You are here

Mixing Gotye's 'Somebody That I Used To Know' - François Tétaz

Inside Track | Secrets Of The Mix Engineers By Paul Tingen
Published July 2012

'Somebody That I Used To Know' is a worldwide hit that breaks every rule in the book. Artist Gotye and mix engineer François Tétaz reveal how they made it.

Although 'Somebody That I Used To Know' was mixed 'in the box' at François Tétaz's Moose Mastering, several of the other tracks on Making Mirrors were mixed at Andy Stewart's The Mill studio. From left: Andy Stewart, François Tétaz and Wally 'Gotye' de Backer.Although 'Somebody That I Used To Know' was mixed 'in the box' at François Tétaz's Moose Mastering, several of the other tracks on Making Mirrors were mixed at Andy Stewart's The Mill studio. From left: Andy Stewart, François Tétaz and Wally 'Gotye' de Backer.Photo: Alain Bouvier

We're only halfway into 2012 and Gotye's 'Somebody That I Used To Know' already looks set to become the biggest hit single of the year. Like the song itself, which remains understated until the dramatic entrance of the first chorus a minute and a half in, its worldwide success took a while to happen. Though it was released in July 2011, it only reached the US number one spot on April 18th this year. By early May, the song had sold five million copies worldwide, had topped the charts in 18 countries, including the UK, where it spent a total of five weeks at the top to become the biggest-selling single of the year, gone nine times platinum in Gotye's native Australia, and been declared the most popular song in the 47-year history of the Dutch charts, while the video had clocked up 200 million views on YouTube. The album from which it was taken, Making Mirrors, has also reached the higher echelons of the charts in many countries.

Its success is particularly striking because the song breaks virtually every directive of today's hit-making rule-book. The first and second verses are whisper-quiet, the double chorus appears only twice in the whole song, the arrangement doesn't strive to create grand entrances or bombard the listener with hooks, and there's no rapping.

The Science Of Feeling

The song's mixer, François Tétaz, had a vision for it from the beginning. He also thought long and hard about aspects of the mix that are likely to have greatly contributed to its appeal, like the way in which the dynamics of the song are shaped, with the intensity increasing at several points, his refusal to engage with the loudness wars, the imperfections that he retained in the vocals, and the way he managed to make the track sound modern without losing the idiosyncratic character of the many lo-fi ingredients of Gotye's arrangement. Tétaz was and is inspired by two books written by neuroscientists, This Is Your Brain On Music by Dr Daniel Levitin and Sweet Anticipation: Music And The Psychology Of Expectation by professor David Huron (both published in 2006). His main focus, and that of Gotye (who was assisting him with mixing the entire album), was on feeling.

To begin at the beginning, Tétaz recalls: "Towards the end of working on Making Mirrors, in early 2011, we were still looking for singles. So Wally wrote four or five songs in two weeks, which were all good in different ways. They were also all quite light, ie. they didn't have a great deal of emotional depth, and then he came in with the first two verses and chorus of 'Somebody', and it gave me a similar feeling as 'Heart's A Mess', the main single from his previous album [Like Drawing Blood, 2006]. It was very much Wally, and it really grabbed me. But we were wondering how the song could work. We felt that it needed another element, and that was the female perspective. So Wally wrote the third verse and the way the hook and all the vocals work in the last section, which is really beautiful. The writing wasn't difficult, but getting the right emotion from the vocal was. Wally first tried singing the female part himself, as a dialogue in his own head, but that didn't work. From there it was a matter of finding the right female singer and getting the right feeling from the female voice. It took a long time to get that done. We tried a local Melbourne singer and then a very famous Australian singer, but neither worked out. I then suggested Kimbra. I'd been working with her as a songwriter in development, and she has a really great voice with the kind of sassy, slightly dark edge that I thought might work. Her first attempt had real potential, so we went from there.”

Moose Mastering is in some ways an unusual studio, with high-end monitoring but little analogue equipment.  Moose Mastering is in some ways an unusual studio, with high-end monitoring but little analogue equipment.

A very short Luiz Bonfa acoustic guitar sample provided the first spark; de Backer then constructed the entire arrangement as a gigantic sample collage, partly using extensively manipulated one-note samples from vinyl records and other sources, and partly sampling himself playing individual notes on various instruments. The vast majority of Making Mirrors was programmed and recorded by de Backer in Ableton Live and Pro Tools at his own studio in a shed at his parent's property, called The Barn and located one and a half hours from Melbourne. Some additional recordings took place at other locations, engineered by Tétaz, who also has an additional production credit on five of the album's 12 songs. Finishing 'Somebody' proved immensely frustrating, a problem compounded by elaborate scheduling problems in getting Kimbra to sing her part. The end result meant, however, that Tétaz and De Backer had some wonderful material to work with in the mix.

