by Maxwell Steer © 1995
When Noise Annoys
At the top of the article I said that dubbing is the art of the possible. It stands as a crucial link in the chain connecting the actor in front of the camera with the cinema audience. For many years the weakest link in that chain was the sound system of the average cinema auditorium. This meant that in practice 'the possible' was severely restricted, art or no art.
Various experiments had been made with multi-channel sound from the 1930s onwards, most famously Disney's Fantasia in 1940, recorded in 9-track omni-directional 'FantaSound'. Problem was, only about 4 US theatres bothered to install it. A three-track system by Warner's, VitaSound, fared little better. One reason was that US anti-trust legislation had forced the studios to divest themselves of their cinemas and the new owners felt under no pressure to innovate or raise standards. Ironically the only studio ever to dedicate itself to recording true stereo was Russia's Mosfilm in the 1950s - tho history doesn't relate how it played in Outer Mongolia!
In the West Cinerama tackled the problem of sound head on in the early 50s by recording a 360 degree soundtrack on six channel magnetic. Its successor, CinemaScope provided The Robe with four track surround-sound on magnetic strips between the picture and the sprocket holes, whose resulting modification acquired the nickname 'Fox holes'. Mike Todd copied the format for Around The World in 80 Days when developing his 70mm Todd AO which was the first to offer six track magnetic sound, configured as five speakers behind the screen and one at the rear of the auditorium. But such 'hard ticket' spectaculars required dedicated installations and made little impact on the quality of sound in the vast majority of smaller theatres.
The major breakthrough in this area came in the early '70s when Ray Dolby, a US physicist working in London, adapted a noise reduction technology developed to for multi-track recorders to cinema sound. Dolby, who has shown himself to have a genius for 'the art of the possible', saw that while merely improving the quality of sound on film would accomplish little if it was not replicated in the audio installation of the average theatre - equally theatre owners, who had been bitten by the collapse of Cinerama and CinemaScope, were not going to improve the quality of their audio if there was insufficient product to justify the level of reinvestment it demanded.
Dolby at first adopted the standard approach to film stereo by splitting either side of the optical track into a separate channel. However his masterstroke was to employ the QS quadraphonic matrix-encoding system (which was then enjoying a brief popularity during the era of quadraphonic vinyl records) to create prints that remained compatible with mono, but could also be decoded to generate Left, Centre, Right & Surround channels, with an optional sub-woofer.
The first film to be encoded with Dolby A was Lisztomania in 1975 but the system really came into its own with Star Wars and Close Encounters in 1976. Based on this success Dolby Laboratories moved quickly to encourage exhibitors to standardise cinema acoustics, and to date have installed systems in over 21Ê000 cinemas world-wide. A further innovation followed in 1986 when Dolby introduced Dolby SR (spectral recording), a backwards-compatible development offering more than twice the noise reduction. The specification of Dolby SR on 70mm prints allows six channels encoded Left, Left-Subwoofer, Centre, Right, Right-Subwoofer, Surround. This is now complemented by the digital Dolby SR¥D where in addition to optical tracks a digital sound track is encoded between the film perforations.
Standard practice for encoding Dolby mag masters is to create a separate four track master for Dialog, Effects and Music, each of which is encoded Left, Centre, Right, Surround.
Dolby's effect on dubbing has been immense. A dynamic range of 28dB in monophonic films rose to 35dB in stereo. Dolby SR (analog) is nowadays quite happy with 46dB, but SR¥D can handle a 60dB dynamic range and allows for 98dB s/n with a theatre spec demands that each channel be capable of peaks of at least 105dBc SPL. Similarly, audio bandwidth that was once limited to around 9kHz has now been doubled to 18kHz.
Hugh Strain adds a cautionary note here: "you have to bear in mind that you can never use a full dynamic range because nowadays you've got more ambient noise in the cinema from air-conditioning and other factors. When Dolby SR first came in, some people used the range too freely and there were complaints about dialog becoming inaudible. However clever you are with your Music and Effects it's no good if people can't follow the dialog."
Compression is absolutely taboo for all film dubbers. However the only negative effect of this dynamic range of some recent films is that the standard compression used on television transmission is at times quite audible on domestic television.
Dolby maintain a tight hold on compatibility by insisting on a rigorous specification for dubbing studios before authorising the use of their DS4 encoding equipment, and by vetting all mixed masters, for which producers pay a fee of £2900. There are currently over 130 authorised Dolby studios throughout the world of which eleven are in the UK. Dolby prides itself that the last 17 conescutive Oscars for 'Best Achievement in Sound' have been won by SR encoded films.
