Achieving a naturalistic sound on a Hollywood film set isn’t easy — but was central to director Damien Chazelle’s vision for La La Land.
Unless you have spent the last few months lost in the jungle, you will know that La La Land has become the biggest movie in a very long time. The brainchild of director Damien Chazelle and composer Justin Hurwitz, the musical has, at the time of writing, won a record-breaking seven Golden Globe Awards and garnered a record-equalling 14 Academy Award nominations. Sales of the soundtrack album have also snowballed, taking it to the top of the UK album chart and to number two in the US. The virtuoso opening scene, in which hundreds of dancers and singers strut their stuff in a huge traffic jam on a sun-drenched LA freeway overpass, accompanied by an infectious Latin jazz track, has already become iconic.
La La Land traces the fortunes and misfortunes of jazz pianist Sebastian (Ryan Gosling) and aspiring actress Mia (Emma Stone), as they try to balance the pursuit of their professional dreams with the needs of their romantic relationship. Both the visuals and the music are to a significant degree a homage not only to Hollywood musicals from the ’40s and ’50s, but to the romanticism of the French New Wave musicals of Jacques Demy and Michel Legrand. The narrative and visualisations move gradually from the primary-coloured and idealised to the gritty and realistic, resulting in an emotionally affecting ending that does not conform to the Hollywood feel-good stereotype.
A major factor in the movie’s strong emotional impact is that the jazz-flavoured soundtrack favours the gritty and realistic over Hollywood schmaltz and smoothness. In the quest for authenticity and emotional honesty, the on-screen musicians and actors had to look and sound 100 percent convincing, and one obvious but technically complicated way of achieving that was to record as much of the singing live on set as possible. Executive music producer Marius de Vries and music engineer and mixer Nicholai Baxter faced many challenges during the movie’s two-year gestation, and there were times when recording, post and mix sessions required extreme working schedules of 16 hours a day, with some DAW sessions ballooning to Pro Tools' maximum of 768 voices.
“The recording and mixing sessions were among the most intense and complex I have ever done,” recalls de Vries. “Everyone was pushed to extremes. Nick hardly slept for the last six months! Damien has such an eye and ear for detail. The tiniest vibraphone note that was played too soft or too hard, or with the wrong mallet, had the potential to bring the whole thing down. Add the many on-set recordings, Damien’s tendency to shoot in long, unbroken takes, and the complexity of much of the orchestration, and it meant that the music tech involved was extreme.”
De Vries is talking from his Berry Drive studio, in the Arts District of downtown LA. Originally from London, de Vries actually enjoyed a brief spell in the 1980s writing for Sound On Sound, and also played keyboards in the Blow Monkeys. In the early 1990s he became renowned as a programmer and producer, and he has since worked with the likes of Björk, Annie Lennox, David Bowie, U2, Madonna, Robbie Robertson, Rufus Wainwright and countless others. De Vries’ first foray into movie soundtracks came with Baz Luhrmann’s Romeo + Juliet (1996), and he later went on to become music director of the same filmmaker’s Moulin Rouge (2001). As soundtrack work came to be his staple, he moved to the US in 2010, where he worked on movies like Kick-Ass (2010), Sucker Punch (2011) and George Lucas’s animated fairytale musical Strange Magic (2014).
In April 2014, de Vries was invited to a meeting with Damien Chazelle, after an introduction via a mutual friend. “I met with Damien and the two original producers for the movie, Fred Berger and Jordan Horowitz, as well as Justin. We very quickly found that we had a common musical language and understanding, and I was the first crew member they hired for the project. I worked on the movie pretty much full-time for two years, though later in the process in some rare moments of down time I produced Rufus Wainwright’s Take All My Loves: 9 Shakespeare Sonnets album, and made some progress with a forthcoming jazz-influenced album for Chrissie Hynde.
“The direction of La La Land, visually, thematically and musically, was very much in Damien’s head when I first met him, and Justin also already had written a significant amount of the music and had gotten a head start with the orchestration. It was always Justin’s ambition to orchestrate the whole thing, which is one of several great stories about this movie: how he managed as someone relatively inexperienced, particularly at this level, to single-handedly orchestrate the music to such an impressive degree. He was working with an old version of Finale and pencil and paper at the time! Of course, there’s nothing wrong with that, and it speaks to the way he thinks through his compositions and orchestrations, but I introduced him to Logic and taught him some tricks of the trade. But, for all the support network we built around him — and it was a phenomenal team — the orchestration is all Justin’s work, and it’s remarkable that he pulled it off under that pressure.”
The company of Chazelle, Hurwitz, Berger, Horowitz and de Vries spent the rest of 2014 further defining the movie’s musical language. “One of my main challenges was to respect all the influences they were paying homage to,” explained de Vries, “and at the same time try to make sure it did not sound like a museum piece. You want to be true to your influences, but you don’t want to sound precious or old-fashioned. If we’d tried to make it sound exactly like the Michel Legrand movies, in the context of the modern cinematic experience it would have likely sounded thin and uninvolving. Everyone agreed that the music needed to sound contemporary, but that’s a very hard word to quantify, and in Hollywood, the default wisdom can be to look at the most successful last thing that’s vaguely comparable and try to imitate that. It’s a kind of job-protection strategy, because if things go wrong, you can always say: ‘But look, it worked there!’