Moose Music

Seven songs, including 'Somebody I Used To Know', were mixed 'in the box' at Tétaz's all-digital studio, Moose Mastering, while the remaining five were mixed on a board at The Mill, a mostly analogue studio owned by Andy Stewart, who mixed 'Giving Me A Chance' and assisted Tétaz and De Backer on the four other songs mixed at The Mill.

Tétaz's relationship with de Backer began with Gotye's second album and Australian breakthrough, Like Drawing Blood, on which Tétaz is credited with additional production, mixing and mastering. "I work best as Wally's sounding board. He plays me stuff and I suggest anything that comes to mind, which is kind of like a production role. In my opinion, Wally is at his strongest with the Gotye project when he has his hands all over it and does things by himself, as was the case with Making Mirrors. Though Wally may well disagree! The way he approaches things when it's just him is what makes his music so special. It's very idiosyncratic. I just bounce off ideas with him, suggesting that this or that could be better or maybe he should try so and so, but generally I don't get hands-on during writing and arranging. I know he finds it frustrating sometimes, because working by himself can be very lonely, but I think it's the fact that he has to grapple with all the different challenges that makes what he does special.

"Most of the songs on Making Mirrors have a lot of detail, with many different bits and pieces resulting from Wally having cut up and arranged loads of samples — 160 tracks per session was not unusual! But the songs sound uncluttered because of Wally's arrangements and also because of his sense of space and placement. He's very big on that. All his ideas are very composed and most of the ideas in the mix come from him. Apart from being Wally's soundboard, and the intellectual idea of bringing Kimbra to the song, all I really did in 'Somebody' was to articulate his vision and try to mix it so that it moved me and gave me and him a feeling, and hopefully other people as well. It's a little puritanical, but I really try not to put my own stamp on what he does. It's about trying to make the most of what's there, making sure that I love it and that he loves it.

The large Schroeder diffusers on the rear wall at Moose Mastering help to ensure an accurate sound even at low frequencies.The large Schroeder diffusers on the rear wall at Moose Mastering help to ensure an accurate sound even at low frequencies."Generally, if both Wally and I like it, it's good, but it can be a long and hard process for us to get to that point! Quite often, I overstep the mark and do something that Wally doesn't like — for example, making a snare drum sound too hi-fi — and he'll literally say: 'I hate it, this sounds too plastic and way too pop.' Wally is not making decisions based on whether sounds are hi-fi. Instead it's more like: 'This has a real feeling about it.' His samples often are lo-fi and a little bit dirty. Part of my job as a mixer was to stretch the canvas, to add depth and dynamics so his tracks can comfortably sit next to tracks by Jay-Z or Rihanna, rather than them sounding like lo-fi records. But I didn't want to fundamentally change things and make it sound too pop and take the life out of it. Instead, the heart of my job as a mixer was mainly to add emotion and depth. Of course, the song has to have great lyrics and a great melody, and this is a very good song with a universal theme. But the success of the song has very little to do with what I did in the mix. I just did my best not to destroy it!”

Telling Stories

"The way the song is constructed and my mix have a lot to do with sound placement and the tension and release narrative of the song. In terms of placement, when I was a teenager in the 1980s I was a massive fan of Trevor Horn. His pop productions, like Grace Jones' 'Slave To The Rhythm' and Frankie Goes To Hollywood really grabbed me. I also listened to dub music and musique concrète at the time, and what all these had in common was the emotion and drama of spatial relationships, whether from small or large spaces. The same with Phil Spector, Lee Hazlewood and Beatles records. They all worked with the dynamics of space and the textures of sounds and distortion, the places where things fit and the emotional qualities this gave. This is a very important element for me in terms of communicating the central idea of a song, and setting it in the drama of space.

