patch. Can you give us sort of an idea of...

G: Sure. Basically in each template I kind of have an understanding of how much my computer can handle in terms of samples. So I can just build up these things where I’ll do eight LoopPlayers into a Mixer and just have a bunch of those. Like I said, it looks a little sloppy over here [Scrolls down Patcher Pane] but basically every ten minutes of the set will be mapped out here [inside a Mulch document] and like I was mentioning earlier, sometimes I’ll prepare more than I want to have so I’ll skip over certain things; this or that. I just try to keep it as uniform as possible over here [In the Properties Pane]. Occasionally I’ll layer up some other samples if I need more than eight loops to execute that particular part.

But it’s kind of the way it’s organised in my head which I think has gone on to heavily influence just the way it is on the albums; just in terms of the amount of samples and this sort of work pace. Of course I don’t ever want to think of it like song A, song B. I want to keep it all fully integrated. But, just as far as like a main vocal theme or main melody theme, I kind of execute those on each of the Mixers. It’s kind of how I keep it organised in my head.

K: So obviously this is AudioMulch 1. Do you have everything you need in AudioMulch 1? You’re not planning to switch to version 2?

G: I haven’t been messing around with 2. Just because like I said, I might be out of my mind or insane, but I just like am so just constantly busy, I don’t want to fool around. I want to just make the music and prepare it. But I would like it. I haven’t really fooled around with it much at all. You know there are some small things that, like I said, I feel that some of the things that hold the program back, some small things are features to me at this point.

K: Like what?

G: So it might almost be a bummer if they’re resolved. Like the whole thing with... each of these loops doesn’t have an internal clock on it. So which is like, not to talk about the competition but Ableton will have an internal clock for a sample so you will never have to have that audio cue to know when a sample is coming up.

R: You mean like a display on it to show where it’s up to?

G: Yeah. For me...if I had that I would never be like, "Oh I’m dropping the sample, where is it coming in?" But I feel like that’s part of the excitement at this point. Like I said, I’ve just grown to have this instrument and I like when it messes up. Sometimes it’s very interesting. I think you know, a set where it just flows perfectly might even be bland. ‘cos even the mistakes aren’t even like that the beat dropped at a wrong BPM, it’s like more just samples coming and going at different times. So, I don’t know if I would want to resolve that at this point; I feel like it’s definitely a feature for people.

Especially the more well-known segments from the album’s like Notorious B.I.G. Juicy verse from the Nightmare Problem. People are very familiar with that song to begin with but I think the mashup I have on there is one of the more well-known things I’ve done. So, a lot of times live when I’ve dropped it at the incorrect time it’s awesome to me because it’s like people are like "Ohhh", kind of bummed out, but when it loops around over and starts over it’s definitely something that wouldn’t have happened if that would have been executed flawlessly. So...I kind of like that!

Girl Talk AudioMulch Interview Part 3 of 4: Gregg discusses his compositional process

G: You know, it’s almost just by chance I would say, most full mixes I run would have maybe eight loops going at once or something. So just because of that, you know each time it’s almost like okay...and on the albums you hear it. I want everything to be...I want it to be one long composition but naturally like drums, it’s like these little sections and they keep moving. Each of those little sections you hear on an album will be just like one Mixer set.

Sometimes that will be a straight mashup with a melody and vocals, and other times it will be more just like, oh this is all beats, and this will be more me kind of executing weird stuff with beats to kind of transition over. Something that’s really fun to play around with, even on the albums that I don’t do as much live, is just to get a whole bunch of beats in there and offset the phases and just improvise on the fly and just hear how it feels and you know, out of that go back to what you did and take little loops and say "Oh, that was cool the way that kind of dropped". Even hearing samples naturally fall on top of each other sometimes get natural phaser effects, or the kick drum might sound different if it’s distorted or layered. It’s kind of like a slightly organic approach when you’re just triggering all these drum loops and figuring out something, and then you take a loop of that. It happened in a way that’s different than actually just cutting it up in a sound editor. That’s something that’s kind of appealing to me where it’s just slightly a bit more organic.

This is something I don’t do live [opens a simpler patch], this is more just at home and it’s pretty easy just using nothing else. Occasionally I’ll use specific VST plug-ins, but even the stuff in here you know you can make enough sounds to make an interesting beat. So this is just a sample run through the DigiGrunge and the PulseComb. If I was to record out of here...[Plays loop]

So something like this where I would just record [via a SoundOut contraption] and just mess it up [manipulates PulseComb parameters]. So if I were to just take that little bit...

K: And you just improvised that and you recorded what you’re doing live?

G: Yeah, and that to me would be the source of my new beat like that. So then I would kind of go through here, [opens recorded file in CoolEdit Pro] I mean it distorted out a little bit, but regardless you can kind of go through and use little clicks or something you could arrange into being that part, you know, from the album. So then I love to just cut up beats by hand in a sound editor [highlights a small region in the wave file with a few small glitches/tones]. So if I were to just briefly do this. This isn’t a beat yet but this will be a start. And this is how I make a lot of the percussion on the actual album. Sorry I’m being specific about this [pastes a copy of the new processed moment into an existing drum loop to replace a hit with a glitch or tone].

So throwing that beat in there [loads new cutup drum loop into a Mulch mix document] those are like some Electric Light Orchestra [triggers other samples - Van Halen "Jump"], just stuff like that is on the album. Maybe not as blatant with the beat but just that little beat behind it, I feel like it adds a lot of character. Just some little clicks and pops. So I think that definitely puts a stamp on a lot of the record stuff and live as well, just the actual you know, little bits of production behind everything.