“So here was a young director belying his age, with a very detailed and profound vision for his movie, and trying to be modern and voguish would clearly go against the grain of the aesthetic he was after. Instead, to my mind, the first and most obvious thing to do to make it sound ‘contemporary’ was simply to execute it as well as possible. This meant making full use of the entire lexicon of modern recording techniques, which enabled us to make it sound like a modern film score. But we’d have to do this to a great extent invisibly. Rather than use modern technology to polish and perfect things, we tried to use it in a way that would preserve the realness of the instrumental performances and the integrity and fragility of the vocal performances. We could make things sound big where they needed to sound big, and detailed and precise when they needed to be detailed and precise, and we could have vocals that sounded largely unproduced, and yet not technically inept.”
In terms of the music production, an enormous amount of work went into getting the music right for the opening scene, titled ‘Another Day Of Sun’. De Vries: “That song was, perhaps more than any of the others, a journey of exploration, because we knew it was going to be the opening number, and so set the tone of the music for the rest of the film. Trying to resolve the issue of how demonstrably modern we were going to be with this track meant that ‘Another Day Of Sun’ went through many different iterations, even right until the final mix stage. Some of the versions were more connected to the pop and radio cultures of today, particularly with regard to the rhythm section, but the final version was by and large quite traditional, just with some modern elements in certain places to add scope, depth and size.
“That whole razor’s edge balance between on the one hand historicity and homage, and on the other feeling contemporary and of our own time, was at the heart of everything we did in the movie. It’s addressed head-on in so many words in the movie itself, in the scene where Ryan Gosling’s character, Seb, who is a traditionalist, sits down with John Legend’s character Keith and Seb’s view is that if jazz isn’t preserved, it will die, whereas John’s character says that if jazz does not evolve, it will die. That’s the dilemma. The thing about La La Land in the hands of Damien’s direction is that it addresses that issue with intelligence and complexity and does not take a reductive position on it. In fact, several of the crises in the narrative are drawn from that dilemma.
“For me I guess it’s a particularly meaningful topic, because I have during much of my career found myself on the bridge between an embrace of past styles and traditions, and the latest musical developments, often, for example, working with orchestras and very modern musical methodologies at the same time. It was the case with Rufus’s 9 Sonnets album, and it’s similar with Chrissie Hynde’s upcoming album, which mixes traditional jazz arrangements with a mix and production approach influenced by dub music.”
De Vries and co defined the musical language for the movie during pre-production by creating sample-based mock-ups of the songs, the dance numbers and some of the orchestral scores, and by tracking all the performance jazz music. “Our first deadline was set as the Spring of 2015, and the aim was to have recordings of a high enough standard to shoot to, and for Damien to use as background to his montages in order to judge the pacing of his shots and so forth. Justin and I spent the best part of a year sitting daily at my studio, with Damien, Fred and Jordan coming in and out, and together we were a sort of committee, with each one of us having to sign off on every decision that was made. If we were not in unanimous agreement, we’d go back and try again until we found a place where we were. This set the quality bar very high.
“When Justin and I were doing these MIDI mock-ups of his scores, every single instrument in the orchestrations was paid exact attention to, so this was a very detailed, painstaking and demanding process. We were assisted in this by Eldad Guetta and Matt Robertson, who were over in London at the time. I’ve worked with both for a long time, and they regularly help me with my productions and mock-ups, and for this reason we have almost identical, synchronised technical setups. With Eldad and Matt being on the other side of the planet, we had a kind of 24-hour, round-the-clock, international operation going!”
The end of 2014 saw the first recording sessions with live musicians, at LA’s Conway Studio C. De Vries: “As I mentioned before, we wanted above all for the soundtrack to sound real and authentic, and for the jazz material, getting the input of live musicians at an early stage was a crucial part of that. We wanted to have great musicians playing in the service of communicating emotion, rather than any kind of pursuit of perfection. And we were not going for mellow, we were going for fierce. So we assembled a stellar jazz band, including Peter Erskine on drums, Kevin Axt on bass, Wayne Bergeron on trumpet, Bob Sheppard on sax and Andy Martin on trombone, and of course Randy Kerber on piano. The great jazz flavours you hear in the movie are first and foremost to do with them. No matter where you point your microphones, if you start with great players who have great studio chops, your job as a producer or engineer will be a lot easier.
“In terms of the actual approach we took in recording them, during one of many pre-production meetings, Gustavo Borner, the owner of Igloo Studios, where we later did the music mixes, coined the phrase ‘vintage plus’, which later became ‘vintage plus plus’. The vintage part meant acknowledging the sound of Michel Legrand’s scores and other influences, and to an extent embracing and celebrating imperfection. At the same time, we live in an age where we have surround mixing, perspective mixing and opportunities for extremely wide dynamics, and we needed to be able to fill the theatre. This meant that we had to have all frequencies and perspectives covered and wanted many different options for where we could place instruments in the room. It all goes back to using technology to the max, yet trying to keep it as unobtrusive as possible.”