François Tétaz: Mixing Gotye's 'Somebody That I Used To Know'"My approach to music in general is very practical: it's about whether you can hear the story. And does it grab me musically? Sweet expectation, which is the title of a book by David Huron, is fundamentally what music is about. So I'm trying to work with what I think are people's expectations about what's going to happen, and then twisting them into something new that you didn't expect but that grabs you and gives you a new feeling. 'Somebody' is all about the chorus and the story is kind of creepy with tension and then release and celebration. In his book This Is Your Brain On Music, Daniel Levitin, who was a musician, engineer and producer before he became a neuroscientist, talks about this being nothing to do with a pop sensibility but everything to do with general musical sensibility.

"Of course a song has to have hooks and changes and melodies and bits and pieces, but really it is about you wanting to be thrilled when listening, and wanting to be fulfilled on many different levels, and being grabbed emotionally when listening multiple times. The difference between songs, and genres, is mostly about intent and timbre, in the texture of a voice, in the instrumentation and in the production overall. The reason why engineers and producers use a particular console, microphone, preamp or compressor, as well as make spatial/reverb decisions, is essentially to do with overall sonic texture and musical intent. But so much music today is just noise: you can't actually hear the vocals and the dynamic shifts between the verses and choruses and so on. Modern mixes often are in your face all the time, which means that I may go 'Wow!' on first listen, but on second listen I already get annoyed, because the song doesn't take me on a journey or play with my expectations. People put a lot of effort into making the beginning of the track great, so it has immediate impact, but when it gets to the chorus, or anywhere else where it needs a dynamic shift, they have run out of headroom. By the time the song gets to the chorus, it actually often gets softer, because hitting compression is the only tool the mixer has left to reach for. I do the same when I make decisions based on fear rather than musicality.

The lengthy plug-in chain for Gotye's lead vocal included Antares' Auto-Tune — used only on a few words — two instances of Cranesong's Phoenix, two Waves Renaissance De-essers, a Renaissance EQ, and McDSP's Compressor Bank. The lengthy plug-in chain for Gotye's lead vocal included Antares' Auto-Tune — used only on a few words — two instances of Cranesong's Phoenix, two Waves Renaissance De-essers, a Renaissance EQ, and McDSP's Compressor Bank. "On the Making Mirrors record, there are a few places where I struggled with that, because of course you do want your track to have impact at the beginning, but then where do you take it after that? I was constantly working with Wally to try to make the tracks sound as big as possible, yet retain enough room so you could musically give it a dynamic shape. That is always a compromise. When you do that, it means that your track will never have the same initial impact as a track that is mixed like a wall of sound from the beginning. But the musical journey is much more fulfilling. It's a more old-school philosophy of mixing, in which you're mixing for the whole curve of the song, rather than for just the first 10 seconds and using bravura to catch the listener's ear. Of course, with 'Somebody', it made it much easier for me that I was dealing with a very good song about a universal theme.”

Starting Points

"When I mix, I always make sure that the vocal performances and edits are top-notch before I do anything else. Wally comped his vocals himself, and he'd then bring them to me and I'd finesse his edits: this is one area where I'm really finicky. Once I was happy with the vocals, I began work on the Luiz Bonfa nylon guitar sample. It has vinyl crackle, which Wally left for the first phrases and after that removed. The first few bars give you the impression of listening to a record, and for the rest of the track these noises are not there. I treated the nylon guitar with the Waves Renaissance EQ, and the Waves C4 multi-band compressor, which is expanding the mid-range and the top end. I copied the nylon guitar for the choruses [to the track named 'NylGDrive'] and added the Digidesign Smack compressor and the McDSP Compressor Bank to give it extra drive and power. It's very much a feel thing. The Compressor Bank is actually being triggered by the kick [track] just above the nylon guitar, and applies hardcore compression, completely smashing the sound, with the phrasing of the kick drum pumping it. It's one of these things that give you a feeling of depth in the track.

The plug-in chain for Kimbra's vocal section employed many of the same plug-ins as for Gotye's, but with the addition of an SSL bus compressor.The plug-in chain for Kimbra's vocal section employed many of the same plug-ins as for Gotye's, but with the addition of an SSL bus compressor."'BscVerse' is the bass guitar, played by Lucas Taranto. I felt that the track needed a little bit more bottom end, so Lucas overdubbed a bass part. It had the crap edited out of it and I compressed with the Bomb Factory BF76, so it sat with the sample. Below that is the 'BassWithDrive' track, which has distortion on it from the Sonnox Oxford Transient Modulator. It enters at the beginning of the first chorus, at the same time as the kick and the 'LowGuitar' duplicate. The kick has a 12dB boost at 55Hz and 18dB at 5kHz with the Renaissance, which is my go-to EQ, plus it's gated with the Drawmer Dynamics Expander and it has the Compressor Bank, all for it to have a huge thump in the bottom end. The low guitar has some minor EQ from the Renaissance and has been doubled with the bass, plus I added a Digirack Delay to give it some depth and to give the bottom end some spread. All this low end was to support Wally's voice, which goes quite high in the chorus, and if it didn't have the support of the bass, it could easily sound a bit thin and screechy. His voice needed some real low-end power below it at that point.

