Thursday, April 30, 2015
Wednesday, April 29, 2015
Inside Track: Secrets Of The Mix EngineersNot only did the Les Misérables crew have to capture live singing on a film set: they then had to figure out how to overdub orchestral arrangements to it!
President of Film Music for Universal Pictures Mike Knobloch was on hand at AIR Studios to take session photos of the orchestral overdubs. Here, Les Miserables director Tom Hooper (standing) consults conductor Stephen Brooker.
When Tom Hooper signed up to direct the film version of the musical Les Misérables, he insisted that all the singing would be recorded live. This, as we saw in last month's Inside Track, set a colossal challenge for the sound team led by Simon Hayes and Gerard McCann. With a lot of imaginative thinking, clever technology and a ruthless approach to silencing the film set, the team managed to capture clean recordings of all the vocal performances, mostly on DPA lavalier microphones concealed in the actors' costumes. Rather than constrain the singers to any pre-recorded backing track, accompaniment was provided live from a MIDI keyboard and relayed to the actors through earpieces.
However, getting the vocal performances in the can was only the beginning. These performances needed to be edited before having orchestration overdubbed to them. The team were dealing with 20 or more takes for each scene, nothing locked in time, and picture edits that never seemed to end, each of which required them to re-edit the vocals and guide piano. Hayes, McCann and Supervising Sound and Music Editor John Warhurst explain how they untangled this one.
Simon Hayes: "We were shooting so many vocal tracks on Les Mis that we knew that dialogue editing would be a huge job, so we brought in a very experienced dialogue editor, Tim Hands, who was on the set to familiarise himself with the tracks while we were recording. He could start getting the best out of the vocals even before post-production started. Tim and myself worked extremely closely. He was based at Pinewood Studios and I spent many of my lunch hours listening to his work and sharing information about my tracks with him. Normally, my recordings will go to the dialogue editor, and the music department won't hear them until the final mix, because they have nothing to do with them. But since the dialogue recordings were also the vocals, in this case, the music department needed to get my recordings as soon as possible, so they could treat them the same way as if they were comping vocals for an album.”
A Dialogue About Dialogue
Gerard McCann: "The vocal editing on this project was unique, and was hard to force into pre-existing job descriptions. Dialogue editing is a recognised job in film, but given that the dialogue was almost entirely sung, there's much more to dealing with it than what a dialogue editor would normally do. I'd asked Tim Hands to join the team and he did all the things he's good at. He received the cuts from the picture editors, loaded all the multi-channel vocal tracks, and edited and manipulated them like you would with normal dialogue. After that, John Warhurst, who is more experienced in working with music, started massaging the vocals to improve them. The vocal editing was really done between the two of them.”John Warhurst was one of the Music Editors on Les Misérables.
John Warhurst: "Yeah, the roles were really blurred on this project, and there was a sort of ping-pong match between ourselves and editorial calling what we were working on 'vocals' and calling it 'dialogue'. Tim loaded all the vocal and piano tracks in using Titan software. This imports the audio from an EDL [Edit Decision List] and puts it in sync with the picture. It then outputs it as a Pro Tools session, in preparation for sound editorial. Once he'd done that, he would put the tracks together as a dialogue editor would normally do, listening out for any background noise such as camera noise or footsteps. After that, I started working on it from a musical point of view, which could entail trying different combinations of takes that work better together as a musical ensemble piece or listening to the different performances. On quite a few occasions, I also edited or straightened up the piano to make it fit the picture edits, using the MIDI data that had been recorded by Gerard and Rob.
"Rather than tune or in other ways treat the vocals, our first port of call was always to try to edit in other takes, and we had a lot of options, because there would, on average, be about 20 takes of each vocal, so we could usually repair things simply through editing. Also, all credit to Simon, because 99.99 percent of all vocals were perfectly recorded, so we had an amazing safety net of alternate takes. We tried to keep any tuning work to a minimum, as you can hear these processes when overtly used, and it went against the very natural feel of the film we were trying to create. We experimented with software like Celemony's Melodyne, Synchro Arts' Revoice Pro and Serato's Pitch 'n Time, but always tried to keep the most natural, least processed option possible. Tom [Hooper] always encouraged us to let him know if we had better vocal takes, in which case he'd look again at the images and might reconsider his picture edit. I was also regularly called into the picture-editing department with questions about whether certain cuts would work or not, because the picture edits needed to make musical sense as well. A major part of my work consisted of making sure the scenes with multiple singers worked and that the singers were in sync together, particularly if they had been shot on different locations and days. The rule was that whoever was on screen, everybody else's vocal had to fit with that. As always, when working on film, three weeks later we might receive a different picture edit, and we'd have to do all the work again, re-prioritising whoever was on screen in the new edit!
"Although Tom was obviously very committed to having as little ADR [Automated Dialogue Replacement, ie. overdubbing] as possible, there was always some ADR planned in the opening scene with the slaves and convicts who are in the historic docks in Portsmouth, and who have freezing-cold sea water right up to their chests! The microphones would have broken immediately, so they had to lip-sync. There also was the sound of the engines from the big fans blowing the waves and spray, and the hydraulics for tilting the massive water vats, rain machines and so on — all in all, a very noisy place to try and record a vocal performance! When we recorded the ADR for that scene, we had people singing while pulling ropes, for it to sound as authentic as possible! A lot of the chorus material, like in 'Master Of The House' and some other big chorus sections, was sweetened with overdubs to give extra width and size. These were all recorded by Rob Edwards in an ADR studio, exactly the way Simon had recorded everyone on the set, with lavalier microphones on their chests and some tight and wide stereo mics.The original live recordings were augmented with additional chorus takes recorded in ADR studios: this photo shows the setup for a session at Goldcrest. We recorded smaller groups, like 15-30 people, at Goldcrest Studios, and larger groups at Pinewood Studios. Tom always wanted the singing to be as raucous as possible, with regional accents showing. He also wanted everything recorded as dry as possible, without reverb, which was why we recorded it in an ADR studio, which had a much tighter sound, rather than a big music studio.When you're working on a 150-minute feature film, preparation is everything. John Warhurst thus delivered his vocal edits in a consistent format, where the main actors always occupied the same tracks. "This is of the factory scene in the song 'At The End of The Day', and it gives a good overall view of the vocal template that we worked in, with the full layout of all characters. This template layout was devised by Tim Hands and Andy Nelson quite early in the process, so that Andy would know exactly what he was going to get for every reel — so when he premixed the vocals, his desk layout would always be the same.”
"We did also try some ADR lines on the lead solo vocals, just to try out all the options, but every time an ADR line went by it was as if there was a big signpost on the screen saying: 'ADR!' So we never used those. It was never any better. The thing is that all these solo vocals were recorded live in an atmosphere with quite a bit of pressure. I went to the set a few times and there always was a certain amount of tension in the air, which gave an atmosphere to the live sung performances. I've been on film sets before, and there was a very different atmosphere at Les Mis. There was a lot more pressure on the actors, not only to deliver good singing performances, but also to deliver them 26 times in a row, perfectly, with only them being able to hear the piano in their earpiece. A vocalist in a studio would never sing a song 26 times in a row to get a perfect take. We all know how a perfect vocal for song is recorded in a studio, with drop-ins of individual lines and so on. Tom preferred to shoot things in complete takes, which really helped a lot when constructing and comping vocals.”
Now Bring The Orchestra
At this point, the team had a rough vocal and picture edit, but the only musical accompaniment was their on-set guide piano. What's more, the tempo of the live performances fluctuated wildly, and more picture and hence vocal edits were yet to come. How is it possible to synchronise a live orchestra with that, particularly if you're also dealing with a director who insists on experimenting with the orchestral recordings themselves? Gerard McCann, Music Editors James Bellamy and Rael Jones, and music recording engineer and mixer Jonathan Allen explain how they managed.Music Editor Rael Jones (shown here in the control room at Abbey Road Studio 2).