The December 2014 recordings at Conway Studio C marked the first time that de Vries’ regular engineer and mixer, Nicholai Baxter, joined the La La Land fray. A graduate from Boston’s Berklee College of Music, Baxter has worked at Igloo Music Studios for 10 years and with de Vries since Sucker Punch. A three-time Grammy winner, Baxter’s versatility in working with rock, jazz and orchestral music was essential.
“Justin came in with some clear-cut melodies and structures,” recalls Baxter, “and the band experimented and improvised with these. This became much of the live music that you see and hear in the movie, performed by the jazz combos playing in clubs. One of the goals of these sessions was to experiment with the live sound. Marius encouraged me to commit to a great extent to any additional processing, and almost treat these recording sessions as a mix as well. We put up a variety of microphone options, incorporated a lot of analogue EQ and compression into the recording chains and aimed to get things sounding fairly finalised. I also had a vintage mic option and a clean, transparent mic option on many of the elements. Having both available later on to suit a variety of different contexts in the film proved invaluable. The many mics also gave me additional material to pan and fill the surround sound field in the mix later on.
“The microphones were an Electro-Voice RE20 on the kick inside, and a Neumann U47 FET on the outside; on the snare, a Shure SM57 and Schoeps CMC5 on the top and a Neumann KM84 at the bottom. The hi-hat also had a KM84 and the toms were all Sennheiser MD421s. Overheads were two U87s and two Coles 4038s, and I had a mono kit mic, a U47, and two Royer R121 ribbons for room mics, plus two RCA77s as drum surround mics. The upright bass had a Neumann M49 on the body, and a CMC5 on the neck, plus a DI. I had an SM57 on the guitar cabinet, and two AKG C414s on the acoustic guitar. I recorded the trumpet with a 121, the sax with an M49, the trombone with an AEA 44, and my brass room mics were Telefunken ELA M 251s.
“The piano was a special case [see ‘Method Piano Playing’ box]. I recorded our main six-foot 1920s Steinway with a pair of [DPA] 4011s for a transparent feel and a pair of 4038s for a more dark and vibrant option. I also tracked a Neumann TLM170 as a mono option. There was a Kawai upright which I recorded with a pair of C414s and a pair of U67s, plus there was a spinet piano recorded with a pair of U87s. I recorded most of the mics through the desk, a Neve 88R, straight into Pro Tools, which we ran at 48kHz/24-bit for the entire project. I used the desk pres for pretty much everything, and most of the studio’s outboard, including Pultec EQP1s, UREI 1176, Tube-Tech CL1B, dbx 160, Neve 1081 and 33609, [Empirical Labs] Distressors, [Teletronix] LA-2A and so on.”
De Vries, Hurwitz, Guetta and Robertson then spent the first half of 2015 further developing the mock-ups of Hurwitz’s compositions, and by this stage de Vries was also recording vocals at his studio, for demo and on-set playback purposes. For the vocal recordings de Vries used his tried-and-tested recording chain of a Neumann U87 through an Avalon 737SP preamp into an Apogee Quartet.
In July 2015, a month before shooting, there was another recording session at Conway, to replace the programmed mock-ups of the jazz material. Baxter: “We got the same band in as half a year earlier, this time with Gary Novak on drums, and part of the aim was to replace the rhythm section in the programmed mock-ups of the songs, to make sure the on-set playbacks were a little more organic. We recorded drums, bass, piano, guitars and a small brass section, but for the moment we kept the orchestral mock-ups. The songs we worked on were big numbers like ‘Another Day Of Sun’ and ‘Someone In the Crowd’. I did a full recall of my recording setup from December.
“Marius and I then combed through all the sessions, figured out what we needed for on-set playback, and created playback sessions with a decent complement of stems, usually about 10 or 15 per song. I made sure that we had enough separation between the vocals and other elements that we might want to edit on set. We would need to be able to adjust vocals and musical elements to aid and influence performances or help with sync. And then we went to set, which was a couple of months of wonderful chaos!”
Shooting took place during August, September and early October 2015. Baxter, de Vries, and the rest of the sound team introduced many challenges for themselves by recording many singing performances on set, particularly in the more intimate scenes with Gosling and/or Stone.
De Vries: “Whenever you set out to do a musical, whether on stage or in a movie, that ‘membrane moment’ when you go from dialogue to song is crucial. In a standard movie musical, or West End or Broadway show, the audience has a tacit expectation that when the performers start singing, they will tend to perform in a way that is somewhat mannered. There are many people who don’t like this, and La La Land has been called a musical for people who don’t like musicals, in part because the singing feels very natural, and totally connected to the way that the characters behave and speak. Ryan in particular expressed a keen interest in that sense of continuity. The 10 or so instants where speech turns into song, and then turns back into speech, were the 10 or 20 most crucial moments for us from a technical and artistic standpoint and from a believability perspective. Our biggest influence was the film The Umbrellas Of Cherbourg, which is a sung-through musical, and so avoids that membrane problem. But that was not what we set out to do. Instead I think we arrived at a very successful solution, which hopefully works for people who are slightly allergic to those transition moments.”