K: It’s kind of fine detail isn’t it that adds that layer of...

G: Yeah...I think a lot of the time on the style of music I make, you know when you’re going to have these samples come and go and you want them to be cohesive, a lot of the time you just want subtle elements that remain there so you don’t even really process it. They’re still there but somehow connect to those two parts. For me, a lot of times that’s just percussion. Having these little bits, clicks and things, they’re going to continue rolling even when the melodies are changing and kind of unifies the segment.

So I think a lot of the processing I do in AudioMulch is kind of crucial to that whole thing of actually doing the production that will tie a lot of the bits together.

K: Okay so you kind of get a unified sound overall. Is it...?

G: Yeah I think just as long as...I just never want, when I do an album, the past few albums at least, I want them to be seen as one piece of music. You know it’s just one collage and there’s many segments and I definitely build them as little chunks, but I don’t’ want it to be like you know, this song or an album with 300 songs. I want it to be one song. So ideally, the production, the stuff like that behind it will actually tie things together as smoothly as possible.

Girl Talk AudioMulch Interview Part 4 of 4: Gregg describes how he transitions between songs during a set

R: I was just going to ask you...before we were talking about, before we started the interview, you were talking about running multiple copies of AudioMulch at once and syncing them together. Can you just explain how you do that in your live set?

G: Sure. So, because it’s a lot of samples there, I don’t like to cut it close or anything like that in terms of overloading the computer. Naturally it’s going to reach a limit; you can only load so many samples. So when I started using AudioMulch, I didn’t know how to do this for many years. So in the early shows I’d kind of play for a little bit, and then make it all crazy like I mean I could give you an example. I guess it was less dancey back then but it’s like you can obviously start to process sounds weirdly on here [creates grinding glitchy sounds by editing LoopPlayer parameters]. You know, I would kind of get into one of these segments and then do live processing and then switch over to the next template and move on. But yeah it took know, it’s a pretty obvious idea but yeah, I didn’t actually think about just cueing up multiple documents and I think that has less pressure to cram more into one.

So yeah, basically every...I would say eight to fifteen minutes of the set I kind of reach what I want to do within that template and I kind of move over and I usually organise those by tempo. Sometimes I use the same tempo but I think it’s kind of I move in tempo chunks. So it’s like this template is 90 BPM and the next one is going to be 105. So basically if I was coming to the end of this set, or this little piece here...[starts playing with a track] so if I was to open up another template...

R: So this is another copy of AudioMulch you’ve got running there?

[Ed. This can be achieved in Windows by launching another instance of the program from the Programs menu or shortcut. It’s also possible on OSX but the subsequent instance(s) need to be launched from Terminal using the following: open -n -a "APPLICATION NAME".]

G: Yeah. So basically if I was running this, a lot of times with larger samples or something, I kind of have it planned so when I’m gonna transfer over I have something that you can actually listen to for thirty seconds without it seeming like I’m not doing anything. So it sounds somewhat busy in order to take the time.

I mean, it seems silly that just opening up the other AudioMulch and getting that template ready, the pace of the set goes by so quickly that that thirty seconds to me seems like a lull. You kind of have to pre-plan it so there’s something kind-of-busy going on that’s not totally boring.

If I was to open up a different template...let me see here [opens a new document in the second instance of AudioMulch]. So this is still playing basically just a loop. Then this is a separate AudioMulch file, I would start it up, press play and just go to the chase network sync. Then I would go over to the other one AudioMulch and generate the network sync and now this AudioMulch is chasing that network. So they’re basically cued up in perfect MIDI time and if I was to drop a sample over the top [un-mutes a LoopPlayer with an acapella sample in the new document]. So this is like an audio vocal sample from one AudioMulch with the beats coming from the other Mulch and I would slowly transfer over and drop these samples out [mutes LoopPlayers in original document]. Now at this point I’d stop chasing network sync and from here I’m on a new file and can kind of then progress through it.

So, it makes it so you can have a 20-hour set, if you truly wanted to have one, and never have to overload the computer. It can be as small a file as you want and you can kinda sync it all up.

K: So that means you don’t need a separate master mix kind of thing where you’re mixing between things. You simply just seamlessly move from one patch to the next by...?

G: Yeah it makes it...because in an ideal world it would be one giant file, but that’s just not possible with any program or any piece of computer; it just would be too much. So, it’s not a crazy headache to just open up a new one and do that little process there every like few minutes. But it basically makes it theoretically like one giant set of samples in front of me.

Even having separate templates allows me to...just in my mind prepare, and I can kind of jump around templates. Like I don’t want to do this one or I want to go into this one. The amount I have there is enough for me to think about. I know what’s going to be in this one and opposed to having them all in one that could be a little much, to be like trying to figure out everything. 300 samples or whatever in front of you trying to figure out where to go. So as is, it’s like a typical set these days I think I have six to eight templates ready. Different tempos and basically work within those.

R: So would you always go from one to the next or do you sometimes switch the order?

G: Sometimes switch the order but typically like I said, it’s pretty organised as far as how I rehearsed it, so it’s like this transitions well. Like I was saying it’s kind of like that crowd-surfing moment where you get used to one bit is like, "Oh, I can get into the crowd" or, "Oh, I can step away from the computer". I think I try to have things at the end of a template where it’s like, "Oh, this will go well picking up the pace" or I can gradually pick up the BPM without anyone actually hearing that.