Gerard McCann: "Once we had a basic picture and vocal/piano edit for a scene, and after we'd finessed these cuts with the editors to improve the musical continuity, we had to go through the audio and mark out every bar and every beat of this edited performance. From that, we built a tempo map in Pro Tools, which was used to do a sample mock-up of the orchestration. Large parts of the film were newly orchestrated by Stephen Metcalfe, who had orchestrated several of the stage versions, and Anne Dudley. The original composer, Claude-Michel Schönberg, was also involved in that process. So the three of them, with input from Tom, worked with the programmers to produce these mock-ups of the orchestrations, and eventually these mock-ups would be approved and the real orchestral parts would be recorded. But as the picture and the vocal edits changed, the tempo maps were constantly changing and the orchestrations as well. This was quite challenging for everyone — 'quite an adventure' was the term we used. You only want to spend your money on orchestral recordings once you're reasonably sure of the orchestral arrangements, and in order to aid the orchestra and the conductor, Stephen Brooker, the tempo maps were available as click tracks and as streamers, which are coloured wipes that go across a picture screen, allowing the conductor to visually follow the tempo of the orchestral mock-up. This was all very hard work: much, much harder work than doing a bespoke film score for a song that has an 89bpm set arrangement!”
James Bellamy:Music Editor James Bellamy.Photo: Tony Lewis "Yeah, we were reinventing the wheel on this one. I did many of the tempo maps, and also was the general-purpose music editor in the corridor, who was often called in by the picture editors with questions about whether certain cuts they wanted to do would work from a musical perspective. Sometimes they wanted to extend a section, which meant that we needed musical placeholders. These initially were done via MIDI piano, and were then done for real, with the orchestra often arranged by Anne Dudley.
"The tempo maps were not easy to do. Sometimes the actor would leave a huge amount of space, and we'd have these bars of 10bpm that'd last 20 seconds. Once it came to actually recording the orchestra, we needed to split that up into something that made more musical sense, and I spent a lot of time turning tempo maps that were utterly ridiculous into something that was merely ridiculous! I also helped create the MIDI tracks for the streamers, and went through these cues with Stephen Brooker. I'd hand my tempo maps to Rael, who then adjusted them as needed to make his sample demos work, and I'd get WAV files back from him. Sometimes there'd be a screening announced for an hour later, and the orchestra mock-up was not entirely in sync, and in those cases Pro Tools' Elastic Audio was a massive help. We discovered that we could import the demo from one session into a different session with a different tempo map, and the whole demo would sort of adjust itself in time with the new vocal, which was pretty invaluable.”
Rael Jones: "I spent an incredible amount of time adjusting tempo maps. In the beginning, I'd import them from a MIDI file from Gerard or James, but as time went by I would reconstruct them from scratch. This was necessary, as the only thing they had to work with was the guide piano tracks, which were not always perfectly in time with the vocals and had abrupt tempo changes that would sound impossible when played by an orchestra. So I had to smooth the tempo maps out to make the orchestral programming more natural, while making sure it fitted with the voices. New picture edits would come in at least once a week, but sometimes several times a day, often with different vocal takes and tempos, so I'd have to adjust the tempo maps again! To be able to accommodate all these changes, I worked in MIDI all the time for maximum flexibility. I did all the tempo maps and orchestral mock-ups in Logic, which I find far better for working with MIDI and sample instruments than Pro Tools. Conversely, I find that Pro Tools is far better at dealing with audio.
"Anne supplied her scores handwritten, whereas Stephen used Sibelius, so I could input his orchestrations quickly from an exported MIDI file. I had an assistant, Nick Hill, whose main task was to input notes into Logic. I believe Stephen has orchestrated Les Misérables five or six times before, so he would often go back to his earlier orchestrations and use these as a starting point. In some cases, they wanted orchestrations that were close to his approach for the West End show, but sometimes the score diverged far from that. Later on, I also orchestrated some additional material. I tried to make the orchestral mock-ups sound as good as possible — there are all sorts of expression options these days, like legato transitions, portamento, dynamic sampling and so on, which can make an orchestral sample mock-up sound very realistic if you spend the time on them. For the strings, I used LASS [LA Scoring Strings] and Spitfire Solo Strings, and occasionally Spitfire Albion. For the rest of the orchestra, I mainly used CineBrass, CineBrass Pro, CineWinds, VSL Woodwinds and Spitfire Percussion. As things went on, I moved more to the role of music editor, and James went over to AIR Lyndhurst to assist with the orchestral recordings. The picture edits continued to change after the orchestral recording, and despite experimenting with using Elastic Audio in Pro Tools, we really didn't want to compromise the sound quality, so in the end we just edited the audio as best we could to make it fit the pictures.”This screen, from the song 'I Dreamed A Dream', shows the meticulous work that was necessary to create Pro Tools sessions that the orchestra could record to. James Bellamy: "I took this session to Air Lyndhurst for the orchestra recordings. There is a tempo map at the top. 'Edit ED' is the Avid dialogue track, hence the D. Edit E is the cut of the film that we were recording this song to. They start with Edit A and work through several versions, and this is edit E. In some songs they got quite far beyond edit E, but this particular song didn't change, because it famously was done in one take. Underneath that is the MIDI track with a Pro Tools plug-in, which is bussed to the next track down, Click Rec, which is a recorded version of the output of that, which is the click for the conductor. 'RJv6' is the mocked up orchestra, meaning 'Rael Jones version 6'. The nomenclature of the orchestral mock-up files was very precise. 'MO4' is the ID for the cue of 'I Dreamed A Dream', 'R2vTH17' is the reel number, scene 28, 'NoVox' shows that this was an audio version without vocal, and it's version 6 done by Rael Jones. There's obviously an edit, I had to move the orchestral mock-up in sync. The tracks below that are MIDI tracks for streamers to give the conductor visual clues of the tempo. If you look at the 'Making The Music' clip on YouTube, you can see that there are the coloured wipes that go across the screen, and they were fired by Pro Tools via MIDI. The green streamers at the top are all on the downbeat of the bars, the blue stream is the direction to him to bring the orchestra off before she sings 'the dream I dreamed' at the end. You can see in the tempo map at the top that the tempo varies quite a lot.”Rael Jones created his orchestral mock-ups in Logic. This is the arrangement for 'I Dreamed A Dream': "You can see the tempo map at the top and on the right; it's quite a list, with hundreds of tempo changes. Tracks 1 to 3 in blue are the vocal files. 'D' is for dialogue, 'M' is the music track. That would have been the guide piano early on, and as time went by it would be my orchestral mock-ups. There's an earlier version of my demo in the greyed-out blue music track (2) and beneath that (3), version 8 of my demo. I'm working on version 10 here. I did about 15 to 20 orchestral demos per song, going up to 30 versions in some cases. Everything beneath the music tracks is MIDI orchestra, colour coded per instrumental section.”
Jonathan Allen: "Tom Hooper is a director who has the ability to imagine sounds as clearly as he can imagine pictures. He was also questioning all the time the conventions of movie scoring, with big orchestras and large dynamics, which is 90 percent of the work we do. In a case like this, you end up trying out many different options to achieve the ideas someone has in his head. By insisting on recording the dialogues live on the sets, and also by the way he filmed things, Tom had created a world that had its own laws, and the way the orchestra would work within that world took time to discover. While Tom was editing the movie, he had been listening to the actors' voices accompanied by a piano guide track, which was obviously totally subservient to the vocals. When it came to orchestral recording, he didn't want anything musical to suddenly detract from the vocal performances. In that context, an orchestra is quite a dangerous beast, because you have enormous dynamics and a huge range of colours. The theatre show is a very fast and busy experience for the audience, but much of the movie was pared down to very simple elements. So when we recorded the orchestra, we had to find a way to reflect the scale of the musical, but at the same time not detract from the emotions of the original performances.
"In terms of the way we recorded the orchestra, this had a couple of consequences. Most movie orchestras will have a string size of 45 to 50 players, but the orchestrators pared this down to 30-34 players, and this immediately also limited the size of the instruments behind the strings, so we were down to double woodwinds, four horns, simple brass and simple percussion. The other interesting thing was that Tom asked for a carpet to be put in the hall at AIR Lyndhurst. At Tom Hooper's insistence, the acoustics of the large hall at AIR Lyndhurst were damped with carpets to achieve a drier sound.Photo: Mike KnoblochThat studio has quite a live acoustic, and he didn't want the energy of the orchestra to overpower the vocals. So I put some carpet down, while still trying to retain some of the nice colours of the early reflections in the hall. But after the first day of recording, Tom said that he wanted more carpet, and we ended up putting carpet everywhere. Actually, this turned out to be a very good idea. Because Simon had recorded the voices on the set so clean and so well, without EQ or compression, and this natural approach had been carried through during vocal editing, it now moved into the orchestral recordings. I also didn't use any compression, and very little digital reverb, apart from on a few of the larger scenes, just for five or six moments in the entire movie.
