The first marathon of the London Olympics took place not on the track, but in the recording studio. Jake Jackson was the man charged with recording 201 national anthems...
We had a sweepstake with the musicians,” smiles Jake Jackson, "and we got to 974 takes in the end.” It's tempting to wonder whether this represents a record feat of studio endurance on the behalf of Jackson, composer and arranger Philip Shepherd and the musicians of the London Philharmonic, but the figure seems less surprising when you think about the sheer number of pieces they had to tackle over the course of six days in Abbey Road Studio Two.
"We had 51 hours of recording,” continues Jackson, "to record 201 anthems. Factoring in breaks, that's about 12 minutes per anthem. They're all about a minute long, and they're all generally pretty hardcore, because they're each country going 'We're the best! Come and have a go!' so it was pretty full-on for the musicians.”
The anthems in question represent the 207 countries participating in the 2012 Olympic Games in London, where they will be played when athletes are awarded gold, silver and bronze medals. "Every country will get a performance of their anthem. There's a flag ceremony when they arrive at the Olympic village, so they're all going to get played, but some will get played more than others. And even Burundi might get a medal! We actually recorded 201, because a few of them are shared, like, for example, Greece and Cyprus, and Hong Kong and China.”
Jackson describes each anthem as being "about a minute long”, but getting them all to the stipulated duration required quite a bit of creativity from him and Shepherd. "They had to be between 60 seconds and 90 seconds in length. That's how long the TV wants it, to be able to show the man putting up the flag. But some anthems are 20 seconds long, and some are four minutes and 18 verses, so Philip had to go through and make them all fit, and that had to get approved. When we were recording, it was like 'This is 52 seconds at this tempo. Let's slow it down, or do a rallentando at the end, or let's hold the last note for five seconds!' Sometimes something doesn't feel right at the wrong tempo, so we had a few little tricks to make it longer at the mix —- I had a long reverb that we used to bring in over the final bar, which would ring and ring.”
Massaging the length of each anthem was only one aspect of a planning process that had to be meticulous. There was, for one thing, the question of how best to record material that would eventually be played back both on televisions in millions of homes, and over PA systems in stadiums, velodromes and swimming pools. "I wanted to know how to make them sound, whether they should be wet or dry,” says Jackson. I managed to get hold of a guy from BBC Sport and had a quick chat on the phone, and he said 'Basically, every mixer does it differently. They'll have the dry signal from the CD player and they'll have the ambience microphones from the individual venue, and they'll make the decision.' I was like 'Well, that doesn't help me very much. I need to know how to make it sound, because if it's going to be played over a PA in a swimming pool it's going to be as washy as hell.'
"Me and Phil discussed what the best thing to do was, and knowing that it was all going to get done differently, we just thought we'd make it sound as good as possible, not too much reverb but enough to make it sound good. We ended up going to Abbey Road Studio Two, which is a great ambience that isn't wet, and it works really well for adding reverb to.
"We recorded at Abbey Road Studio Two and mixed here at AIR Studio Three. On the first day of mixing, the hall was quiet downstairs, so we took a rough mix and played it out through the SLS speakers into the massive hall, took the moving ceiling and put it right up to make it as ambient as possible, just to see what it sounded like. We played it loud in the hall and based our amount of reverb on sounding good in here and in that environment. We might have taken the reverb time down by point two of a second or something, to make it sound good in that room.”
Although this was a prestigious project, the sheer amount of music involved meant that using a full symphony orchestra was out of the question, as was any attempt to vary the instrumentation to reflect musical traditions within the participating countries. Instead, it was decided to use a small ensemble consisting of strings, brass, woodwind and percussion. "We had five first violins, four second violins, four violas, three celli, two basses, double woodwinds — two flutes, one of which played piccolo sometimes, two oboes, two clarinets, two bassoons — two French horns, two trumpets, two trombones, a tuba and three percussionists, which was a timpani player, a bass drum player, and a suspended cymbal player who played a snare drum occasionally. There was vibes in one of the 'cool' anthems, and a glockenspiel in one.”
Unusually, the string players were separated so that the first violins were on the left and the seconds on the opposite sides. "We had them left and right to make it not so heavy with the violins being on the left side.”
To compensate for any sonic imbalances arising from the size of the ensemble, and to cover all eventualities, Jackson used an extensive setup incorporating both close and ambient mics. "I gave myself a lot of microphones, more than I usually would if I were recording a film score, because we were recording 200 pieces, and I was never sure where the melody was going to be and we needed to be able to pull it out. We also had to be able to falsify the balance a little bit, because the brass were overpowering the strings.