Part of the challenge was to make sure the sound of the voices and of the surrounding acoustics were the same during dialogue and singing. This meant recording dialogue and singing with the same equipment, in the same place and at the same time. De Vries: “The first time we tried this was in the scene during the ‘Someone In The Crowd’ section where Mia sings to herself in front of a mirror. We recorded her vocals with a lapel mic and a boom mic just above her. Justin was in another room, accompanying her on a piano, with a video feed of Emma singing. She was hearing Justin’s piano in her earwigs. We rehearsed that a lot and in the end we got to the stage where the actors could act through the songs in the moment of performance without being enslaved either by backing track or click. Generally, the singer would lead and Justin would follow. We also worked like this in the songs ‘City Of Stars’, with Ryan and Emma at the piano, and ‘Audition’, with just Mia singing.”
This approach was similar in some ways to that used during the on-set recordings for Les Miserables (see SOS April and May 2013), and de Vries consulted some members of the Les Miserables sound crew, including celebrated re-recording mixer Andy Nelson, who did the final mixes for both movies. However, there’s no spoken word in Les Miserables, so the sound team of that movie did not have to deal with what de Vries calls the ‘membrane moment’ from dialogue to song. This had far-reaching consequences for the recording approach on set, but also, as we shall see, during mixing.
A particular example was ‘Audition’, another tour de force single take where Emma Stone moves from talking softly to singing in a whisper to belting, while the camera first zooms in on her and then circles around her. Once again music engineer Baxter and production sound mixer Steve Morrow, who was responsible for recording the spoken dialogue, had to work closely together. De Vries: “The boom operator was hanging from the ceiling and managed to keep the microphone stable in relation to her face. Together with the lapel microphone Nick and Steve were able to maintain an accurate perspective. The bigger problem was that as the camera had to move around the room, first towards her and then around her, there was furniture in the room that was in the way. So as soon as the camera had gone over the desk there was a team of people very quietly moving the desk out of the way, and I am sure on the final version you can still hear the vestiges of a scraping noise of the desk being moved. Then, as the camera came back at the end they had to put the desk back. Small things like that took a lot of rehearsal and planning, and meanwhile we had to keep Emma in the zone, and for her not to be too aware of the fact that there was all this extraneous stuff going on to make the shot possible.
“Even with the ensemble numbers, for example ‘Someone In The Crowd’, we had boom and lapel microphones on the girls throughout, because it was important to encourage them not just to lip-sync but to sing as well. That’s crucial. And then we would often do three or four takes where we would turn off the music and just put the music into their earwigs and record their vocals live on a silenced set. A few bits of those performances ended up in the final mix. So it is not just the quiet sequences in the obvious scenes. And Emma was very insistent that she wanted to do the lines in that song live from the outset, even the loud bits with everyone else around her.”
Baxter: “The boom mic we used was a Sennheiser MKH50, which has a slightly wider polar pattern than the MKH416, but it has more low end and is more natural-sounding. The lapel mics we used were Sanken COS11s. The same mics were used during ADR [post-production dialogue replacement] later on. We recorded everything on set that we could without disrupting the shot. In post-production you often don’t have all the elements that you need. Little mistakes are made on set, or little details are overlooked because there is so much going on and everybody has their hands full. Our goal was to make sure no details or opportunities were overlooked, whether it be in the musical performances or our on-set recordings. We had microphones on almost all the singers, on all the musicians that you see on screen, and in some cases room mics to capture ambience. In addition, I created impulse responses in Altiverb for every environment that we were shooting and recording audio in, so I could later digitally place newly recorded instruments and singers back in the spaces that you are seeing on screen. It proved to be amazingly valuable during mixing.”
Before the final mix stage could take place, however, there was the small matter of months of picture and audio post-production, the latter occupying February and March of 2016. Audio post-production consisted of vocal recordings, more sessions at Conway with the jazz band, building tempo maps for the orchestral overdubs, the actual orchestral sessions at the Barbra Streisand Scoring Stage at Sony Pictures, and endless rough mixing.
De Vries: “During post-production we were constantly comparing the live material to the pre-recorded material. In many cases the live material was dramatically more convincing, but obviously it often was compromised by on-set extraneous noises or by a microphone being off-axis, or because of some other issue. So we were constantly making judgements between what was technically usable versus what was dramatically desirable. We had anticipated that by always recording the main performers with a boom mic and a lapel mic, in addition to the main vocal mic, and this method carried through ADR and all other vocal dubbing. And where the live recording wasn’t usable, we could at least visually track from the waveform what the actors were doing and use the on-set audio recording as a reference point for comp or post. We were assisted by Synchro Arts’ VocAlign plug-in and technology like that to line things up.”