"The 'chucks' guitar sample ['ChckDrive', playing a G-F-F note sequence] below the 'NylonGuitarDrive' track also supports Wally's voice in the chorus. Wally created two 'chuck' guitar tracks, one dry ['ChckDr101'] and one with reverb ['Chcks-Wt01'], and the dry one has quite a big pre-delay on it, plus a [Waves] Renaissance compressor with a really fast attack and a Renaissance EQ with a 10dB boost at 1.3kHz for a bit of mid-range poke. The 'ChckDrive' guitar for the choruses has all the lower-mid taken out with a huge cut at 300Hz using the Renaissance EQ, and I added a short stereo delay to brighten and widen it. As the chorus goes on, this guitar becomes brighter and wider, giving the chorus a very subtle lift.

"The 'BoofyDrums' below that is one of these lo-fi parts that I tried to make more hi-fi, and where I eventually had to conclude that where Wally had put it was the best place for it to sit. You can see that it has three greyed-out plug-ins: all that stuff wasn't necessary in the end. It was a low drum part with a lot of hiss on it, and the easiest way to manage that was simply to have a low-pass filter. It was the same with the 'TangoDrums'; the way they sat in the track was totally fine, I just added a bit of 'HallDelay' from the aux track at the top of the session, just a bit of classic '60s plate reverb spread. The funk guitar [playing 'C' and 'D' notes] was just a tiny piece of scrap with lots of hiss, and it also didn't need treatment. The 'WobblyGuitar' is one of the main instrumental hooks [playing A-C-A-G-D], and has just the Renaissance EQ cutting out a whole bunch of 13k top end, again to get rid of the record noise. I tried denoising it with Sonic Solutions NoNoise but that didn't work, so I reached for the EQ.

"Underneath that are the 'Pizz' track, with pizzicato strings, and the autoharp track. The latter is brilliant. It has just two small bell-like tones, and the first one is played the moment before Kimbra comes in, and settles you into that third verse. It's the most simple and elegant way of introducing a new idea. Underneath is the 'LatinGroove' loop, which is doubled, but the duplicate is muted. I also doubled the bongo part, and on the first track I worked on the transient, the front end of the sound, using the Sonnox Oxford Transient Modulator to get that to sit in the right place. I also have the Renaissance EQ on it and it's sent to the Speakerphone aux track. The second bongo track is EQ'ed with the Renaissance EQ and has a small smashed bathroom space using the [Avid] D-Verb. The idea is to give the bongo more depth and space, so it's not just a flat little sample that just sits there. I tried to make it feel a little bit more organic. I do this a lot on percussion sounds in particular.

"The two shaker tracks are really subtle, adding some very light texture in the chorus. The 'LightShaker' sits very lightly on the left and the 'BlackShaker' is a little heavier and more dramatic. Both help lift the chorus without being very apparent. The xylophone is one of the main instrumental hook parts, first heard right at the beginning of the song. It has some Renaissance EQ, with a lot of mid-range added, +10dB at 1.2kHz, as well as some extra brightness, +4dB at 10kHz. The main thing about that part is that it has no bus compression. The rest of the tracks go through a McDSP Compressor Bank, which is on the 'MixSqueeze' aux track at the top, but the xylophone has its own bus. I find that if you compress xylophone and delicate sounds like that, they start to sound a bit driven and lose their simplicity and naivète. Then there are two Mellotron flute parts that are sitting quietly in the back, like a pad. They just fill out the chorus and otherwise don't do very much. They have a bit of automation on them to push them up in places. Finally, the 'Synth' part is another instrumental hook [playing G-A-C-G-A-F-C]. It has some Speakerphone reverb, and some Avid AIR Stereo Width, plus the Phoenix Cranesong Luminescent plug-in to give it more width and shine.”


François Tétaz chose to work at The Mill on the tracks that he felt would benefit from its high-end analogue equipment.François Tétaz chose to work at The Mill on the tracks that he felt would benefit from its high-end analogue equipment.