It should be said that there are two rival (and, needless to say, incompatible) digital encoding systems: Sony's eight channel SDDS, using 5 front speakers, where the digits are encoded in between the perforations on both sides of film using ATRAC compression developed for Sony's MiniDisc: and a system which originated in France but is now owned by Universal called DTS, where 6 track sound (configured similarly to Dolby) is recorded on a CD-ROM slaved to the film by time-code.
Both systems retain an analog optical track, the omission of which in Kodak's 1990 all-digital Datakode system led to its rapid demise when exhibitors were not prepared to invest in an entirely new technology.
To explore state of the art dubbing technology let's look into the latest Bond movie which, as I write, is dubbing at Pinewood.
The Pinewood post-production facilities includes two dubbing studios each with automated analog 60i/32o desks, one a SSL5000M, the other designed by Sam Wise of Technical Projects and the Pinewood team. There are 17 playoff machines, including six 6 track and four 4 track. Normal procedure for a James Bond production is to work in both theatres simultaneously, with a team of five led by Graham Hartstone, Head of Post-Production Sound. Given Goldeneye's two hour running time the dubbing is set to last 5 weeks and will cost in excess of £100Ê000. The mastering requirement is for both Dolby SR¥D and DTS timecode.
The sound cutting has been supervised by Jim Shields. There are three specialist editors, each with an assistant, Peter Musgrave covering Dialog and (Auto Dialog Replacement) ADR, Bob Risk, on Effects/Foley and Bob Hathaway, Music editor. These people will be present as consultants but are not operative in the dubbing process. (This way of working is known as the 'English system' and is held to produce more artistic results. In the 'Hollywood system', which is claimed to be faster, subordinate editors are each given individual reels to work on simultaneously rather than assuming responsibility a specialised aspect.)
At the start of the dub the Dialog is worked on to achieve a smooth and intelligble premix track with as much separation as possible. Most of the dialog and ADR is in mono. Location dialog is retained wherever possible, but where it has been post-synched the editor may be offering one or more alternatives for the director to choose from. The dialog editor will provide loops featuring a variety of room tones and atmospheres recorded as wild track (ie without the camera running) in each of the locations where filming took place. Also on the dialog tracks there will be grunts, shouts, treated voices (phones or loudspeakers), and crowd noises which will be premixed with as much separation as possible. When all the decisions have been taken the dialog will be premixed, reel by reel.
The team then moves on to premix the Sound Effects tracks. These will include explosions, car noises, footsteps, gun clicks, ricochets, body falls, animal noises - anything, in fact, which isn't related to the human voice. This procedure is repeated for each reel.
For the final component, music, Graham's policy is not to premix unless the cues are unduly complicated. For the Bond film the music has arrived from different sources in a number of formats which are to be played in from time-coded masters. The traditional Bond title song which for Goldeneye has been recorded by Tina Turner is on DAT. The film's composer, Eric Serra, has recorded the score in two parts, the symphonic music was recorded at London's Angel Studio on DA88, with synthesized music recorded at Xplorians Studio in Paris on ADAT.
Once the team is happy with the premixes the final dub commences. At this stage the director is called. Graham encourages directors to stay away during the preparatory period in order to keep their ears fresh. The grand finale usually involves three dubbing mixers controling up to 20 faders each. After agreeing levels reel by reel, the faders are programmed, and the master recording made.
The $64M question is "couldn't all this be just as well done with one person using a computer?"
Graham Hartstone doesn't think so. "There's a kind of ritual drama to a dub, especially on a big film. An enormous number of people have worked on it, and this will be the first opportunity for the key production staff to see how it's really come together. It's tremendously important for the director to get a feel of how it 'plays' from the other people involved before it's too late to change things."
Speaking about television dubbing which now mostly uses time-coded units slaved to tape Graham says "all film dubbers have a real problem with digital dubbing. You miss hearing the rhythm of audio as it reverses up when you're doing a drop in. You use the reverse to listen to the EQ and check other details. I can't really explain, it's a 'feel' thing."
Like most dyed in the wool film craftsmen he's not opposed to digital technology, but he feels that until some standardisation occurs it will continue to be more applicable to low budget production or television. "We go thru waves of enthusiasm as people experiment with some new system, followed by waves of backlash as they get burnt." He feels it works excellently when a production is 'in-house' - "but the problem with films is that it's here, there and everywhere." Since most films rent equipment "the real nasties occur when people change their minds after the machines have gone back."
There is a further consideration which applies to Feature films, and that is remixes following previewing. Altho here it's still unusual for a film to be reedited /remixed more than once. In the US it's not uncommon for it to happen three or four times. Theoretically digital should be in its element here but the problem remains the enormous hard disk memory requirements and the attendant time/cost constraints of uploading and downloading. "We've never had a problem with digital equipment using PAL time-code, but I've known horrendous problems to occur with the various NTSC time-code rates. And at present I think it isn't worth the aggravation."