"The great thing about doing it at Abbey Road was that the microphone selection is amazing. We had a Decca Tree of [Neumann] M50s and then some outriggers of [AKG] C12s, so that was a kind of traditional orchestral recording arrangement. We set up essentially so it could be mixed 5.1 if they wanted to in the future, and we also recorded at 96k, so that was future-proofing a little bit.
"The M50s were about 10 feet above the conductor's head, and the C12s were set to halfway between cardioid and omni — a kind of wide cardioid pickup. They were between the first and second desks of the violins. I had them a little lower than I normally would, because I used them as a kind of overall violin pickup. There weren't that many violins, and that was where a lot of the melodies were, so I wanted to make sure that I could hear them on the room mics as much as possible. Then we had some ambient mics, which were Schoeps 2Ss, which were pretty high up, about two-thirds of the height of the room. They were just there to pick up more of an ambient sound.
"The horns were on the left-hand side, the brass on the right. There was a Neumann U67 in front of them, and the woodwinds had a stereo pair of Schoeps MK4s as an overall pickup, and then the percussion had some MK21s, which are the wide-cardioid pickups. Then each individual violin desk had a microphone on it, and the flutes had a microphone between the two of them, the oboes had a microphone between the two of them, the trumpets shared a microphone, and so on.”
The position of each microphone was documented precisely, so that the setup could be reproduced for subsequent sessions. "We had to be able to recall it to the 'nth' degree, because if we had to spend 10 minutes at the start of the session checking that everything sounded OK, that was a whole anthem we couldn't record.”
With only 12 minutes' recording time available for each anthem, session preparation was also crucial. "Philip had roughed them out in MIDI, but all we used from it was the tempo information [to give the musicians a click track]. They all had two bars' click into bar one, basically. All the musicians were hearing the click — it had to be done that way because of the time involved — although sometimes we did the endings without a click if there was a rallentando. Philip did all the printing of parts himself, it was nuts. I think he used 30,000 sheets of paper!
"The first one we did, we spent a bit of time listening through to it and making decisions, and Phil came up and listened to it, but after that it became a real production line. Everything had to work: the Pro Tools couldn't fail, the music had to be in the right order on the sheets, everything was like clockwork. We got there two hours early every day, made sure all the microphones were working, all the headphones were working, all the MIDI maps were there and right, all this kind of stuff.
"One of the hardest jobs was being the Pro Tools operator, because of the pressure of keeping up with the take numbers. We discussed it with the tape-op at Abbey Road, Gordon Davidson. We tried to make sure that everything we did in preparing the template session saved us time — because if you add 10 seconds every time, then 201 times 10 seconds is quite a long time!
"Phil would turn up with the MIDI files in the morning or send them last thing at night and I would forward them to Gordon at Abbey Road. The first hour of the day was spent making the sessions up, and then we'd have 50 or 40 to record that day. We used three-letter codes, like 'GBR' for Britain, all the way through the process: so the score had that three-letter code, every audio file had that three-letter code, every mix had that three-letter code, so it was always identifiable by that. Then there was a take number, which started at one and went through to 974!
"We recorded 40 tracks of 96k audio, 24-bit, with a live mix as well, and all those tracks had to be named as well. We used Quickeys to go through and name all of those at the beginning, so if we lost an audio file or if Pro Tools crashed, we could still find it. So we had our three-letter code at the front, like 'GBR'; then we had a number that corresponded with the desk channel, generally, because it's easier for your brain, and then it'd be 'Tk.01', or 'Tk.356' or whatever. So, of those 40 tracks, each would have a different number, and we'd also say what it was, so it would say 'Tree Left' or 'Flute 1'.
"[Gordon] had a template which had all of those set up, so it would say 'XXX 25 Tree Left Tk.00' in the template, and then he'd make a new Pro Tools session and replace 'XXX' with 'GBR'. Then he'd do a Quickey macro that copied and pasted that to every channel. Then, with Pro Tools Playlists, you can delete '00' and put in the number of the previous take — so let's say we were working on take 700, he could put in take 699 and do a 'New Playlist' and it would change all of them to 700, which is a much quicker way of doing it than copying and pasting.