"Kimbra recorded her own vocals and did a comp, which Wally then comped and I comped again — minor tweaks and word selection. The first track, 'WalyftCmp3', is the actual vocal channel, the '32' track underneath is a reverb send that has different automation, compression and EQ than the main vocal and is also going through a pre-delayed reverb. That second track is like an effects control. The main vocal track has a De-esser, SSL Limiter, Phoenix Iridescent, Auto-Tune, Renaissance EQ and McDSP Compressor Bank. The reverb track also has Auto-Tune, Renaissance EQ and the Compressor Bank, and goes to the Altiverb chapel reverb aux track. Underneath those vocal tracks are two muted Kimbra vocal tracks. Wally really liked the idea of given her voice quite a distorted, radio-like quality, so we put them through some guitar effects and the Line 6 Pod Farm plug-ins, but I wasn't into that. It made her sound less powerful or more powerful, depending on your perspective, and I wanted the female voice to be equally powerful. We worked with it for a while, but I kept pushing Wally for the vocal to be straighter and in the end we muted these two tracks. A few tracks down you'll see a track called 'Kimbrado', and that does go through the Pod, but just for a moment. It's where she sings the phrase, 'somebody that you used to know' as the second chorus kicks in. At that point it's like a new texture that adds some extra grit. If you'd been listening to it the whole time you wouldn't have had that surprise and you wouldn't get that real moment of release.

"Wally's verse vocal, 'Vocal Comp', has Auto-Tune, used only on the odd word, the Cranesong Luminescent and Iridescent for some drive, two Renaissance De-essers, Renaissance EQ, and a slow compressor, which is one of the McDSP Compressor Banks. I again made a duplicate. The main vocal goes to a bus called 'Wally', and the duplicate goes to the 'HallPreDelay' aux with the Altiverb church reverb. The duplicate also has a De-esser, the Renaissance EQ and, again, the Compressor Bank. The reason I created a duplicate is that it is EQ'ed going to the reverb. There are a number of sharp 's' sounds in the verse, and so it has a big de-esser in front of it and then the EQ rolls out all the high so the reverb is fed a much smoother sound. The 'Botanicals' track underneath was an experiment [he plays a very psychedelic track with lots of reverb and delays]. I wanted some depth and grandeur, but it was too much and clouded up the subtle things that the flute parts were doing, so we ended up muting that track.

"Wally's chorus vocals were also doubled, and both tracks were treated very similarly to his verse vocals, with the same inserts: Cranesong Luminescent and Iridescent, SSL Channel, two Waves Renaissance De-essers, one for high and one for low [frequencies], and Renaissance EQ. The two vocal tracks are panned slightly left and right; the vocal on the right is dead dry, and the one on the left is sent to the Speakerphone reverb and also has some Echo Farm bright tape delay with a very subtle slap to slightly fill out the chorus vocal. It's a very Beatle-esque thing to have a highly compressed and EQ'ed dry delay on a vocal. One of these chorus lead-vocal tracks also has a side-chain send going to the chorus backing vocals, to duck them in relation to the lead vocal.

François Tétaz: Mixing Gotye's 'Somebody That I Used To Know'

"Below the two chorus vocal tracks are all the backing vocals for the final section, you can tell from the names what words are sung. Several didn't need treatment. The 'somebodys' were recorded with Wally's Mac laptop mic, and were pretty lo-fi. You can hear him stomping and stuff like that, but we liked how they sounded. Some hi-fi vocal tracks also sounded good enough to be left untreated. We tuned some of the softly sung 'used to know' backing vocals with Auto-Tune. We liked the vibe of them, so we decided to tune rather than redo them. You can see that the plug-ins that I did use on various backing vocal tracks are similar to those on the main vocals: Auto-Tune, de-essers, Luminescent and Irridescent, SSL Channel and Renaissance EQ. There's also a BF76 compressor. The three tracks called 'BVs' go to the 'Oh Oh Oh' master track, and these are the ohs in the final chorus. Wally demo'ed these vocals, and Kimbra did sing some of them, but we felt that she sounded a bit harsh, so we used Wally's demo vocals. Wally and Kimbra convinced me to mix them in rather loudly, so it gives the effect of a bit of a battle between the 'ahs' and Wally's lead vocal. It was a new element that came in and that became a hook in its own right.

