Innovative production can be made or broken by the mix — and in Rob Kinelski, Billie Eilish's team found the perfect foil.
Billie Eilish's breakthrough naturally drew attention to those responsible for the ground-breaking musical direction and sound of her recordings. Eilish breaks the mould on many fronts and this is one: many contemporary albums feature dozens of writers, producers and mixers, but only four people are credited on When We All Fall Asleep, Where Do We Go? They are the singer herself, her brother Finneas O'Connell, mixer Rob Kinelski and mastering engineer John Greenham. In contrast to the two new kids on the block, Kinelski and Greenham are both dyed-in-the-wool professionals, but neither is part of the select studio elite responsible for the vast majority of today's hit records.
Being one quarter of the small team that has taken Billie Eilish into the stratosphere therefore had a huge impact on Kinelski, who is suddenly enjoying a surge of interest. He still has an air of slight disbelief when discussing what happened. "After I finished mixing the album," he comments, "I listened to it in its entirety, and I thought: 'Wow, this is a really dark album!' There is no high end in it, apart from a couple of songs with high synthesizers. I was imagining being slammed on gearslutz.com. I loved it, but I wondered how it would translate to the world."
The answer, as it turned out, was 'amazingly well': and Eilish's astonishing success has not only been life-changing for the 17-year old singer and her brother, but also for Kinelski. "The phone hasn't stopped ringing," he acknowledges. "It's been crazy!"
Kinelski was first introduced to the music of Eilish and O'Connell early in 2017, not long after their breakthrough song 'Ocean Eyes' was re-released on Darkroom/Interscope. "Finneas had mixed 'Ocean Eyes', and once they were with Interscope they were looking to hire a mixer," Kinelski recalls. "I'd just mixed the Lost On You album by LP, which John [Greenham] had mastered, and he recommended me, saying I'm good with low end. I think they wanted to find someone with an urban background, because Billie is a big hip-hop fan. But they also wanted a mixer who could do more than just urban. The first song I mixed for them was the follow-up to 'Ocean Eyes', a song called 'Bellyache', as a trial. They loved it, and called me again for the next song, and just kept calling. Then I did their EP, Don't Smile At Me [August 2017], and eventually the entire album.
"I haven't really spent much time with Billie or Finn. I met them before we did the EP together, because they wanted to get to know me, and we had some discussions about me and about what they wanted. But really, we quickly got to the point where they just send sessions to me, and they know what they're going to get from me. I know they want the low end to be massive, but also not overwhelm other things, they want the vocal to be super-present, and I found out that any unusual things in their sessions tend to be deliberate. For example, the song '8' came in with a kick that was off-centre, and I made it mono and centred it. Initially they liked this, but then Finn asked me to put it back where it was. On the song 'Xanny' there was a debate that possibly the chorus vocal was too distorted, which was intentional by Finn, but I managed to find a happy middle ground, blending a clean vocal in with the distorted vocal. They know what they want.
"Our way of working is always the same: they send me a new song, and I take a day to mix it, depending on my schedule, and I'll send over my first mix pass, and then they come back with comments, usually very quickly. Sometimes they send me a new stem to swap for an old one. Finn sends me great stems, with the vocal production already dialled in, including all distortion and other effects. I usually do not know what treatments they have added to the stems, but everything is really thought-out. Billie also is heavily involved, though generally I communicate with Finn. It was funny: for the album they sent me voice notes for the revisions. They'd be talking into their phones and send me the messages they recorded. Sometimes Billie would sing the vocal parts she was referring to. It was very cool."
The first thing Kinelski does when he receives a song to mix is load it into his mix template. "Finn sends me stems, and when I load them into Pro Tools and push them all up at unity, it'll sound the same as his reference mix. I then filter everything through my routing and summing, and the aux effect tracks in my template. After finishing a mix, I get rid of everything I don't use. With Billie Eilish, this usually meant that only my 'Width' aux track remains in the session, which has the Soundtoys MicroShift, for some space. The plan with the album was for most things to be super-dry. I added some reverb to an early mix, and they immediately said: 'Get rid of the reverb.' When I heard they didn't want much reverb, I was super happy and inspired. That is so cool.
"Before I get to the effects, my initial process is quite old-school. I start with kick, snare and hats, and get them to feel good. Finding the right balance between those three is the most important to me. I then bring in the vocals, and make sure they sit with the drums. I guess that's a hip-hop approach. After that I take them out again, and add the bass. I like to get the low end dancing away, with the kick and the bass in the right place together. To me they're one instrument, they need to be together, and work in harmony. Does the bass need to be under the kick, or the other way round? If the kick has too much sub, I roll some of that off. Next I bring in the keyboards, and work on them together with the drums. I work in groups, and build the entire track together like that.