"As we opened each session up, Gordon would stick it in record so we could hear properly, renumber the takes, and by that time the musicians had got their music up and were ready to record. Then we'd record, see how long it was, discuss changing the tempo — 80 percent of them were the right tempo, because Phil had obviously checked it out — record it, see whether we needed to record it again, do any pick-ups, and generally we'd try to get a rough edit done. I'd say to Gordon 'Take six for the first half, take seven for the second half', he'd do it very roughly, then go back during a break and get the edits closer. If you were having to open up 40 sessions at the end of the day after nine hours' recording, it'd be nuts, so we'd try to do all that stuff while they were getting the next anthem up and discussing it.
"I ordered a map of the world beforehand, to mark off when we'd done the countries, so we could have a visual representation of how we were doing. I wrote the master take number on the dot. We had two colours: orange was for the countries that won gold medals at the previous Olympics, green for the others.”
With a first take of any given anthem in the bag, the immediate task was to check whether its duration fell within the 60- to 90-second window deemed acceptable. "If it was 50 seconds or 1 minute 40, then we had to come up with a solution, so the Pro Tools guy would be like 'OK, let's take 10bpm off, or 5bpm off, and get that to the right length.' The next time through, generally, they'd get it right.
"Then there would be a few tweaks, like 'Let's make the violins louder.' It was difficult, because there was a lot of production to be done, but I'd never heard most of the anthems before. There's about 10 or 15 that you can name straight away, but apart from that, you're having to make a judgement call on something you don't know. Phil left all that up to me — he'd done all the arrangements, so he knew what he wanted, but he left me to make it balanced.
"We ended up often adding octave violins to make the tune stick out a bit, because the forces of the orchestration we had didn't give us a huge number of violins. The second time through the verse, maybe we'd have them go up an octave, that kind of thing, just to make it stand out a bit and make it interesting.
"It was hard work. Just listening to that amount of music was hard — because it was loud, and just because of the pressure of time. I would be there saying 'No, we can't do another take, let's move along,' or 'Let's just do those four bars again because there was a funky bit of intonation,' or whatever. Phil left that totally in my hands, to be judge of takes. I had a score that I wrote in and scribbled rough edits in as we went along.”
In the end, Jackson's efficiency and forward planning paid off, to the extent that the team were able to spend some time embellishing the anthems that were likely to be heard most often. "We went through a list of countries that won golds at the last Olympic Games, and spent more time on the gold-winning countries, because obviously they're going to be heard more often. We had one session left where we'd got through ahead of schedule, so we went through in gold medal order from Beijing, going 'Right, let's do the overdubs for China, and the USA, and Britain...' working our way down until we ran out of time. We went through and overdubbed a string pass to make the strings bigger and a bit more lush.”
With almost four hours of music recorded, it was equally important to come up with an efficient working approach at the mix. "We had about the same amount of mix time as record time. We did a day's mixing in the middle of the recording, to make doubly sure that we were happy: we mixed some of the important ones to make sure they sounded good, and then took them back to Abbey Road to double-check how they sounded on the speakers in there. The B&W speakers in the room sound great, but they're often a bit too flattering. I wanted to have some small speakers we knew that we could reference, so we took the Genelec 8240s down there, so that we had the same reference speakers all the way through.
"We did the mix on the desk [the AMS DFC Gemini in AIR Studio 3]. It was quicker to have fewer things to open up in Pro Tools — I didn't have to import a balance into the Pro Tools session or worry about automation nodes getting carried over, stuff like that.”
Rather than open each Pro Tools session in turn, reconfigure it to work with the I/O arrangements at AIR and then mix it, Jackson found it easier simply to import the recorded audio into a new session. "You have to import one way or the other — either the audio onto your playback lanes or you import the mix data into the recording session. You've got a recording session and a mix session, and you either bring the recording data into the mix session or the mix data into the recording session. The record session was done at Abbey Road, so it was not configured for my outputs and reverbs.
"Basically, I had a session that I had open all the time, so after I'd worked on Great Britain, I moved onto France. I finished Great Britain, did a 'Save As' and called the new session 'France', removed the audio for the British session, imported the French audio... you match tracks and bring only the audio in, no automatio, just Playlists and audio in the Import Session Data dialogue. So all you're doing is essentially replacing the multitracks. It's like taking a tape off the tape machine and putting another one on.”