"The final mix only had a tiny bit of EQ, and that was it. The 'Mix' and 'MixSqueeze' buses go to the HEDD bus, which had the Renaissance EQ on it. The 'MixSqueeze' track was hard compressed, with a tight 5ms attack and a 100ms release, and it's automated to make the parallel compression louder in the chorus. The overall volume is not getting more; I'm just pushing more of the compressed track into the mix. At the end of the mix, I had all those tracks there and I tweaked things, in terms of volume and compression and EQ, to make everything sit how I wanted it. I then recorded the mix internally into the session. It's a very simple mix in terms of the outputs: all the sounds already had their treatments track by track.”

Rarely will a 60-track session have been whittled down to a stereo mix of such transparency and apparent simplicity, and with such sensational results.  

Gotye: How I Wrote 'Somebody That I Used To Know'

Wouter de Backer, aka Walter, aka Wally, aka Gotye (a stylisation of Gautier, the French translation of Walter) recorded Making Music and 'Somebody' in a barn on his parents' property close to Melbourne. His main tool was an Apple MacBook Pro laptop, running Ableton Live and Pro Tools. "Ableton was the starting point for most of the songs,” explains de Backer, "and I used Pro Tools for engineering instrumental and vocal recordings and editing. My main vocal microphone was a Neumann M147 that went into a mono SSL Alpha Channel. I also sang many vocals directly into my MacBook Pro mic, and used things like Speakerphone from Audio Ease to emulate different microphone EQs and different spaces and settings. Plus I borrowed some other Neumann mics, Audio-Technica mics, and a Carillon Axis 70, a very cheap Chinese ribbon microphone for some of the vocals on the song 'Brontë'. My dad put together some acoustic tubes made from glass wool and wood and stretchy fabric. I also spent a lot of my time searching for strange instruments and old vinyl records for me to sample.

"I occasionally used a TEAC A3340 quarter-inch [tape machine] for some bass recordings and other stuff, mostly when sampling notes of acoustic instruments to turn them into virtual patches. I hit the tape hard to get harmonic distortion. I also recorded samples with other bits and pieces, like a Dictaphone — a little handheld cassette recorder — and the Edirol 09. I used the latter to record the Winton Musical Fence in the outback of Queensland, which became the bass line for the song 'Eyes Wide Open'. I dumped all these recordings into Ableton or Pro Tools, and then had to edit them to bring them into time. Another piece of gear that I used a lot was the Novation Launchpad, to trigger samples in Ableton. I'd chop my samples up in Drum Rack in Ableton and triggered them with my Launchpad to try out melodic ideas. Many of the main hooks for 'Somebody I Used To Know' and part of a song like 'Brontë' was me playing samples with the Launchpad to come up with melodies.

"I made my first two records just moving coloured boxes around on the screen with a mouse, and being tied to a desktop computer, so it was nice this time to record using different things. One of the main differences with Making Mirrors, in terms of the sampling approach, was that I tried to incorporate more live performances to come up with more original melodies and ideas. I am still using samples for texture and like a platform to work from, but I then used particular interfaces like the Launchpad or a MIDI keyboard or a set of drum pads to interact with those sound snippets and come up with different sequences and different ideas. The MIDI keyboard I used was a Novation Remote SL, which is actually broken now. One of my favourite patches that I created for the album was the autoharp one. Live, I just pull that patch straight out of Ableton Live and play it with the MalletKat. It is like having a large hammer dulcimer on stage.”

Other gear that Gotye used in The Barn during the making of Making Mirrors included a Minimoog Voyager Old School synth, a Suzuki Omnichord, and the huge Lowrey Cotillion D575 home organ, bought for $100 in a second-hand store, on which he created most of the track 'State Of The Art'. Unlike some sample-oriented musicians, de Backer does not lack instrumental skills — he's an accomplished drummer and a respectable keyboardist. "I really wanted to involve more actual playing and more musicians on Making Mirrors,” he says, "but probably half the songs on the new record were still inspired by a break that came off a vinyl record. In the case of 'Somebody', it was the nylon guitar sample that came from Luiz Bonfa's track 'Seville'. It was just two downbeats and two offbeats, and I took the offbeats out and turned them into their own little kind of separate top line, and I then crafted a bass-note line out of the two bass notes in the first two beats in my track, and put these two sets of notes on top of each other to create a repetitive two-note loop that I found quite hypnotic. The other parts, like the xylophone and the wobbly vibrato guitar, are also from records, but chopped up into individual notes and re-pitched. The wobbly guitar line is the second hook in the song, it occurs between the verses, and the vibrato came from me manually manipulating the turntable as I recorded the guitar sample.