So far we've considered the process entirely from the perspective of 35mm theatrical release films. But nearly all tv drama shot in the UK is 16mm. And film is certainly not dead for all sorts of other applications. But of course the greatest growth over the last two decades has been in the increasing popularity of video formats now that editing has become so flexible and affordable.
So far as dubbing for TV drama is concerned the significant difference is the scale of post-production, at least in the UK, employing less than half Feature crewing. In this medium lower budgets and a faster pace of production mean that the wholesale replacement of backgrounds and dialog are avoided wherever possible.
Television dubbing has been far quicker to embrace digital technology because of its proximity to electronics of video and Electronic News Gathering (ENG) procedures. Here time-coding is a way of life. All this has led to a divergence of philosophy from film dubbing, but of course the Brave New World of digits may one day bring everything back into convergence É or not, as the case may be!
Interviewed by Norman Humberstone in AudioMedia (date) about the acquisition of 28-fader, 112-input AMS Logic 2 fully automated digital desk, Steve Haynes, Head of Post-Production Sound for YTV pointed out the fundamental difference between music mixing and dubbing for pictures. "Probably the most important thing [in dubbing automation is that] you fiddle with the EQ as much as, if not more than, the levels. I realised that fader automation alone wasn't the answer.
"It was damned hard work to start with, but then the whole idea of having one fader and one set of controls with four layered signal paths suddenly clicks, and you start thinking 'this is great, I can put the echo send and echo return on the same fader and just switch between the two.'"
Describing his working method Haynes said "I'm keen on is doing 'virtual mixes'. I'm doing it all the time now, for example on Heartbeat. The idea is to do a video production track, laid on AudioFile, where you do a mix with everything together. Because of the automation you can concentrate on one area, such as all the effects, and make sure that's right, then sort out the dialogue whilst the effects section is doing what you've already done. And of course you can go back and re-do any area.
"Without that system you would do a pre-mix where you're 'locking' things off, or you'd try and juggle 10 faders at once, and end up driving yourself crazy, Then the director will invariably say 'just go again and keep the dog bark down!' Now the flexibility is fabulous: I can just do a mix, and sit back and watch the faders and EQ waggle. Then if the director still says he can't stand the dog bark, you can say 'OK' and remove lt. Everything else stays the same, and you haven't lost anything.
"The automation, especially facilities like the Auto Takeover, is superb - it's taken a lot of stress out of mixing. When someone asks you to go over something again and again and again, you just smile at them now there's no need to try and talk them out of it!
But asked if the new system has led to speedier productions Haynes replied "Well that's a bit of a moot point really. To be truthful, it sometimes isn't quicker because what happens is that now you're able to just go back and selectively change one thing, like the pan or EQ, we're actually honing things to absolute perfection. I think if we were mixing to the [previous] standards we'd actually be a whole lot quicker. But there is a great temptation to be very exact because technology has given you the opportunity."
YTV's Dubbing Studio 1 accommodates 16mm, Super 16mm, and all video formats. The overall design and integration of the Logic 2 system was aimed at incorporating the existing Albrecht film transport machines and ways of working. "I designed the system around being able to cope with both video and film. You can have a video picture and Audiofile tracks, or you can have film tracks, or a combination of both, So you just press a button and you're working on film or video, which is great because what we've got now is a theatre which combines the advantages of all the different systems. For example on Jimmy's the sync tracks are track laid by a film editor, but the backgrounds are done on AudioFile. We also have a DAT with about 30 wild tracks of atmosphere - hospital, operating theatre, wards etc - and you just get the programme and decide where appropriate atmospheres are needed. We also lay the music down, and then mix a combination on the desk of the AudioFile tracks and the film dialogue tracks. It works well and makes a nice job of something for which really there's not a lot of time, because of the pressure and speed of production."
The Commercial Rate
As facilities become increasingly specialised in order to cater for niche markets, so working practices adjust to meet the different conditions. And nothing has played a more significant role in recent 'adjustments' than the coming of digits. As a tireless sleuth I felt I should also explore the operation of a 100% digital environment, and so I went to visit The Tape Gallery, also in London's Soho.
Founded in 1981 by Lloyd Billing, a dubbing mixer with a background in radio commercials, The Tape Gallery became the first post-production house, five years later, to go fully digital. Now with 5 dubbing suites 2 AVID suites and 3 video transfer bays. Lloyd thinks the difference is simply "that now it's a lot easier to be more creative. The analogy is really between the typewriter and the word processor. Digital is not just a 'word processor for audio' but is the only format that really allows total manipulation of sound, ie resynthesis."