Baxter: “Doing the vocal recordings for ‘Another Day Of Sun’ during post was an exacting process. They brought in hundreds of singers to try and sing the leads, the main issue being that Damien did not want it to sound too Broadway, too old-school musical. He did not want it to sound too showy or professional. So Marius and I ended up singing some parts in that song! But in fact, ‘Another Day Of Sun’ is the most Broadway song of the entire movie, so we had to embrace it to a certain extent, while keeping things authentic, natural and raw, which was what Damien wanted. We also re-recorded Emma and Ryan’s vocals in ‘A Lovely Night’, because it had not been possible to record them live, as the camera was generally positioned too far away from them in the scene. Here I also used Marius’s Neumann 87/Avalon 737 chain, to stay consistent with what Marius had used during demoing at his place. We kept Ryan’s demo of ‘City Of Stars’, because the take he had captured with Marius at Berry Drive right at the end of pre-production was already great.
“After the Igloo vocal recordings — and well into post-production — we tracked for a few more days at Conway with the same musicians we used before, in April 2016. We re-recorded the rhythm section for ‘Another Day Of Sun’ and ‘Someone In The Crowd’ because the current versions were judged to be a little too pop-rock, and the note was given to maintain a more consistent jazz-influenced thread throughout the movie. So we jazzed up these first two songs. On ‘Another Day Of Sun’ we ended up combining the post-production recording, which had more of a jazz-combo feel, with the big rock drum fills from the session in July for extra impact. Randy Kerber also returned to replay a few sections of his piano parts to make sure they locked perfectly with what was happening on screen. This definitely drove him crazy, because he had to replace some of the virtuoso stuff which he had already played, which because it was mostly improvised was just near impossible. But Randy is a genius, so he pulled it off.
“Throughout the whole of post-production I was doing rough mixes and preparing things for previews. We did a few temporary dubs of the entire soundtrack for previews and test screenings, which were a lot of work. Finally, just before the final dub began with re-recording mixer Andy Nelson, we did three days of orchestral recordings at the Barbra Streisand Scoring Stage. Those were incredible days, and were probably my favourite part of the whole process. There is nothing quite like manning the controls in front of a 95-piece orchestra at Sony Music, where they did all the MGM scores! My approach in recording the orchestra was pretty traditional. We were aiming for a classic big-room orchestra sound with modern flexibility. I added a second layer of more colourful mic options to give us choices, and some extra mics in the rafters for Atmos, Dolby’s 64-channel format that most movies are mixed in today. I also had an additional mic setup on the brass for various big-band moments in the score.”
Nick Baxter conducted the final music mixes at Igloo Music Studio D, where he has an Avid Icon controller and B&W 801D monitors for LCR and B&W 802s as surrounds. “For stuff that needs to be recalled and tweaked at a minute’s notice there’s just no time to recall a full board mix, so during mixing there was little outboard gear used. It was mostly plug-ins. One thing that was unusual in doing the music mixes for La La Land was that I also worked with dialogue, to make sure the transition to song was continuous. Damien really didn’t want that jarring moment you feel in classic musicals where the entire sound of the movie changes at the start of a song.
“A lot of making that transition seamless came in the recording stage, making sure that we did not have problems with proximity effect or an overly hyped or bright vocal. Hence our use of slightly more distant, natural microphone techniques with a boom, instead of a standard close vocal microphone. For some transitions I delivered a handle, ie. a little bit of dialogue before the song starts, to make sure that the sonic hand-off was smooth. My impulse responses definitely came in very handy, because they allowed me to mix in the matching ambience around the vocals where it was missing. I delivered these impulse responses separately to Andy as well.”
Baxter: “The session for this song is maxed out at 768 voices, which is the most you can have in Pro Tools! Most of them are vocal tracks, because we needed to have all the harmony parts and vocal groups available. Damien and Justin were still figuring out what vocals they wanted to feature, so everything needed to be in the session, even if we did not feature it throughout. We also recorded a full 40-piece choir at Sony, which was pulled out for most of the song, again because it sounded overly theatrical. We did use it extensively in the ‘Epilogue’. The song ended up being a patchwork of different vocal layers, harmony parts and gang vocals that build in complexity and density throughout.
“This session was extremely dense, and one of the harder mixes that I have been involved in. The mix was also a bit of a creative exploration to find the right balance between theatrical and authentic, and it took a couple of passes of redrafting and recalibrating. I pulled the mix back two times and started again! In the end we settled on a jazz rhythm section foundation with a layer of energetic rock drum fills and a natural but punchy-sounding orchestra over the top. As was the case with most of the soundtrack, we were aiming for an authentic and natural and raw feel, making sure things never got too glossy and polished.
“Showing you the entire session with a few screen shots is impossible, but I lifted out some of the vocals. You can see I leaned on the FabFilter Pro-Q 2 EQ, the UAD 1176 compressor, the FabFilter MB multiband compressor and the Massey De-esser. It’s also tricky to show you plug-in screenshots, because most plug-ins are automated, and their settings are constantly changing. Damien and Justin really did not want that standard ‘studio’ vocal sound that is compressed or too hyped. Again, they wanted it to sound natural. This meant that I could not hit the compressors hard, which, at times, made mixing more difficult. I ended up having to do tons of volume and plug-in automation, using compression only when I really had to.