"Writing 'Somebody' was a gradual and linear process. I started with the Luiz Bonfa sample, then I found the drums, and after that I started working on the lyric and the melody, and added the wobbly guitar-sample melody. After that, I took a break, and a few weeks later I came back to the session and decided on the chorus chord progression, wrote the chorus melody, and combined that with sounds like the Latin loop and some of the percussion and the flute sounds that further filled the space. At that point I hit a brick wall. I was thinking: 'This is pretty good, how can I get to the end really quickly?' and I was trying to take lazy decisions to finish the song. I considered repeating the chorus, an instrumental bridge, a change in tempo or key, I even considered finishing the song after the first chorus. But nothing felt like it was strong enough. So the third session was all about writing the female part and changing the perspective. The arrangement of 'Somebody' is reflective of me moving towards using sounds that provide me with inspiration for a texture or a platform for an idea, and then through sonic manipulation and coming up with original melodies and harmonic ideas to make it my own. I guess the balance of sounds taken from records and samples I created myself is perhaps 50-50.

"When you see the 'Somebody' session, you realise that there are many more sounds in there than is apparent on first listen. It may sound like quite a minimal song, but there are many different small things happening at key moments that provide minor accents for the lyric. During the mix, Franc [Tétaz] helped me make decisions on the right amount of noise versus cleanness in samples and on the overall dynamic arc of the song, like how soft can the verse be for it to still be engaging and intelligible to an audience that's using all kinds of different speakers, while still also having a huge powerful dynamic jump in the chorus. Clarity and well-controlled dynamics are often lacking from my rough mixes. Frank is good at massaging my arrangements of samples and soft synths and bits of live performance into a coherent sound stage. It is not easy to create that from sources that come from so many different places.”

François Tétaz

Born in 1971 as a second-generation Australian of French Swiss descent, the teenage François Tétaz devised his own music education, his teachers including Graeme Leak, then artist in residence at the Sydney Conservatorium of Music. Tétaz was particularly interested in electronic music and musique concrète, and in 1992 co-founded experimental music trio Shinjuku Thief. He released a solo album in 1997, and, following this, divided his time between producing and composing scores for movies and theatre.

Tétaz adds: "In my early 20s, I also did a lot of remix pop work and I learned a lot about engineering from Australian engineer Mark Forrester, who used to be the in-house engineer at Prince's Paisley Park studios. That was really invaluable for the beginning of my career. I see mixing and engineering skills as a practical necessity. I wanted to gain those skills because I had become obsessed with sonics and textures and I wanted to be able to articulate and realise my own musical ideas without being dependent on others. Also the reality of working in Australia was that there was no way I could write and produce music full time and make a living from that, so my wife suggested I begin my own studio facility, which became Moose Mastering. I opened it in 1998 as a place where I would be able to do all the different things I wanted to do with my own music, and also mix and master other people's projects. Mastering was something I had a real interest in, and I spend a lot of time in my late 20s learning the craft. One of the main mastering projects I have done here was Merzbow's 50 CD box set of noise music, Merzbox [2000]. I think it means I'm well qualified to talk about noise.

"The main room here is six metres wide and eight to 10 metres long, and has a wall with Schroeder diffusers at the back. I have a large Pro Tools rig with a whole bunch of plug-ins and DSP. Other than an Avid Command 8, I don't have a desk, and very few bits of outboard gear. My main piece of outboard is the Cranesong HEDD. I do all mastering in Pro Tools. I'm a minimalist: if I don't use something pretty much every day, I don't see the point of having it, and it's cheaper not to have it. What I do is very much based on a musical-feeling thing and not on what gear I use. Designing and building the room here was quite expensive, and I invested in Duntech Sovereigns, which are my main monitors. They have ATC mid-range drivers with a very clear mid-range and bottom end that extends flat to 27Hz. The acoustics of the room are very even as well, so I have a really good sense of what is happening in the lower two octaves. I also have a 5.1 set of Mackie HR824 speakers, which I don't like that much, [Yamaha] NS10s, Auratones, Genelec 8050s and Event Opals. I used the last a lot when mixing Wally's record.