"All the way through, I wanted a sound that was honest, so I used some nice ribbon microphones and placed them very close on the strings, like the Coles 4038s and the AEA RCA 44C. This allowed us to have a more tactile and dry string sound.The orchestra was tracked with numerous different miking setups simultaneously, to provide plenty of options at mixdown. AIR Lyndhurst's custom AEA 44R ribbon mics offered a vintage flavour.Photo: Mike Knobloch I also had two sets of ambient microphones, the omni Schoeps MK2Hs for the larger sequences and subcardioid MK21s for a more intimate surround sound. I edited the various orchestral takes together, as this not only helped me mix but also keep on top of the picture changes. These edits were then further tightened for sync by James and Rael.
"I mixed the music stems, and Andy [Nelson] created an initial 7.1 template with the surround information in the side speakers keeping space for the effects and atmosphere in the far rear. That was our first configuration, but when Andy tried it during dubbing, it did not seem to be working very well for Tom, who asked me whether I could reconfigure the template to spread the surround further, giving you the feeling of being enveloped within the orchestra but leaving a clear space for the dialogue at the front. So I emptied the centre channel quite considerably, leaving very little information, and spread the panning around so that the surround microphones were in the far back left and right, and then my Schoeps Mk21 outrigger wide string microphones would be available as extra side information in 7.1. Suddenly it became far easier to hear the dialogue, and there still was the feeling of being immersed within the orchestra. In essence, I did a 5.1 mix, with two untreated MK21 microphones which Andy then controlled. I really loved the fact that Tom was questioning everything, because it made the whole process much more interesting for me. It's very refreshing, and it's something that I have since carried to working on other movie scores. We should regularly question what we are doing and why, and whether we are getting the best emotional impact possible.”
The Final Hurdles
To minimise the length of cable runs, AIR Lyndhurst employ remote mic preamps which sit in the live area and send line-level signals to the Neve console. Photo: Mike Knobloch
The orchestral recordings took place at the end of October 2012, after which it was time to combine all the ingredients and see whether the whole thing would actually work. The pressure was on, because an early December London première was already planned. Eminent movie mixer Andy Nelson flew over from Los Angeles to mix the vocals and music for the soundtrack, with help from Mark Paterson who handled the special effects. They mixed at Halo Post Studio 1 in Soho, which sports the movie industry's flagship console, a 72-fader AMS-Neve DFC Gemini with a massive 1120 inputs. Nelson and Paterson might have had a major mess to sort out, but mercifully, nearly a year earlier, McCann and Hayes had already foreseen some of the problems that they might encounter.
Hayes: "When we were testing our working methods in February 2012, it immediately became apparent that as soon as you put an orchestra around a vocalist, it was completely and utterly unacceptable to have shifts in acoustics on the vocals when camera angles change. This turned out to be one of the major differences between recording dialogue and recording live singing for a movie. Normally, we celebrate the changes in natural acoustics when you switch from a close shot to a wide shot, and the microphone also moves further away, because it makes watching the movie seem more natural and realistic. But when dealing with singing it was very unnatural and drew the audience's attention to the editing process in a very uncomfortable way, as in: 'Look, now they have cut to a wide shot.' It was the same when we tried to pan the singers' voices in accordance with them shifting position in the picture. So we had to keep the acoustics we recorded on the vocals exactly the same throughout each song, and this was another reason why we prioritised the lavalier microphones, because they obviously were always in the same position in relation to the singer. These were decisions that we discussed with Andy at the time, and that he agreed with.”Andy Nelson at the Wise Mixing Stage in Fox Studios, Los Angeles.
Nelson: "When I first met Tom Hooper to discuss me mixing Les Misérables, about one and a half years ago, he immediately told me about his plans to record the vocals live. We spoke about the musicals that I mixed, like Evita , Moulin Rouge!  and Phantom Of The Opera , which were done using playback tracks and lip-sync, and I mentioned Alan Parker's The Commitments, which I mixed in 1991, for which the band was playback but the vocals were recorded live. From the audience point of view, this really made a difference to that film, but of course, it had been easy to record because the singer is standing in front of a microphone. I told Tom that I thought his plan was great, if he could pull it off. It clearly was a massive challenge to record good enough sound with lavalier microphones, keep the set quiet and have playback in the actors' ears. But the quality that Simon got using these microphones was phenomenal. Of course it's not like listening to a Neumann, because of the nature of the size of the microphone and where it's placed, but it was pretty genius. I often was astounded and would ask myself, 'My goodness, how did he do that?'
"When I began the mix in London and was sitting in front of the Neve with up to 24 tracks of live vocals in some scenes, I had to think about whether to treat the vocals as a dramatic movie that happens to be sung, or in a conventional musical way. I decided to approach it as if I was mixing a dramatic movie. When I do drama, the first thing I do is spend a week going through all the vocal tracks. I put each character on screen and his or her voice on a reference level and I will listen carefully and make adjustments for quality, like rounding off the edges with some desk EQ, adding a bit of compression if necessary, applying my Junger outboard de-esser, and so on. Because Simon had given me fully flat tracks, without compression, limiting or EQ, I could mould each track exactly as I wanted, to fit the screen, just like when mixing a dramatic movie. The film is split down in reels of roughly 15 minutes each, so I'd go through all the reels and just listen to all the vocal tracks, on their own. They are a form of communication and I have to make sure that the sound of the actor is coming at me at the right level. I have to get that aspect of the storytelling correct.
"I introduced the music after that. I received the music tracks from Jonathan, who had mixed them at Abbey Road, and they were spread out over many different channels, with close mics on strings, wider mics on strings, woodwinds separate, solo instruments separate, and so on. I had in the region of 40 music tracks, and because we mixed in 7.1, Jonathan had given me separation on the room mics to help with the surround sound. The thing with the music in this movie is that it never stops, so the question for me was how to balance that with the vocals. If I was balancing a stand-alone song in a film, I'd have the music more upfront, but you can't follow sung dialogue with orchestra over two and a half hours like that. So I had to find the appropriate balance where you could immerse yourself in the movie and follow the story without the music overwhelming everything. That took a bit of learning. I did three passes of the entire movie quickly, so we could see where we could go with it. Then we introduced the sound effects, and this would, again, lead me to go back and adjust the sound of the vocals or the orchestra. We were constantly adjusting the mix.
Gerard McCann (top) and Simon Hayes oversaw the capture of the live vocals on set. "My natural tendency when working with vocals set to music is to add some reverb to make them a little more beautiful in a traditional way. The Lexicon 960 is my favourite box for that. But the minute we tried this, we lost the rawness and realism that Tom was going for. So we took it off again. I did have some automatic desk compression on the vocal, not to control the loud material, but to push the very soft material up a bit, so you could hear every breath and mouth move. In terms of panning, I always kept the principal singers fairly stationary in the middle, wherever they were on the screen, and spread the chorus-style vocals a bit wider. The Foley sounds remained fairly low in the mix, because once a song was in full force, there usually was no reason for a sound effect to suddenly be there. We used them more during the transitions between songs, but again, you don't want the audiences to be drawn into that world and then jerked back into a song again, so even there these natural sounds couldn't be too intrusive. I again tried to have the orchestra 2dB louder to envelope the vocals more, but we again lost the nature of what Tom was trying to do, which was for the actor to run the show. Every time we tried things that we normally do, we had to back down. Every bit of gloss that we tried to apply didn't work. The soundtrack was always, from day one, going to be a dry, raw, in-your-face experience.