“There are very few plug-ins on the individual orchestra tracks, apart from some FabFilter Pro-Q 2 EQs and a few reverb sends. The orchestra sounds really were achieved when I recorded them at Sony. There’s also a lot of volume automation on individual spot mics to feature individual instruments or sections. The entire orchestra gets bussed to the sub, in 5.1, where most of the processing is done. Here I have the FabFilter Pro-Q 2, a UAD Shadow Hills compressor, the FabFilter Pro-MB multiband compressor and the UAD Neve 1081.
“There’s heavy automation on these plug-ins as well. The Q2 EQ is automated differently for each side, so at a certain point you may have really bright trumpets on the right and darker strings on the left, and they will need different EQ. The Shadow Hills applies just some overall compression to squeeze the orchestra a little bit and bring a little life to it. It is a warm and fat-sounding compressor. It has punch, but also helps brings out the little details. The FabFilter multiband compressor gave me another means of controlling frequencies that pop out too much, or too little. It is a really transparent multiband compressor and when you’re not compressing you don’t actually hear it, so you can leave it on and use it only at specific moments, which is great. The Neve EQ is adding some presence and low end, just to mould the overall sound of the orchestra.”
“This was also a fairly dense session. You can see Emma’s bathroom vocal in the screenshot, which is the raw vocal from the set, recorded with the boom mic. It again has the Massey De-esser and the FabFilter Pro-Q 2 and, because the vocal was pretty noisy, the iZotope RX4, to take out some of the noise. There are compressors on most of these vocal tracks, but again I had to use the absolute minimum, and I adjusted levels mostly with volume automation. The goal remained to keep the vocals as natural and non-studio-sounding as possible. The reverb sends feed an aux track with [Audio Ease’s convolution reverb] Altiverb loaded with the impulse response from the set, in this case the apartment.
“Looking at the orchestra section of the session, almost all individual tracks have the Slate Virtual Mix Rack. It’s more indicative of how I normally work with orchestras. I use it a lot, because it’s an all-round plug-in that can do what’s required, whether console emulation, EQ or compression. The submix track has the same plug-ins as in ‘Another Day Of Sun’, but there’s an ‘Orchestral Reverb’ on the send here, which feeds a TC System 6000. I wasn’t actually using that much reverb in this movie. It’s mostly dry. Too much reverb tended to rub Damien and Justin up the wrong way. If we needed more ambience, I would push the room microphones, or the natural impulse responses from the set.
“I create stems to go to dub using Master tracks in Pro Tools. There are two orchestra stems, drums, bass, piano, guitars, brass, mallets, percussion, and ‘Vox1’, and there are actually a couple more: ‘Vox2’ and a background vocals stem. Each master track has the Eiosis Air EQ, which is a super-transparent EQ that I love using on master tracks, Cytomic’s The Glue compressor to glue things together, the Crane Song Phoenix II for any additional saturation, the McDSP F202 for bass management to each surround channel, and the Massey L2007 limiter at the end. There’s also a Waves PuigTec EQP-1A on the first orchestra master track. Most cues or songs have a similar group of masters at the top. Each track has a full 5.1 surround send and a stereo fold-down. I also deliver a few Atmos stems in addition to the 5.1 stems. Atmos is fantastic for music because you can allocate the music to speakers that are just off-screen, which puts it in its own space, where it doesn’t interfere with sound effects and dialogue.
“I was mixing with picture throughout. One of the main differences between film mixing and record mixing is that every decision you make in film mixing is to serve the picture, and this often means you’re doing rides and creating balances that sound crazy by themselves, but that work great in context. I also mixed the La La Land soundtrack album. For this, I took an additional pass of mixing with the intention of creating versions more tailored for stand-alone listening. The soundtrack also calls for a stereo rather than a 5.1 mix, which changes balances. You can pump a lot of sub low end and dynamics into movie mixes, but you have to dial that back for home listening and ear buds. Eric Boulanger mastered the soundtrack album, and again, we did not want it super-compressed. The soundtrack is not flat-lined. It’s not trying to be a pop record, instead it’s more open, natural and dynamic-sounding.”
More than two years of hard work have, eventually, paid off to a greater extent than anyone could have anticipated. “Had I expected the enormous success of La La Land?” reflects Marius de Vries. “I knew it was special, but you can never tell whether something is going to be successful. But there was a feeling of uncertainty during the making, because you fear that you’re so ambitious that no-one would ever want to watch the movie. I have learned to recognise that as a good sign, because if you’re not nervous, it means you’re not taking risks, and if you’re not taking risks you’re not putting enough of yourself into it. In the end, La La Land celebrates life in all its richness and complexity and sometimes melancholy, and I think its enormous success is definitely very welcome at this stage in world history.”
La La Land’s Music Co-Producer Eldad Guetta worked closely with Marius de Vries in mocking up some of the pre-production demo recordings for La La Land, both for the jazz-tinged numbers and John Legend’s song ‘Start A Fire’. De Vries and Guetta have worked together closely for so long that they have similar setups. De Vries’s Berry Drive studio is based around a Mac Pro loaded with Logic, with an Apogee Quartet interface and KRK E8 monitors, plus MIDI controllers such as an 88-key Korg Kronos and the ROLI Seaboard Grand.