"While I am listening, I deal with anything that bugs me. I stop the session and may solo something, and do a little surgery. But the funny thing is, particularly with the vocals, I do most of the detailed surgery last. I first deal with all the levels, and try to get the right vibe, and dial in my mix bus compressor to make things dance, and then I mix into that. I'll EQ and do automation and then at the end, when everything is really exposed, I will go in and do detailed surgery. It is really easy in this day and age to overdo surgery. You're mixing visually and you're adjusting things by a tenth of a dB and you swear you can hear the difference. It's really easy to fall down that rabbit hole.
"I try to avoid that, so most of my vocal clean-ups happen at the end. I'll solo them but again will only deal with things that jump out. I prefer to leave everything else and keep it organic. With Billie's vocals, I did not do any timing correction or tuning. As I said earlier, her vocals are fully produced. But towards the end I do get rid of little pops and clicks that don't really show on the waveform. I just pencil tool them out. It's quite subtle, but there are quite a few of them, because she sings so quietly, and once I add compression and bring the vocals forward, these things get exposed. I get rid of these things, but I don't go too crazy on sibilance and all that stuff, because if it feels good, I do not want to over-doctor it."
Written by Billie & Finneas O'Connell.
Produced by Finneas O'Connell.
Kinelski illustrates his approach with a detailed dissection of his mix of 'Bad Guy', the biggest single from When We All Fall Asleep, Where Do We Go? The Pro Tools mix session contains a fairly modest total of 68 tracks, of which 49 are audio tracks. Of these, 21 are drums and percussion, the same number are vocals, and there are four bass tracks and just one synth track. The other two audio tracks are mix prints. Clearly, O'Connell's and Eilish's focus is almost entirely on drums, bass and vocals, something that also shows in the mix session of the other main hit from the album, 'Bury A Friend', the only difference in that session being the presence of five synth tracks and seven tracks with Foley sound effects, called things like 'screech', 'scary shit', 'voldemort' and so on.
The remainder of the 'Bad Guy' session consists of aux tracks, starting with seven aux group tracks labelled Kick, Main Snare, Sub Bass, Lead Vocals, Lead Vocal Doubles, Lead Vocal Harmonies and Backing Vocals. As well as the aforementioned lone Width aux effect track, there are eight aux tracks that send to the Dangerous 2-Bus, plus a VCA fader to control the output level on these tracks and two aux master tracks for the 2-Bus return.
Leaving small differences aside, the structure of the mix sessions for 'Bury A Friend' and 'Bad Guy' is almost identical. According to Kinelski, his mix approach for each of the songs on the album was indeed very similar, with similar plug-ins on many of the tracks, identical master bus chains, and also, very few plug-ins in general.
"You can probably count all the plug-ins I use on two hands. I like to keep it simple. I get new plug-ins when clients give me sessions with plug-ins I don't have, and I then buy them to see what they do. Also, because all the vocals are one person with one mic, after I had dialled in Billie's vocal chain on the first song for the album that I mixed, I did not change it any more, but just tweaked the settings on that chain, as required. Everything that Finn sent me was super-consistent, so I also kept my approach consistent, and that meant that the entire album has a consistent sound. The beauty of what Finn gives me is also that all transitions between sections are already built in. All dynamics are there from the way he stacks his productions. The sound effects in particular create many of the dynamics and transitions and the rest is loop-driven, which is really hip-hop."
"The three kick tracks all go to the Kick aux track above them. I have the Waves SSL Channel Strip on one of the individual kick tracks, cutting around 260Hz. On the Kick aux I have another instance of the SSL Channel, as well as the FabFilter Pro‑Q 2. The SSL Channel does some gating, and compression with a 4:1 ratio, fast attack, a pretty quick release, a little less than 0.2. I'm also boosting 3kHz by 9dB and 90Hz by 3dB. The Pro‑Q 2 cuts 500Hz, boosts 60Hz and 200Hz, as well as 800Hz. The kicks sounded a little too full, so I was trying to get rid of what I didn't need and boosted the other frequencies.
"The four audio tracks below the Snare aux are all snares, and are sent to that aux, and I have just the SSL Channel on the aux. I don't EQ, but I compress and gate. The compression is at a 5:1 ratio, slow release and slow attack, resulting in about 4dB of compression on the transients. I don't use the gate in a traditional fashion, which is to clean up stuff, but instead I use the gate to add some snap. I think gates are great for adding punch to drums. It's really subtle, but it helps them to cut through."