"The whole thing was about getting the orchestra to support the actor, never overwhelming him or her. I thought of it as 'cradling' the vocals. It was immensely challenging, also because of the picture editing that went on even as I was mixing, and the clock ticking ahead of us, but it was really great fun, and I'm very proud of the result. There was a tremendous amount of passion in the room during the mix, with producer Cameron Mackintosh, Claude-Michel Schönberg and Tom all there and seeing what they had been working on for more than a year finally come to life with all the elements in place. When we finished a reel, we'd sit back, turn the lights low, and play that reel back, and we'd all be taking notes on what worked and didn't work. I'd then address any issues, and we'd move on to the next reel, and eventually we went through the entire movie, and this then revealed more things that needed adjusting. I don't like working on small segments, because it's not the way we experience a movie. Instead, I'm a big fan of doing things in long sweeps that may lack in detail, but that give you a sense of the arc of the film. You can always go back in to do the more detailed work. Finally, the director, the producer, and the mixers sign off on the final playback and that's it, that's the soundtrack.”
The Final Analysis
All the efforts of those involved would have counted for nothing had the resulting movie been a flop, but at the time of writing — only a few weeks after release — Les Misérables has taken over $340 million at the box office. The film won three Oscars, including a thoroughly well deserved win for Andy Nelson, Mark Paterson and Simon Hayes in the Sound Mixing category. "Given how well the movie has done at the box office, one can say that recording movie vocals live on set does work,” says Gerard McCann. "I imagine that it will be difficult for any director to do a musical after this and handcuff him- or herself to pre-recorded tracks.”
Simon Hayes agrees: "I hope that what we did will have a really positive impact on sound in film in general, and that Tom Hooper's emphasis on original performances will become the norm. Shorter musical numbers in dialogue movies may still be recorded using pre-record and miming, but for films that are entirely sung through from start to finish, I honestly don't think that these movies can be shot using the pre-recorded vocal method any more. From here on, sung-through movies will have to be sung live from start to finish. I don't think there's any other choice.”
Go to Part 1 .
Mixing The Soundtrack Album
Lee McCutcheon.A typical vocal EQ chain for Lee McCutcheon's album mixes. Note also the extensive level automation!
"I mixed the album at FX Studios, London in Pro Tools 9 and using Focal SM9 monitors and the Avid Artist Mix for physical faders, but all EQ, compression, effects and so on were done in the box. My first step was to put up the print master audio of the film as soon as Andy [Nelson] had finished mixing it and listen for any edits and changes to vocal takes that had gone into the film at the last moment. Because the album was to be only one hour long, we then had to make decisions about how it was edited. The schedule was so hectic that time didn't allow us to get everything on the Highlights album that we would have liked to. We're currently looking at a creating a more complete soundtrack album.
"The soundtrack album is balanced differently from the film, with music being slightly louder against the voice, and fewer or no sound effects. While I was mixing, it quickly became apparent that a large part of the raw feel of the voices was that they were not covered in reverb. It worked for the album to put them into a small space, but anything unnatural took away from the emotion. For quite a few moments I left them completely dry, which has the effect of keeping the listener completely focused on what they are singing. I used [Audio Ease] Altiverb and [Avid] Revibe for putting these small natural spaces around the voices when needed. The Waves C6 was also used a fair bit on vocals, to grab some harsh frequencies, along with the UAD 1176 for compression. For other EQ, I used the UAD Pultec EQP1A and the SSL G-Channel plug-in by Waves. The UAD Precision De-esser came in useful on the vocals too. In a couple of places, I used an old plug-in called [Line 6] Amp Farm on the vocals, just a little bit under the main signal, to give it a little more energy and edge instead of using EQ to brighten it. A plug-in that I liked on the orchestra was the Steven Slate [VCC] virtual channel plug-in, and I had the Waves Kramer Tape plug-in to glue things together a little on the master bus.”
Orchestral Miking For Les Misérables At AIR
1 Main Room L: Neumann TLM50
2 Main Room C: Neumann TLM50
3 Main Room R: Neumann TLM50
4 Main Outrigger Wide L: Schoeps MK21
5 Main Outrigger Wide R: Schoeps MK21
6 Surround L: Schoeps MK2H
7 Surround C: Schoeps MK2H
8 Surround R: Schoeps MK2H
9 Surround subcardioid: Schoeps MK21
10 Surround subcardioid: Schoeps MK21
11 Side L1: Schoeps MK21 (not used)
12 Side R1: Schoeps MK21 (not used)
13 Side L2: Neumann M49 (not used)
14 Side R2: Neumann M49 (not used)
15 Main Room Ribbon L: AEA R44C
16 Main Room Ribbon R: AEA R44C
17 Violin 1: Neumann KM84
18 Violin 1: Neumann KM84
19 Violin 1: Neumann KM84
20 Violin 1: Neumann KM84
21 Violin 1 Low : DPA 4011
22 Violin 1 Front Ribbon 1 : Coles 4038
23 Violin 1 Front Ribbon 2 : Coles 4038
24 Cello: Neumann FET47
25 Cello : AKG C451
26 Cello : Neumann FET47
27 Cello Solo : Neumann U47
28 Viola: Schoeps MK41
29 Viola: Schoeps MK41
30 Viola solo: Neumann M147
31 Violin 2: Neumann 84
32 Violin 2: Neumann 84
33 Violin 2: Neumann 84
34 Violin 2: Neumann 84
35 Violin 2 Low: DPA 4011
36 Violin 2 Front Ribbon 1: Coles 4038
37 Violin 2 Front Ribbon 2: Coles 4038
38 Bass: Neumann U47
39 Bass: Neumann U87
40 Bass Overhead: Neumann M50
41 Horn L: Neumann U67
42 Horn R: Neumann U67
43 Horn Close 1: Shure SM57
44 Horn Close 2: Shure SM57
45 Horn Close 3: Shure SM57
46 Horn Close 4: Shure SM57
47 Horn Close 5: Shure SM57
48 Horn Close 6: Shure SM57
49 Woodwind Overhead L: Schoeps MK4
50 Woodwind Overhead R: Schoeps MK4
51 Flute 1: Coles 4038
52 Flute 2: Coles 4038
53 Oboe 1: Neumann TLM170
54 Oboe 2: Neumann TLM170
55 Clarinet 1: Neumann TLM170
56 Clarinet 2: Neumann TLM170
57 Bassoon 1: Neumann TLM170
58 Bassoon 2: Neumann TLM170
59 Trumpet 1: Neumann U87
60 Trumpet 2: Neumann U87
61 Trumpet 3: Neumann U87
62 Tenor Trombone 1: Coles 4038
63 Tenor Trombone 2: Coles 4308
64 Bass Trombone: Neumann FET47
65 Tuba: Neumann TLM170
66 Harp Low: Schoeps MK4
67 Harp High: Coles 4038
68 Harp Amb L: Schoeps MK2H
69 Harp Amb R: Schoeps MK2H
70 Guitar: Neumann KM84
71 Accordion: Neumann U87
This screen, from the song 'I Dreamed A Dream', shows the meticulous work that was necessary to create Pro Tools sessions that the orchestra could record to. James Bellamy: "I took this session to Air Lyndhurst for the orchestra recordings. There is a tempo map at the top. 'Edit ED' is the Avid dialogue track, hence the D. Edit E is the cut of the film that we were recording this song to. They start with Edit A and work through several versions, and this is edit E. In some songs they got quite far beyond edit E, but this particular song didn't change, because it famously was done in one take. Underneath that is the MIDI track with a Pro Tools plug-in, which is bussed to the next track down, Click Rec, which is a recorded version of the output of that, which is the click for the conductor. 'RJv6' is the mocked up orchestra, meaning 'Rael Jones version 6'. The nomenclature of the orchestral mock-up files was very precise. 'MO4' is the ID for the cue of 'I Dreamed A Dream', 'R2vTH17' is the reel number, scene 28, 'NoVox' shows that this was an audio version without vocal, and it's version 6 done by Rael Jones. There's obviously an edit, I had to move the orchestral mock-up in sync. The tracks below that are MIDI tracks for streamers to give the conductor visual clues of the tempo. If you look at the 'Making The Music' clip on YouTube, you can see that there are the coloured wipes that go across the screen, and they were fired by Pro Tools via MIDI. The green streamers at the top are all on the downbeat of the bars, the blue stream is the direction to him to bring the orchestra off before she sings 'the dream I dreamed' at the end. You can see in the tempo map at the top that the tempo varies quite a lot.”