"I like the fact that we mixed 'Somebody That I Used To Know' at my place in the box. A major reason why that worked was that all Wally's sound sources are samples with a lot of character, so I wasn't really looking to add more character. Sometimes putting things through transformers and other bits of analogue gear can be detrimental to what you're trying to achieve. If a track is sounding great and I'm getting an emotional response from it, than that is terrific. I don't care whether it's done in the box or on a desk. But there were a number of songs on Making Mirrors that I felt would benefit from being mixed on a desk with analogue outboard, and for those we went to The Mill, Andy Stewart's studio. It's an analogue haven, and I really enjoyed going between Moose Mastering and Andy's place. His studio sounds completely different than mine, and it has lots of different colours and flavours that I can reach for. His board is very transformer and creamy-sounding, and is great for tracks in which I want to bring depth and a feeling of lushness to the sound, but for other sessions things might start to sound a bit mushy and lose drama. Old boards can steal a bit of the top end or transients as well, and again, that works on some tracks and not on others. I did compensate for that, like using his SSL bus compressor and EQ to make the mid-range a bit harsher or add some top end. Also, there may be 160 tracks in Wally's sessions and I may cut these tracks up further to work on transients and so on, so we had some sessions with a very high track count. You can't bus all that over an analogue board, so we made core decisions in the box and then made stems of the tracks that we were sure about that way laid out over the board.”

The Screens

Like most of his songs, 'Somebody That I Used To Know' was programmed by de Backer in Ableton Live, after which he switched to Pro Tools for recording live audio, in this case just the vocals. François Tétaz received a 24-bit, 44.1kHz Pro Tools session with 160 tracks in the playlist, though only about 40 were used. Several tracks were doubled, since Tétaz likes to work with parallel treatments, and with four aux tracks and a few mix buses, the entire session consists of just over 60 tracks. The final mix ('V2.23') is at the top of the session. Below that are a 'Mix' bus and its parallel, 'Mix Squeeze' (with McDSP Compressor Bank compression), and then a 'Xylo' bus purely for the xylophone; all three are controlled with the 'Volume' track and routed to the 'Hedd' track, which feeds the stereo mix to Tétaz's Cranesong HEDD. From there, it comes back in on the 'SmbdMix08' track. Underneath are four aux tracks: 'Speakerphone', with Audio Ease's Speakerphone set to an EMT plate reverb; 'HallPreDelay', with the same company's Altiverb set to a 2.5-second chapel reverb; 'TapeSlapBright', with Line 6's Echo Farm; and 'Plate', with an Altiverb EMT plate reverb fronted by a 170ms (dotted semiquaver) pre-delay.

Below this the individual audio tracks begin, starting with two bass tracks, two low guitar tracks, a kick-drum track, three nylon guitar tracks, including the Luiz Bonfa sample, three 'Chucks' guitar sample tracks, drum tracks labelled 'BoofyDrums' and 'TangoDrums', three more guitar tracks, pizzicato strings and an autoharp track, six more percussion tracks, the xylophone track and two flute tracks. These are followed by 26 vocal tracks, with the Kimbra vocal comp at the top (strangely called 'WlyftCmp3'; 'WlyftCmp32' is a double), then two muted Kimbra vocal tracks, and two Kimbra backing-vocal tracks. 'Vocal Comp' is Gotye's verse lead-vocal comp, which also is copied. Underneath is 'Botanical', a muted vocal effects track, and then two chorus comp tracks. The bottom of the session consists of a further 15 Gotye backing vocal tracks, revealing that what sound like female backing vocals in the second chorus were actually sung by Gotye.

Tétaz: "I always organise my sessions based on the musical roles of the tracks, so don't necessarily put all the drums and guitars together. In this case, the kick-drum track related to the bass, which is why it's put next to it. But if all the guitars are working together I will have them next to each other. It's more practical, because I have just eight faders on my Command 8 [control surface], which I use all the time when mixing. I love faders and hate working with a mouse. Working with just eight faders is OK, though it'd be awesome to have something like an [Avid] Icon in my studio! Because of the sample, the track is 30 cents flat, and some of the other samples are out of tune too, so before I began the mix, we tried pitching them, as well as the whole track, using [Antares] Auto-Tune. But we found that it lost a whole bunch of atmosphere: the track became too straight. It was the same with Wally's and Kimbra's vocals. There's Auto-Tune on a few of their notes, but very minor. Many of the note corrections didn't add anything and everything felt better as it was, ie. a little flat. It was awkward to leave the entire song 30 cents flat, because it impacted the song order of the album. The song sounded strange in some places.”