“At my own studio,” says Guetta, “I’m using a Mac Pro with a UAD Apollo Twin and an Apollo Quad, a Neumann U87, Millenia preamps, PSI A21-M speakers, and an 88-key Doepfer LMK2+ master keyboard. Marius and I both have an enormous sample library, with Native Instruments Kontakt libraries playing a crucial part. We used the Komplete libraries frequently, as well as the LASS strings, Project SAM libraries such as Symphobia, Classic Brass, True Strike and others, Sample Modeling’s brass and reed instruments, plus many customised instruments that we have gathered over the years.
“We also have tons of plug-ins and soft synths, including all the U-He synths, especially Hive, Zebra and Ace, and others such as FXpansion’s synths and drum machines, NI’s Maschine MkII, Spectrasonics’ Omnisphere and everything else by the company, and New Sonic Arts’ Granite for granular synthesis. Plus we have many analogue synths, like the ARP 2600, Minimoogs — an old Model D and a new Voyager XL — an EMS VCS3, an Analogue Systems modular, a Roland MKS80, Studio Electronics Boomstar 4075, Korg MS20, a Sherman Quad Filterbank, a Fender Rhodes, a Wurli 200A and a bunch of Dave Smith synths. Our processing plug-ins include the U-He Satin tape emulation and Presswerk compressor, and we love UAD’s LA-2A on vocals, and use their EMT 140 reverb. In general we use tons of UAD stuff. Other things we often use are the Valhalla reverbs and Shimmer, HOFA IQ-DeEsser, Gforce M-Tron, and many SoundToys plug-ins. Serato’s Pitch ’n Time is usually our first call for pitch-shifting and time-stretching.
“With regards to the jazz tunes in La La Land, even though the entire soundtrack was eventually recorded with real musicians and instruments, we created very precise MIDI mock-ups, with Damien, Justin, Marius and music supervisor Steven Gizicki zooming in on the finest details. We had to create convincing demos for some of the jazz numbers, which is quite an unusual task, but coming from a jazzy background I found it very enjoyable. To that end I used a wide selection of brass sample libraries. Depending on the context and the part, it could be Kontakt’s factory brass, or Session Horns Pro, and I occasionally fired up my breath controller to use Sample Modeling’s amazingly realistic brass and reed instruments.
“There are a couple of bits in the film where Seb is getting in or out of his car and you can hear four or five seconds of old jazz coming from his car radio. These pieces of music were actually composed by Justin and they had to not only be very realistic but also sound like they’re coming from an old record. Tape and vinyl emulations obviously played a big part — UAD’s Ampex ATR-102 and U-He’s Satin are my favourites — but UAD’s Ocean Way Studios plug-in was most helpful. Used in ‘mic’ mode as an insert it brought the MIDI brass to life in a remarkable way. The result was so impressive that we thought it wouldn’t be horrible if we ran out of session time and had to keep them in the film, but the jazz session musicians were so good they managed to also play these bits!
“‘Start A Fire’ had to sound exciting for the shoot, because we had a live band miming on stage, and hundreds of extras in the crowd. We had to move very fast, and a mixture of intensively tweaked Logic Drummer grooves and fills fit the song’s groove like a glove, and so it went to the shoot. Damien probably told the drummer to use ‘big’ gestures and months later after they had edited the scene we had to address a few points where the drums on the screen didn’t quite match the audio. At that point I had to build drum fills that fit the drummer’s movements in the picture. I did this when it was close to the final music mix, and took the opportunity to improve the drum sound further by also layering some Steven Slate Drums. I think it was the famous ‘Old Zep’ kit, which has that amazing ‘roomy yet punchy’ quality. The brilliant Nicholai Baxter and his team at Igloo took it miles further and the result is pretty realistic!”
The first La La Land recording sessions with real musicians, at Conway Studios in December 2014, involved virtuoso piano player Randy Kerber coming in to develop the language and sound for the piano playing of Ryan Gosling’s character. Marius de Vries explains: “I had suggested Randy, who is one of the world’s greatest piano players, as someone who could work with Damien and Justin and myself to achieve the right level of virtuosity in the initial recordings, and in that way give shape to Sebastian’s identity in the screenplay as a genius piano player. We told Randy not to hold back and to perform with the greatest amount of extravagance and, where necessary, flamboyance possible, and he responded by playing these extraordinarily difficult piano parts.”
So far, so good, but there was a slight problem, as de Vries explains. “We knew that much of the piano playing would be filmed in unforgiving close-up, so even before Ryan was cast, it was obvious that one of the biggest challenges was to get a male actor whose piano playing was up to the standard that the script called for. Yet Ryan had at most a beginner’s understanding of the art of piano playing. However, he’s also a very self-motivated, creative person, and when he saw the apparently impossible task of actually learning to play these fiercely difficult parts, he went: ‘I like the challenge of that.’ He spent time talking with Randy about the psychology of it all, and then I brought in a piano teacher I know, Liz Kinnon, from the Colburn School in LA, and the two hit it off.