Tuesday, April 28, 2015
Monday, April 27, 2015
Inside TrackTechnique : Recording / Mixing
In his quest for authenticity, director Tom Hooper drove his Les Misérables technical crew to completely reinvent film soundtrack recording.
Director Tom Hooper (centre), with Daniel Huttlestone as Gavroche, on the set of Les Misérables. Photo: James Fisher / Universal Pictures
When Tom Hooper was asked to direct Les Misérables in March 2011, the British film and TV director was riding high on the success of The King's Speech (2010). He took the role on one condition: all vocals in the completely sung-through musical would be recorded live on set. As Hooper explained in a promotional interview: "You can tell in your bones that there's something false or unreal about people singing to playback. What will be exciting for the audience is that singing it live has such a profound effect on the power of realism of this story. I thought it was an amazing opportunity to do something genuinely groundbreaking.”
Hooper's determination to record live and give the actors complete control over every creative expression, including the tempo in which they were singing, was hugely popular with the actors. Russell Crowe has stated that "There's an emotional level to this that cannot be created in the studio,” while Amanda Seyfried has said "It is so much more powerful, you have complete freedom, complete control.” For the technical crew, however, the director's vision gave rise to innumerable practical problems that had never hitherto been solved in the history of film-making. Fulfilling this brief embroiled the crew in a painstaking and lengthy process, during which they repeatedly reinvented the ways in which soundtracks for movies are recorded, edited and mixed. The word "impossible!” was exclaimed quite a few times; but the crew made it possible, and Hooper's vision been vindicated by impressive box-office takings, critical praise and an Oscar for Best Sound Mixing.
Proof Of Concept
Production Sound Mixer Simon Hayes, with some of the gear used on set to record the actors.
Tom Hooper had his first meetings with his sound and music teams in the autumn of 2011. Gerard McCann was hired as Supervising Music Editor, and Simon Hayes went on to act as Production Sound Mixer ('recording engineer' to those in the music world). McCann and Hayes explain why Hooper's demand that all the singing be recorded live was widely deemed "impossible”.
"We heard that word a lot,” explained McCann, "but to be fair, it was usually used by people from outside our team. They'd heard what we were doing and came up to us and said things like: 'We don't see how this is going to work. How will you be able to record vocals of good enough sound quality? And how will you cope with noise on the set? And how will you edit things together with vocal takes that are sung in different tempi? And how are you going to add an orchestra to that?'
"We also never assumed it was going to be easy, and it certainly didn't turn out to be easier than we initially thought! There were problems that we anticipated and had put plans into action to deal with, but there also were complications that we hadn't foreseen and for which we had to find solutions on the spot. We also had to demonstrate to the film's backers that we were able to do this: if you are going to invest millions of dollars in a movie and somebody says: 'This is how we are going to do it,' but he can't point to it having been done before, you are going to want to see some proof! So in February 2012, towards the end of seven or eight weeks of rehearsals, we did some test shoots to be able to demonstrate the whole process from A to Z. We tempo-mapped the vocals out and added a small 15-piece orchestra. This rehearsal period was also helpful because we needed a sense of how long the whole thing was going to run, since the film version is not identical to the stage musical: it has new verses, a new song, different song orders, and so on.”
"Tom was also absolutely adamant,” elaborates Simon Hayes, "that he didn't want to take the huge risk of recording all the vocals for Les Mis live, and put everyone through the extra time and effort involved — including asking the actors to train so they could sing for 12 hours a day — only [for them] to be told afterwards that their performances would be committed to ADR [Automated Dialogue Replacement, ie. overdubbing after the fact] to be re-recorded anyway. That would have been pure folly, because if you're going to do ADR, then you might as well have recorded the movie in the traditional way and have the actors lip-sync, and save yourself a lot of time and trouble. We had to make absolutely sure that what we set out to do was achievable.”
Supervising Music Editor Gerard McCann.
While the approach taken by Hooper and his crew was indeed groundbreaking, some have gone too far in claiming that it was the first time singing was recorded live on film sets or locations. "There have been some misunderstandings,” says McCann, "because obviously Les Misérables was not the first movie for which actors were recorded singing live. When sound was first introduced to film in the late '20s, the singing and accompaniments for musicals were recorded live on the set, but gradually the standard way of recording musicals became to pre-record all the vocals to a backing track in a music studio, and play these perfected recordings back during shooting via speakers on the set, while the actors mime to their earlier performances.
"What really makes Les Misérables different is that Tom wanted to give the performers 100 percent flexibility. He didn't want them tied to anything, whether a tempo or a pre-recorded backing track, and certainly not to a musical performance that they had done in the studio two months before they came on the set. Instead, he wanted to give the actors the same freedom to express themselves as they have when they're acting drama. Being in the moment with the timing of a performance is what actors and directors seek the most in drama, and Tom wanted them to have exactly that, in a sung environment, including when there are several actors on the screen musically responding to each other. So the actors had the freedom to perform faster or slower, hold up a certain cadence or sentence or word while they have a thought, walk to the other side of the room before they sing the next line, sing loudly or softly, and so on. The actors could do things that were much more intimate than on a stage; for example, Anne Hathaway singing 'I Dreamed A Dream' almost in a whisper. Because the musical is a big emotional piece, I think it would have been wrong to have done it any other way. That freedom was what made it exciting and worthwhile for everyone.”
"Several movies have flirted with live singing over the years,” adds Hayes. "For example, I did a small amount of live vocal recording with Meryl Streep in Mamma Mia . Ninety-five percent of that movie was lip-sync'ed in the normal way, but there was one song that Meryl wanted to record live, because she was climbing a wall in that scene, and she felt that the pre-record she'd done three months earlier was not going to represent her facial expressions or the efforts that she was making while she was climbing. She thought that it would look false. That situation was a good example of the disconnect that can happen with miming. In addition, with normal musicals you have three- or four-minute dance routines interspersed with dialogue scenes, and you're only asking the audiences to believe miming for short periods of time. Tom felt very strongly that for a sung-through movie like Les Misérables, in which there's very little dialogue, it was taking things too far to ask audiences to watch two and a half hours of mimed performances. I also don't think that would have been believable. Recording the actors the way we did meant that they could really focus on making their feelings apparent, without having to worry about falling behind the musical tempo.”
Simon Hayes (left), with his on-set Les Misérables sound team. From left: Hayes, Paul Schwartz (audio maintenance), Arthur Fenn (key first assistant, sound), James Gibb (second assistant, sound), Robin Johnson (first assistant, sound).
This is all very well in principle, but created enormous technical challenges. One consequence was that all film departments had to change their working methods and work much more closely together than they normally do. "Tom was so single-minded that he did not want to re-record anything,” recalls Hayes, "that he called a meeting with all the heads of all departments before we started shooting, during which we all talked through the ramifications of his aim. For example, Tom wanted all department heads to talk to their crews to make sure that there was a completely quiet set, and anything that could be noisy should be addressed. Gerard and I also decided that separating the music and the sound departments wasn't going to work for Les Mis, in part because the singing was treated as both dialogue and as the leading musical element. Our jobs were completely interdependent. It was a massive collaboration.
"Another example is that normally the production team rarely talks to the post-production sound team, but with Les Mis the sound and music post-production teams were brought in from the very early stages to discuss every single decision, so that we had a workflow from pre-production, through production, all the way up till the final mix in post-production. Every single member of these teams had their input in every single decision that was made. This meant, for example, that Andy Nelson, the re-recording mixer, had many conference calls with Gerard and I and the rest of the departments, during which we discussed various scenarios and made sure that he agreed with the decisions we made on the movie set.” ['Re-recording mixer' is another film nomenclature anomaly. He or she mixes all sonic elements together at the end of the project, but doesn't re-record anything, other than print the final mixes, so in the music world this person would simply be known as a mixer.]
Booms Or Lavs?