"The two tracks called 'subbass1' and 'subbass2' are the main bass riff, and there's a distorted 'dirtybass' track that doubles them in the hook sections. The two sub-bass tracks are routed to the Sub Bass aux, on which I again have the SSL Channel Strip, boosting a little bit around 800Hz to get some more tone, and cutting a little around 400Hz. I also added compression, with a fast attack, slow release, ratio 6:1. The second plug-in is the Waves Renaissance Compressor, which I side-chained slightly to the first kick track. The R-Compressor has a 20:1 ratio, with fast attack and medium release. I think it's a keyboard bass, but it feels and sounds like a live bass.
"The purple track is the synth melody that plays in the choruses, and that's all there is in the track with regards to music! I love it. It just makes me smile! I have the SSL Channel on it, just boosting the track by 1dB. I never use the [Pro Tools] Mix window. If you were to look at it, almost everything would be at unity. So sometimes I throw on an SSL Channel and use that as a fader. I also have the [Sonnox] Oxford Inflator on the synth, with 50 percent of the effect, which adds some mid-range harmonics. The synth also has a send to the Width aux track with the MicroShift, which I also use on the vocals."
"Finn sent me the vocal tracks as they are in the session, and I then grouped them as Lead Vocals, LV Doubles, LV Harmonies and Backing Vocals. The latter are more like weird vocal sounds, nuances, ad libs, things like that. The 'Spookylight' sound is some kind of weird, highly-treated, high-pitched laugh. I like mixing in group-fashion, especially when people expect a mix the next day. All four vocal groups have the same signal chain, and I then vary the EQ a bit to give each of them a different space in the track. I might cut some mid-range out of the doubles, for example.
"The chain starts with the Pro‑Q 2, which is a low cut. That's very standard. The next plug-in is the Waves PuigChild 670, which is a compressor I use on all vocals, because I just really like the sound of it. It doesn't hit it very hard, but it smooths the vocals out a bit and adds some colour. With another singer I'd have had the Waves RVox before the 670, as a more aggressive compressor, but because Finn's stuff comes to me already processed, I don't want to add more.
"In this case the Waves De-Esser is the third plug-in, set to 6557Hz. It's my standard de-esser, and it catches some post-compression nastiness. If a vocal is really tricky, I'll sometimes add the FabFilter Pro‑DS at the end. The Waves is purely surgical and pretty transparent. You can hear when it messes up. The fourth plug-in in the vocal chain is another instance of the Pro‑Q 2, taking out some mids between 200Hz and 2kHz, and then the UAD Neve 1073 EQ, which in this particular session is just engaged. I like the colour it adds, so I just throw it on. Finally, there's the Waves Vocal Rider, which does 1.5dB of treatment. If there are some words that drop out, it brings them up, and vice versa. I do my vocal level automation after all this, by hand, using the Avid Artist Mix."
"At the bottom of the session are my eight stereo sends to my Dangerous 2-Bus. All drum tracks go to DBuss 1-2, all bass to DBuss 3-4, 5-6 is for synths, 7-8 is not used in this session, 9-10 is for Lead Vocals, 11-12 for Backing Vocals, 13-14 also is not used, and the Width aux track goes to 15-16. The Drums bus has the UAD Pultec EQP-1A, boosting 60Hz at 2, and also attenuating. I just wanted a bit more thump. All group busses have the Steven Slate Virtual Mix Rack, which basically is me trying to make everything sound more like it's going through an SSL! On the drums it's set to the British 4k console, which is like an SSL modeller, and I am driving the input, so it is clipping and adding some harmonic distortion. The backing vocal bus has the Oxford Inflator in addition to the Virtual Mix Rack.
"The signals are summed in the Dangerous 2-Bus and then go through the Dangerous compressor, and come back into Pro Tools via the two print aux tracks, on which I have my stereo mix chain, and I print them below. I feel I spend half my time trying to make each mix sound like it was not done inside a computer, and so, in addition to the Virtual Mix Rack plug-ins and the Dangerous hardware, the first thing I do on the master chain is put on the UAD Ampex ATR-102 tape machine plug-in, and then the UAD Thermionic Culture Vulture.
"The tape machine is set to 30ips, half-inch, but the one thing with Billie and Finn is that they don't like hiss. The plug-in added some hiss on the very first mix I did for them, and they asked me to turn it off. There's so much space and silence in their music that the hiss ruins that. The Culture Vulture then adds just a hair of drive, and finally there's the FabFilter Pro‑L 2, set to the preset 'EDM-Aggressive and Tight'. The only difference between the two print tracks is the level of the L2. There's also the TC Electronics Clarity M meter. All songs on the album have this master bus chain.