“In the four months leading up to the film shooting Ryan practised piano for a punishing two to three hours a day! He then continued to practise the piano in his trailer every day during the shoot. He was rigorous about accuracy, really studying the pieces and getting inside them, but nonetheless, as the shoot approached, we were all getting concerned that although he was capable of accurate renditions of the simpler pieces, we weren’t seeing many fast flourishes, and his manner looked a little studied and thoughtful. He looked as if he was learning, rather than performing with the required panache and flamboyance. But Ryan seemed unfazed, and said: ‘I can do this.’
“Sure enough, in the last few days he switched gears and did a ton of work on the physicality and attitude of his performances. One of the first scenes in the shooting schedule was his first big piano solo in the club, and I’ll never forget the astonished silence which settled on the room at the end of his first take. It was kind of miraculous. What he played was a very musically intelligent approximation of the difficult sections, with perhaps a few wrong notes here and there, but the basic melodies he had totally down. So even though the piano parts you hear in the film are played by Randy, the visual representation, every single pianistic hand gesture, was Ryan’s. It’s an astonishing achievement that when you watch the movie, you don’t for a second question that Ryan’s character plays the parts.”
There were many challenges for the sound team during the hectic and intense film shoots, one of the biggest being the film’s ambitious single-take opening scene. “We had pre-recorded the music and the vocals for the opening scene,” says Marius de Vries. “The first problem was that the recording had to be distributed over a very large area in an outdoor open space, which caused a potential problem with delay, and of course the speakers could not be visible. The music also had to be satisfyingly loud, because it’s difficult to dance to quiet music. So we used car radios and some hidden monitors to create an invisible PA system across the stage. Oh — and the idea was to shoot the entire four-minute, 30-second scene in one shot, with dancers, singers, a skateboarder, someone hula-hooping, and much more, which was unbelievably complex.
“Production sound mixer Steve Morrow, music supervisor Steven Gizicki and Nick were watching the on-set visual monitors like hawks, and I was running along behind the camera as it moved about, desperately trying to make sure I didn’t come into shot, to keep an eye on all performers, particularly the singers, and the band in the truck, making sure they all did what they were supposed to be doing at the right moment. It was also unbelievably hot, around 40 degrees C. So it was a white-knuckle experience! There had been very extensive rehearsals and preparation, but something unexpected will always intervene to threaten the success of a shoot of this kind. For example, the rolling mechanism of the truck door that is pulled open mid-way through the number to reveal some musicians snapped, and needed to be fixed. We lost a couple of hours to that, with people waiting in the heat. Even without the unexpected hitches, for the performers to do the scene 40 times under those conditions was an amazing accomplishment.”
Nicholai Baxter: “The car radio sequence was one of the biggest curveballs of the entire movie. It was ridiculous, but it worked. We wired them up, along with the 10 or so speakers that we had hidden along the highway, so there never was too much delay. Plus we had two speakers on a rig, moving with the camera. The problem was that Damien wanted the ability to change his shots on set, so we could not map things out beforehand. We messed around with a couple of different game plans but had to adapt on the day. Our final solution was to have a loop of the material for each car radio playing in Pro Tools and going out to separate channels on Steve Morrow’s mixer. Steve then pulled down the material for each car as it moved out of the shot and pushed the material for the upcoming car before it came into the shot, so the actors in and around each car would already be singing or miming accurately when they came into view. We did also have some mics on set, but they were purely for sync later in post-production.”
Throughout La La Land, the only major detour away from the jazz flavour of Justin Hurwitz’s compositions is a modern pop/R&B song performed by singer John Legend, with Gosling’s character as the band’s virtuoso pianist, and, in another twist, Legend on guitar, despite his normal instrument being keyboards. The song was designed to sharpen the script’s discussion about when modernisation involves selling out, and when not. Marius de Vries further intensified the discussion by adding the NI Maschine during a rehearsal scene and getting Sebastian to play a blistering Moog-like solo on a ROLI Seaboard Grand during the live performance of the song.
De Vries: “Here it was always clear that we needed a song that was at odds with the music of the rest of the movie, because it represented a change in direction in Sebastian’s musical journey. So we had one traditional pop songwriting session, led by John in my studio. We had Damien’s brief as to roughly the kind of song he wanted, but in terms of specific content, we had more freedom here than in other parts of the story. We had a tempo and groove style in mind already; John came in singing something that could be a chorus, I put down a bass line using Logic, Justin messed around with some chords, and Angélique Cinélu, a top-line expert who often works with John, started scribbling down lyrics. Every now and then we’d pause and re-record the developing vocal on an SM58. Within three to four hours we had a recognisable version of the song.
“After that, John, perhaps inspired by watching Ryan practice the piano, decided to learn to play guitar for the live shoot, and Eldad and I worked really hard to make our demo version sound like a real band. During the actual shooting our version was played over the PA. The musicians, led by the great Keith Harris and Kaveh Rastegar, did a great job of shadowing the track, and we recorded everything they did, even the drums, though they were muted. The idea was always to re-record the track during post, but Eldad’s skill and attention had created a feeling that was just right, so in the end we kept largely to our mock-up, only adding some of the brass from the actual shoot, and a few other details, like some bass slides, a few drum fills and some bits of John’s live vocal — things that connected the audio to what could be seen on screen.”