One of the most obvious challenges was how to record high-quality live vocals and allow the singers the freedom to move around freely, yet keep the microphones off camera and make sure that the sets were completely quiet. Simon Hayes: "Normally on a film set, the production sound mixer prioritises boom microphones, because they are large, high-quality condenser microphones and they are connected via cables, and give very high-quality sound. Only when boom microphones cannot be deployed effectively, usually because you're dealing with a wide shot and they are pushed too wide, or because of shadows from lamps, do you rely on lavalier microphones. These days, we multitrack and record both boom and lavalier mics, but we still prioritise the booms. Lavalier microphones have smaller capsules, we have to hide them in the actor's clothing, and you have to connect them to a radio pack and transmit the signal via radio waves. All this means that the sound is inferior to that of a boom mic.
"There was an additional problem with Les Mis, which was that Tom wanted to shoot every take with wide, mid and close camera angles, so if he got a great performance from the cast, he had every angle covered and would have coverage for every picture or vocal edit that was done during post-production. Also, if for example Hugh [Jackman] and Russell [Crowe] were singing together, Tom wanted cameras on each of them at the same time, rather than doing things in the more traditional way, where you first have the camera on one person, who will act his or her part, and then you film the other person. When you're shooting multi-camera, you're automatically prioritising radio microphones, because the booms would not be able to get in close enough with the wide shots. If a boom microphone is stuck at the edge of a wide frame, the acoustics aren't going to match any of the close and mid shots.
"Tom was aware that there have been huge technical advances in the sound, picture and electronics industries, and he told me that he wanted me to use every piece of modern technology available to capture the actors' singing live. I told him about the DPA lavalier microphones, which I discovered a few years ago when doing that live take with Meryl Streep for Mamma Mia. They were designed for both musical instruments and vocals, and sound almost as good as [conventional] condenser microphones. They maybe aren't 100 percent as good, but they're much better than any other lavaliers that are on the market. They have a flat frequency response and don't start to sound chesty or constricted in the way lavaliers normally do. They can also cope with very loud SPLs, which are notoriously difficult for lavaliers. Normally, if someone screams into a small-capsule lavalier mic, it will distort and you have a square wave.
The need to accommodate wide camera angles made DPA lavalier mics central to the sound team's efforts to capture high-quality live singing on set. These were disguised as much as possible using fabrics from the actors' costumes, and later removed from the images using CGI where necessary. The signals were transmitted to Simon Hayes' recording rig using Lectrosonics radio systems."So I decided to use the DPA 4071 lavalier mics, and the next issue was the range and frequency response of the microphone radio transmitter. We are now at a stage where the Lectrosonics Digital Hybrid Wireless system has a quality that is almost indistinguishable from that of a cable. We actually tested the DPA and Lectrosonics system at Abbey Road, with the music editing team, against condenser mics using professional singers. The music department guys wanted to be sure that they'd have good enough quality vocals to work with! They are used to having Neumann U87s or other vocal mics in a position of priority, ie. there's no camera stopping them from having mics where they want them, and they were very worried that the sound of the lavalier mics would be extremely poor, also because we would not be able to place them right in front of the actors' mouths. But they were pleasantly surprised. They thought that we would be getting 20 percent or so of the quality of a music-studio vocal recording, but it was about 60 percent. They then asked me to use no EQ, compression or limiting whatsoever while recording, so they'd have vocal recordings with full dynamic and frequency ranges to work with.
"That left us with the issue of where to place the lavalier microphones. This is a huge problem in the film industry. When you hide them underneath the actor's clothes, you are at the mercy of the costume he or she is wearing. Some costumes aren't very noisy, but at other times, due to jewellery, or the use of silk or chiffon, you get terrible noise on the dialogue. The plug-ins available now to dialogue editors and re-recording mixers to remove noise are incredibly advanced, but clothing rustle is notoriously difficult to remove in post-production, because you end up also removing some of the frequencies within the voice. This may be acceptable to a minor degree when dealing with dialogue, but it would become very obvious in singing. We knew we couldn't do that. My proposal to Tom was to take a portion of the money in the budget allocated for ADR, transfer it to the visual effects department, and ask them to remove the microphones using CGI. This allowed us to put the microphones on the outside of the costumes, so we wouldn't have any clothing rustle. Ten years ago, that idea would have been way too costly, but the price of doing visual effects has decreased year by year, so we could now use visual effects as our safety net, instead of ADR.The sound team had access to offcuts of all the fabrics used in the costumes.
"Putting the lavalier mics on the outside of the costumes meant that we could place them in a primary position, on the solar plexus, ie. the centre of the chest, which also resulted in better sound quality. The DPA microphones come with a small plastic mount that you use to stick them on the costume, and that also acts as a concealer. I worked very closely with the costume designer, Paco Delgado, who gave me small offcuts of every single piece of costume material that he was using, and my team glued these to the plastic DPA mounts. So although the microphones were on the outside of the costumes, and you could see them with the naked eye, they were disguised to some degree. You didn't see them on many of the wide camera shots, and with the close-up shots the lavalier mics generally were out of the frame, so the microphones only needed digitally removing on the mid-shots. As a result, it was not nearly as challenging as we originally thought it would be. Normally, when you work on a movie, after the picture cut the director sits down with the sound post-production team for an ADR spotting session, to decide which lines need ADR due to poor sound quality. Instead, in the case of Les Mis, the visual effects department had microphone spotting sessions to decide in which images the microphones needed removing and in which they didn't. So the priority shifted completely, and gave a unique strength to the production sound department.
"In some cases, we put more than one microphone on an actor. If an actor had a severe head turn in the middle of a song, we would put one radio microphone on the left side of the chest and another radio microphone on the right side, so we had a lavalier in place for each way they would turn their head, and the dialogue editor and the music department would have this actor's vocals on two tracks. In other cases, particularly with Hugh and Russell, we'd have one lavalier mic a little closer to the mouth and one slightly further away, where the abdominal section meets the chest, because they had such huge dynamic ranges in their voices and went from very quiet, whispered sections of the songs to being extremely expressive and hitting huge SPLs. In that way we had both the quiet and loud sections covered. I should add that we never recorded one scene where we did not use the boom microphones as well. Whenever we could get them in a close-up position, they became a very good choice for the dialogue editor and the music department. Generally on a movie you will have just one boom operator, though I like to use two, and on Les Mis I felt that the best way of avoiding ADR was to have three boom operators. We had two boom operators who were using the Schoeps Super-CMITs, and their job was to pick up all of the solo vocals, and the third boom operator was using a Neumann RSM191 stereo microphone to pick up the choruses. All of the chorus members were on lavalier microphones, but we also wanted to have a stereo track which was a little further away to add width and texture to the lavalier tracks on the chorus, should the music department want to have that choice.”
Don't Shoot The Pianist
The lead actors were given complete freedom in the way they sang, apart from in one aspect: pitch. They needed support to stay in tune, and a bit of help with the song structures, so they weren't totally singing into thin air. But any monitoring fed to the singers needed to be inaudible on the lavalier and boom mics and invisible on the cameras. There was also the issue of how to keep everyone in sync in scenes with multiple singers in multiple locations, and large choruses. "Impossible!” said many, but Gerard McCann and Simon Hayes once again came up with the answers.
McCann: "The rehearsals in February and March 2012 had been done with piano accompaniment — incidentally, that rehearsal period was very important because the entire cast came together as a musical theatre ensemble, and built up a sense of camaraderie that you normally don't get in a feature film where the actors get called in just for their own scenes. For the actual shoot, which lasted 70 days from March to June, we could of course not supply the cast with a full orchestral track, or any pre-recorded track for that matter, because that would have locked the actors into something that was pre-set. Instead, we had a pianist playing live on set — mainly Roger Davison or Jennifer Whyte, who are both experienced West End Musical Directors in their own right, and who know the music for Les Misérables inside out. They were under strict instructions from Tom to always follow the actors and never push them with their playing. What was important for the actors was to hear the pitch, and anything else was icing on the cake. The pianist was effectively the other half of the drama, and even though the pianists were never leading, the actors' performances came out of the musical interaction between the pianist and the actor.