The actual mix was carried out using the desk. "I turned the automation on and used rides, which was nice. There weren't that many rides. It was mainly just changing the balance, or sometimes we needed to hear more of the violins, or the ambience. I subgrouped things on faders, so if I wanted more ambience or more strings I could just turn them up.
"I had an analogue compressor and EQ across the mix — a Neve 33609 and a Manley Massive Passive — which basically stayed the same each time, except for the output gain. We didn't limit everything, we wanted some dynamics. It's got to feel real, so I used gentle 2:1 compression, and then I changed the output gain. The threshold was only +6 on the 33609, so I was only taking 3-4dB off it, not much really. On television, it'll get compressed to buggery anyway.”
The only effect that was used was reverb, and this was applied by setting up aux sends on the desk, which were routed back into Pro Tools and instances of Audio Ease's Altiverb. "I used two different Altiverbs. I find it sounds more natural if you use two reverbs. It doesn't sound like a reverb so much. I think I probably used 'Large Hall 1' and 'Large Hall 2', changing the length of the IR sometimes. It doesn't make it sound like one reverb, it sounds like a natural room. Then we had one very long reverb that I'd use for making tails sound longer, which I'd just sneak in on the last five seconds. Most of the time I didn't use it at all, but for the 10 percent where they needed to be stretched, we did that.”
Not only was mixing a lengthy process, but the logistics of delivering 201 pieces of music also required considerable thought, especially as numerous different formats were required. "We delivered MP3s, we delivered 44.1, 16-bit stereo interleaved WAV files, 48k, 24-bit stereo interleaved WAV files, 48k 24-bit L-R split WAV files, and 96k 24-bit L-R files, and they had to get labelled as well.
"I'd do a bunch of mixes, like 20 or 30, and then bring them into another Pro Tools session. They were always done as 96k left and right split, and I'd grab 30 of them and export them as WAVs, MP3s, and whatever, and Phil would go through and label them all. That would take forever. There was a lot of data. At the end of that, we then went through and double-checked everything. By ordering them by file size, you could check that they were the right ones. It became more of a maths problem than a musical thing!”
In a world where revenues are shrinking and studios closing, audio engineers need to be alive to new opportunities, and Jake Jackson's Mix At Yours service is an excellent example of how a skilled engineer can shape a role for himself, despite composers' increasing reliance on sampled instruments and home studios.
"There's fewer Hollywood movies being recorded over here, so I do quite a lot of mixing for TV. I've got a service called Mix At Yours, where I go to a composer's home studio, which keeps me really busy. If you're doing a multi-episodic TV show, there's often budget to pay for someone to get involved, but not budget for a studio.”
In this situation, Jackson's role is completely different from that of the pop/rock mix engineer. "It's almost a production service — a second set of ears — to go through and go 'Are you sure about that? Let's take this out.' Some people come to me halfway through the composing process, if they're not happy with a piece of music, they know it's almost there but not quite, or they don't know what's wrong with it. Even if it's not a finished piece of work, it's not necessarily about finishing it, it's about making it sound better. It means they can concentrate purely on the writing.
"I'm not starting again with any of the mixes. I go in and pick up where the composer has left off. They'll generally be working in my template, so they'll have all my reverbs and all my delays. I'll go through and make each sound sound better, and balance it. A lot of these things would get delivered without my input anyway. So the point is not to go and be too diva-ish about it, and go 'I can't work in this space, there's too many reflections, I haven't got a comfy seat.' It's about trying to build up a relationship with the composer and get it to a slightly better standard. I can often do pretty much a whole episode's worth of music in a day, working really fast.
"You've got to know what they're aiming for, and that's not done by deleting all their EQs, pulling down all the faders and starting again, like you would do if you were doing a pop mix. They're close with it, but the melody will be too loud or too soft, or the balance won't be quite right between the two things that make up the melody, and I can go through and add a bit of reverb or set it a bit differently, and it's actually aided by the fact that you're basically taking their working mix and finishing it off. I know that if I took it to a studio I could get it better, but it's all about timescale and cost, and it's all going to be beneath dialogue anyway. It's got to sound as good as it can for that purpose, and I've realised there is a market for that, and people find it very helpful.
"Dubbing mixers are used to composers doing it themselves, and they'll compress it too much or won't EQ it right and it won't fit round the dialogue right. I did something for Dru Masters for the new series of [BBC One's] Silk at his studio, and he sent me a text saying that he was thrilled, because the dubbing mixer didn't need to use the stems, or change the compression or EQ.”