While Kinelski's 'Bad Guy' session looks fairly straightforward on the surface, a lot goes on under the hood. Still, with many modern mix sessions adorned with literally hundreds of plug-ins, Kinelski's mix session is a study in minimalism. This is entirely befitting of Eilish and O'Connell's minimalist approach, but also reflects some lessons Kinelski has learned, for example the mixer's Hippocratic Oath: above all, do no harm.
"Yeah, when you are young you mix with your ego, and you want to add this or that, to make your mark. As you get more experience, you realise you don't need to do as much. Sometimes mixing involves literally doing nothing. In this case, I managed to do just enough, and this reflects the great working relationship I have with Billie and Finn, where not a lot needs to be said."
Rob Kinelski's background hasn't always been in urban music. He grew up on the East Coast, and his first attempt at climbing the music industry greasy pole ended in Spinal Tap-like ignominy. "When I was in high school, I played bass in a band, and we were doing really well. We had a cross-country tour booked, and record label interest. Then, the night before we had a big showcase gig in DC, the drummer got drunk and broke his hand punching a friend in the head. He played the gig with one hand, and the label said, 'Let's try again in six months.' I was 20, and had spent five years in this band, and was burned out.
"I decided to study at the SAE Institute in New York, and after that I set up my own studio in New Jersey. I did well, but never felt I was doing anything that would attract major-label interest, so I joined Sony Music Studios in New York, started as a general assistant and eventually became an assistant engineer. I assisted on Beyoncé's second album B'Day , and that's what opened many doors for me. All of a sudden, I was in a lot of high-profile sessions. After Sony Studios closed [in 2007] I worked freelance in New York for two years, mainly cutting demos for songwriters. In 2009, I moved to LA and started working with Roc Nation management, which is how I connected with No ID. I worked with him for four years, during which time we worked with some incredible artists, including Big Sean, Nas and Common."
By the time Eilish and O'Connell made contact with Kinelski, he had been working as an independent mixer for four years, clocking up credits like Zara Larsson, Joey Bada$$ and Vince Staples. His time with No ID in particular had grounded him in hip-hop and R&B, but, says the mixer, his background was never really a topic of discussion.
Rob Kinelski mixes mainly in software, but uses a chain of analogue outboard equipment as part of the process. "I have Pro Tools HD Native, with the Avid I/O 16x16, and an eight-fader Avid Artist Mix, which I use quite a lot. All my mixes go through my Dangerous 2-Bus, and then into my Dangerous compressor, and back into Pro Tools. Since I mixed Billie's album, I also obtained the Dangerous Convert-AD+. Plus I have the Dangerous Monitor ST monitor controller, and my monitors are ProAc Studio 100s, Yamaha NS10s and I recently bought Avantone CLA-10s. I also have Auratones.
"When I came up at Sony I was working on SSL and Neve desks, but they gradually became giant elbow rests, because it got to the point where clients were continually asking for precise, minute revisions, which were impossible do with desk recalls. But when I first got my own iMac, I could not believe that I would be mixing records on that. I was still working with an SSL in 2013, and that year a friend gave me the Dangerous 2‑Bus, which I first used when I was working on Big Sean's second major album release, Hall Of Fame. I wanted to maintain some analogue elements during mixing, but I needed to get into the in-the-box workflow, which is super-quick and super–easy, with the ability to pull up sessions instantly and work on multiple songs every day.
"I loved the Dangerous 2‑Bus, and made it part of my sound. I also really like the Monitor ST. I think Dangerous make excellent quality stuff, at a good price point. I feel like using the Dangerous equipment gives me more harmonics, and some more depth, and a little bit more separation than if I did everything purely in the box. I tend to hit the 2‑Bus pretty hard, with a +6dB boost on every channel, and I just back off the output about 4dB. When I A/B it with an in-the-box mix, it feels more open and transparent. Then I got the Dangerous compressor, because I wanted to have yet more analogue stuff on my master bus chain. I love that as well. It's really transparent and it makes my mixes come alive. It adds a musical element. It has some cool features, like the bass cut and the sibilance boost in the internal side-chain. I believe the bass cut bypasses some sub frequencies from triggering the compression. That allows the sound to stay open and punchy. I don't know what the sibilance boost does, but I have it engaged anyway. I usually have the compressor set to 4:1 ratio, stereo linked."