"The fact that the two protagonists, the singer and the pianist, couldn't see each other directly was another obstacle that we had to overcome. To resolve this, the pianist had video and audio feeds of what the actors were doing. This was a big challenge, because watching someone on a screen and hearing them through headphones is an entirely different thing to being in the same space together. The actors would hear the piano via very small earpieces that were on an induction loop on the set, and the frequency and dynamic ranges of the induction loops system are very limited. This meant that the pianists had to play in the octaves that registered best via the ear pieces, and avoid playing low notes unless absolutely necessary. From the actors' point of view, while it is hard enough to be singing with an orchestra or piano in the room, in this case all they had were these tiny earpieces with sub-standard sound, and they had to pitch their performances against that.”
Hayes: "The induction loops and earwigs were a means to an end. The actors were incredibly understanding and did an absolutely monumental job dealing with these earwigs, because they don't sound good. It's old technology, traditionally put in theatres and cinemas and banks and post offices so people with hearing aids can hear in those locations. You have an indication-loop amplifier and play the music through that, and it sends it through a cable and this creates a magnetic field that the earwigs can pick up. We bought the best, most up-to-date products to get the best possible sound quality from them, but they have a limited frequency response, and I very much hope that better technology will become available for future projects. Induction looping also is a black art: you have to work out through trial and error what will give you the best results. We'd always arrive early at a new location to get the best possible loop. When we shot the scenes in Greenwich for the start of the uprising and the parade, we had an induction loop that was a kilometre long! So we pushed induction-loop technology further than ever before, insofar as we know.
"We had to use induction loops because it is notoriously difficult to get many radio frequencies to work together without them interweaving, and we had 26 frequencies for radio mics, plus five channels of talkback systems for the directors and camera operators and for the director and pianist and dialogue coaches to hear the singing, plus two radio booms which we occasionally used. That meant 33 radio channels! Moreover, painting out something on images of human flesh is very intricate and expensive. So we had the earwigs custom-fitted to the actors, and they were so small that you'd have to look directly into someone's ear to see them. But it did mean that their bass response was very limited, because of the tiny driver, and this meant that we had to EQ the piano to the limited mid-range bandwidth that worked best for the earwigs. Luckily this didn't matter when recording the piano tracks, because they were thrown away afterwards, the irony being that those piano tracks were hugely responsible for the performances. Over the course of shooting the film, the piano players developed a hugely personal relationship with the actors because of the very intimate way in which they were being creative with each other.”
McCann: "You have to imagine the actors having to sing with this tinny piano sound in their ears, and there'd be maybe 200 people watching them and they'd have a camera practically up their nostrils! That's hard enough on your own, but with scenes in cafés or barricades, with various people moving around and singing off each other, or singing choruses, it's very, very difficult. When recording large crowds we couldn't have 500 earpieces and lapel microphones, but we could accommodate maybe 80 or 90 earpieces. So we would spot the strong singers, and give them earpieces and spread them around and then the others would sing along with them. Simon also had three boom operators recording at all times to complement the multiple radio mics.
"This left the problem of what to do with the ensemble pieces and more rhythmic songs, where there isn't really the flexibility to be pushing and pulling the tempo. I initially argued that we'd have to pre-record a 'bare bones' musical background that we could play back on the set, as we did have some pre-recorded backing tracks for a few songs, taken from rehearsals. But Tom again felt that there was something about having everyone on the set in the moment deciding what the tempo should be. In those cases, we'd shoot a number of takes and once Tom was happy with one, with a clear tempo, we would use that recording as playback for the remainder of that sequence. The pianist would still play along live, and if the actors wanted to go faster, we'd instantly switch from playback to live piano. Because the sound of the piano was identical, the actors never knew whether they were listening to live or playback. In songs with actors in different locations, like 'One Day More', we would establish a guide tempo during rehearsal, and in this case Russell Crowe, who was in the first scene, and live pianist Roger Davison finessed that tempo on set. This gave Hugh [Jackman], who was in the next sequence, a range of matching tempi he could work within, and from there we built a template for the entire song, and recorded the piano backing for it. This completely pre-recorded piano track was then used as a playback for shooting all the other cross-cutting scenes in this song.”
Pro Tools Rigs Are Bonny
Director Tom Hooper was insistent on shooting most scenes with both close-up and distant cameras, meaning that boom mics could not be placed close to the actors.Photo: Laurie Sparham / Universal Pictures
"To be able to play back pre-recorded material and tracks from a previous take, we had three Pro Tools systems on the set. The first system, which was driven by my Music Editorial team colleague Rob Houston, was our HD desktop 'recording rig'. It recorded each new take live on external drives, consisting of Simon's mono vocal mix, live piano audio and live piano MIDI, plus it handled direct feeds of a playback of a guide piano and vocal take from a previously recorded performer, where appropriate (for example in 'One Day More'). When required, these direct feeds and clicks came from Pro Tools rig two, which was driven by Rob or myself. Rig two was our laptop 'playback rig', which ran via a Soundcraft Spirit F1 14:2 mixing desk, and it had external drives that were pre-loaded with all rehearsal vocal/piano recordings for reference. As the two systems were networked together, we could directly transfer the files from each newly recorded take from the HD rig across to rig two, so they could be edited, positioned and be ready for immediate playback if required. Pro Tools rig three was our M Box Pro laptop 'prep rig'. We used this for all other tasks. The two other rigs had to be permanently 'online' for the shoot, with each shoot day being a continuous process of setup, rehearsal, shoot, move location, rehearsal, shoot, so I would generally use this third rig to bring material in that I'd asked John Warhurst, my other Music Editorial team colleague, to send over, or to listen more closely to some element of a live take that I wanted to check, or to prepare tempo maps of chorus numbers, et cetera. Rig three was my personal rig. Generally, as the cameras rolled, I would be following the vocal score, listening to the live vocal and piano, while watching direct feeds from the film cameras via video monitors, making take notes and working with the pianists to help guide the tempo, and discussing any music issues with the Music Direction and Production team members present.
"In terms of technical recording quality, or musical and editorial issues — tempo matching between takes, vocal separation with multiple singers, signal path problems such as from a dirty mains feed affecting the piano quality — that was a full-time discussion between Simon's team and mine, where our combined expertise could straddle all these different issues. So the two of us would frequently go on set to discuss with Tom anything we thought we could improve, regarding adjustments to equipment (mics, earpieces and so on) or notes for the actors (such as asking ensemble members to limit movement noise under a soloist's vocal performance). Dramatic performance issues were very much the domain of the director and actors, as they would be on spoken-word drama, although we had the Music Producer, Musical Director and composer on the set much of the time, listening to the live vocal mix on headphones, so they could express any comments they had from their perspective.”
Climbing Mount Impossible
Clearly, a lot of audio technology was needed to make all of these impossible goals possible. What's more, all this gear needed to be lugged around locations and sound stages in southern England, where most of the shooting took place, almost on a daily basis. Oh, and there was a French mountain that needed to be climbed...
McCann: "There's one scene with Hugh Jackman on a French mountain at three o'clock in the morning, with him looking cold and bedraggled and you see his fluming breath, and the best way to get that was to actually shoot it on a French mountain at 3am, even though it was a challenge for the actor and the crew who took Simon's Deva hard disk audio recorders up that mountain!The camera and sound teams attempt to keep up with Hugh Jackman on a freezing French mountain.Photo: Laurie Sparham / Universal Pictures But the rest of the on-location recordings were almost all done in a 50-mile radius in and around London, and so for each new location we'd spend two hours preparing and setting up, and unhook everything and wrap it up at the end of the day. We discussed very early on whether we would use something like the Abbey Road Mobile to record everything, and have more of a fully fledged studio approach, with all the mics coming into a big studio-style console, but we realised that there was no way that this would work with the sort of locations that we were working at and the speed with which things would be shot.
"Instead, we initially had our gear outside the sets in a tent, which was alternately freezing cold or sweltering hot, depending on the location and time of day or night, and was also slow to set up. So we then decided to hire a Luton van for the music team, to match the flexibility of Simon's sound van and the van of the video playback unit. All three vans could be cabled together easily. We had all the Pro Tools systems, and the video monitors and the piano set up in our Luton van, so all we needed to do was plug into Simon's pre-existing recording setup. This saved us an hour a day, and when you're doing 15 or 16-hour shifts, that hour is your extra hour of sleep! We affectionately named our van the 'Shabby Road Mobile'. Our laptop systems were often pushed to the limit because it got too cold, or too hot, or because they were knocked around when we were moving from one setup to another. But having three Pro Tools systems meant that we could always devise a quick workaround if one of them was temporarily out of action. We also needed to make sure that we had power supply backups, like UPSs and so on. If someone plugged in a hairdryer in a trailer some distance away, causing a buzz on the circuit, we wanted to be able to immediately switch to another circuit.”
Hayes: "My primary signal paths for recording the vocals were the DPA lavalier mics, going into the Lectrosonics radio transmitter and received on the Lectrosonics receivers, and the boom mics, all going into a pair of Audio Developments AD149 mixers, which, in my opinion, is the very best film-industry mixer on the market today, and then going into two Zaxcom Deva recorders. The Deva 16 gave me 16 tracks, and the Deva 5 had 10 tracks, so I had 26 recording tracks available to me at all times. I recorded at 48kHz/24-bit. Because we used the lavalier mics, there was no need to go to 96KHz, but it was very important for us to be able to use the full dynamic range of 24 bits, since we were asked by the music department not to ride the gain or use compression, limiting or EQ. If you ride the gain to make quieter passages louder, you're affecting a performance that you're there to protect. Similarly, when he hits full SPL and you compress. So I turned off the compressor/limiter on the mixer and the Devas, and to prevent the automatic limiters kicking in, I lowered the gain of the Lectrosonics so they could handle the loudest possible SPL you can get from a human voice. The transmission system was actually completely quiet and we didn't have issues when recording at low levels. We also didn't want to have acres and acres of cable, so although I sent a live mono mix of the vocals and piano of every take to the director, the picture editor, Gerard, Rob and the pianist, I afterwards would quickly do a download of the entire recording to an SD card, and I gave that to Gerard or Rob Houston, and they'd load it into their Pro Tools systems.”
Challenging as it was, the process of capturing acceptable recordings of great vocal performances was only the first stage in realising Tom Hooper's vision for the film. These vocals then needed to be edited, before orchestration could be overdubbed to them: no mean feat when you are dealing with 20 or more takes for each scene, wildly varying tempos, and endlessly changing picture edits. In next month's SOS, Gerard McCann and Simon Hayes, along with Supervising Sound and Music Editor John Warhurst, Music Editors James Bellamy and Rael Jones, and music recording engineer and mixers Jonathan Allen and Andy Nelson take up the story.
Go to Part 2 .
About Les Misérables
For the few remaining uninitiated, Victor Hugo first published his classic novel Les Misérables in 1862. It traces the criminal-to-saint transformation of Jean Valjean, who, while being chased by the pitiless policeman Javert, changes and saves the lives of several people. There are political overtones as Valjean inadvertently finds himself near the barricades of the 1832 Paris June rebellion. The novel has been adapted for film, radio and stage, and the 1980 musical, with French lyrics by Alain Boublil and music by composer Claude-Michel Schönberg, is perhaps the most famous adaptation of all. An English-language version opened in London in 1985 and is still running 28 years later, making it the second-longest running stage musical in the world. The musical also ran for 18 years on Broadway and is set to return there in 2014.
The movie version of the musical had been stuck in 'development hell' since 1988, from which it emerged in 2011 when production began. Les Misérables, the movie, finally premièred in London on December 5, 2012, featuring well-known actors such as Hugh Jackman as Valjean, Russell Crowe as Javert, Anne Hathaway as Fantine (a factory worker who falls on hard times), and Amanda Seyfried as her daughter Cosette (who is rescued and raised by Valjean).
The Silent Set
Gerard McCann: "Simon [Hayes]'s job was to make pristine recordings of what the actors were doing on the sets, and the director and I discussed anything to do with musical interpretation and the choices that would enhance our editing options during post-production. But because Simon's and my jobs were so intertwined, he and I would both police the sets to keep them as quiet as possible. This was a real challenge! We had children, cows, chickens, reindeer and everything else that can be difficult to control on the sets, as well as, in some scenes, 40 people loading muskets and bumping into light stands, plus there were often six cameras rolling at the same time. Everyone was making supreme efforts to minimise noises, including the costume department, the special-effects department, the art department, the camera department, and so on, all to make sure that what we captured could be the final vocal that would sound acceptable when hugely amplified in iMax. It's easy to say, 'OK, this is as good as we're going to get,' but Simon and I were constantly challenging each other to go the extra length and not to slacken in any way in the pursuit of the very best recorded tracks we could achieve. He normally captains his ship, and I captain mine, and I really enjoyed that we supported each other on the set of this film.”
Hayes: "The need to keep the set quiet had numerous consequences that affected everyone. We built a soundproofed plywood box for the pianist that we used on these soundstage sets. Although the instrument was a Yamaha CP50 electric piano, with everything else being quiet you could still hear the weighted keys thump. When we were shooting on location, the pianist was in the Luton van in which the music department carried its gear, so that wasn't an issue. We were shooting many of the exterior scenes on sound stages so that we could control the outside background noises, but it's no good shooting an exterior scene inside if the actor's hair isn't moving, so in those cases we needed a wind machine. This has ruined many performances on a film set, because normally if you use a wind machine you're committing the entire performance to ADR. So the special-effects department placed the wind machines far outside the sound studio and rigged hundreds of metres of flexible air-conditioning tubes, and stuck them through the walls of the sound stage. A technician would then hold a tube and point it at the actor. All we had to deal with in this case was the noise of air moving, which sounds like real wind, and is above the human voice, so it can be removed using plug-ins in post-production, if necessary.
"There was also the scene with Samantha Barks singing the song 'On My Own' in the rain, walking down Rue Plumet.One of the most challenging scenes for the sound team was the sequence in which Samantha Barks sings 'On My Own' in the rain. This was a massive collaboration between the sound department and every other department on the set. We asked the special-effects department to make the rain droplets as small as possible to keep the impact noise down, and they got it to a fine mist. At this point the cinematographer, Danny Cohen, said that it didn't look like rain any more on camera, so they made the droplets a bit larger again. We knew then that we had the quietest rain possible. We also got a lorryload of horsehair (the Americans call it hog hair), which is synthetic material used to soundproof cars and buses that comes in sheets of 4x8 feet and four inches thick. We placed that on every single surface that the rain was hitting that wasn't in camera. We put lavalier mics on Samantha, but they kept getting waterlogged. We used new ones for each take, but even so, the post-production team decided that the boom mics sounded much better. But the rain was hitting the top of the boom mic, so we got a one-metre-square frame, put horse hair on top, and rigged that on a second boom pole, and that was held just above the boom mic by the second boom operator. Luckily, Tom used close-up shots for this scene, so we could get the boom mics close enough. The other problem was that the camera was wrapped in a polythene bag and the camera operators had their rain gear on, and you could hear the rain falling on these. We have these rolls of black fabric on film sets, called Bolton fabric, or Duvateen, and we cut big pieces of that and covered the camera and the camera crews with it!
"You can see the level of detail that we went into, including this huge and unprecedented degree of collaboration between departments, to give Tom and the actors the guarantee that these live recordings were going to be usable. The set designers tried to make the sets themselves as quiet as possible, but when Eve Stewart, the production designer, asked me what she could do in this respect, I said to her that it wasn't only a matter of making the sets quieter, because that's not always possible, but also of making sure that stuff that's on camera is real. As a result, the cobbles in Rue Plumet are real, so when people were walking on them or rain was falling on them, it would sound like footsteps or rain on cobbles, and not on plaster. The same, for example, if an oak door was closed by an actor on camera. Rather than use flimsy wood and paint it to look like oak, as you'd normally do, she made it from real oak, so if a door shut behind a line of singing, it would sound like a real door, and we might be able to use the take with that sound in it. In general, we tried to avoid recording incidental sounds, like footsteps, rustling paper, and so on, because we never wanted to risk that these sounds would ruin the vocal recording. We would only record Foley if it was in the shot. So it was the job of one member of the sound department to put carpet down to muffle every single footstep that wasn't in the shot, knowing that Tom would be able to put these footsteps, or any other sound, back in later using the traditional Foley overdubbing method, with complete control over the volume at which he wanted it.”