Lloyd is fully committed to Avid's Film Composer /Audio Vision system which will be at the heart of all the studios. The larger studios are equipt with SSL Omnimix 38 channel desks (32 analogue +16 digital i/o) which have an almost eerie absence of physical controls. Also slaved with the Audio Vision is a Synclavier keyboard-addressable sampler (with an inboard 200 track sequencer) which holds all the audio RAM or saves it onto MO disks and a further 8 track direct-to-disc recorder. The Tape Gallery holds some 165Ê000 sound effects on hard disk which can be instantly auditioned.
Lloyd reckons that with this technology a single operator can tracklay and dub a commercial in "somewhere between an hour and a day." As he points out "What's different about Commercials productions is that agencies are continually coming back to nip and tuck as the client changes his mind. With this system of total dynamic recall, it's as simple as reloading, hitting a couple of buttons, and printing to tape. Using Avid's Film Composer the editor can even adjust picture cuts during the dub to accommodate changes in the script."
Asked if this technology is applicable to Feature films, he replies "absolutely, but you're looking at about three quarter of a million pounds worth of hardware in each studio. And the problem for Features would be that we have to charge around £320 an hour to cover our overheads. We could do it - probably in half the time and with a third of the crew - but would they pay for it?"
A trend Lloyd mentioned in dubbing was the increasing use of Dolby Surround for TV commercials. The Tape Gallery produced the first one in 1993 for Toshiba Television and now they find that the majority of clients are demanding it as a matter of course.
Surrounding up the usual sususpects
Which brings us neatly back to Ray Dolby again. Having successfully sold the idea of surround sound to the cinema he is now taking advantage of Nicam Stereo to persuade manufacturers and broadcasters to promote four channel sound for broadcast TV, radio and VHS video. This year Prime Suspect 4 and Cracker are among the productions encoded with Dolby Surround. (Which has also found it's way onto BBCr1 & BBCr4.) As the demand for high quality sound overtakes domestic television, for centuries an area of even more abysmal sound than the cinema, it will increasingly mean that television dubbing theatres need to equip for this process.
Trying to get straightforward answers from either producers or dubbing theatres about durations or costs led me to think that the proverbial bit of string might get itself so knotted that it could never be untangled. There are so many categories of production from the James Bond SR¥D epics down to humble (or not so humble) mono sales videos. Naturally, everybody complained that schedules were getting shorter and budgets tighter.
On the other hand Hugh Strain recalled that 1957 film The Trials of Oscar Wilde with Peter Finch was in the cinema two and a half weeks after the last shot was in the can, and wondered whether that could be done nowadays?
Norman Brown put it more succinctly "every technical advance has just allowed more fucking-about time for directors."
Three to six weeks seemed an average time for a Feature film, tho both Spielberg and Lucas are reputed to have taken up to three months. And the record must surely go to Michael Cimino for spending 5 days on 200ft (just over two minutes) in The Deer Hunter - no doubt as a trial run for knock-knock-knocking on Heaven's Gate. Plainly the rule here is the more complex the production values the longer dub and the higher the cost.
Talking of records, that of the maximum number of tracks ever layed vertically would seem to belong to the doyen American Sound Designer/Editor/Mixer Walter Murch who layed up 162 at one point in Apocalypse Now. So now you know why Zoetrope went bust!
At the other end of the scale the producer of a current 16mm tv drama series said she based her budgeting on being able to dub roughly an hour of finished film a day, and that she allowed about 7% of the budget for it, adding that nowadays dubs were getting shorter because more was being spent in premixing. In fact in all my researches the figure of 7% seemed a fair average, since the smaller the lower the production values, the smaller the budget and the shorter the dub.
One problems which emerged as a recent bugbear for dubbing mixers is the fad, now mercifully on the wane, of recording music in the darkest recesses of the former Communist empire. Altho orchestras there were phenomenally cheap, they were also often phenomenally duff.
Hugh Strain recalls a production called The Longships with Richard Widmark where the music had been recorded "oh I don't know, in Czechoslovakia or Poland. The politest thing you can say is that it must have been a very cold day when they recorded the title music because the french horns were diabolically out of tune at times. We had to do our best to drown it with Longship-type waves.
"While all this was going on I got upset because I thought I could see the producer crying, and I assumed he was getting upset at what we were doing to the music É until I saw that actually he was laughing so much at the horn playing that he had tears rolling down his cheeks."
So you see, even if Cinderella doesn't always get invited to the ball she still knows how to enjoy a jolly a good laugh from time to time.
I invited my interviewees to exchange their crystal slippers for a crystal ball and peer into the future. Hugh Strain of De Lane Lea replied at once "the future of dubbing is digital without any doubt. There's so much more scope." Pinewood's Graham Hartstone was more cautious: "Undoubtedly the future lies in the digital domain, but there will be a long a gestation period when both analog and digital must interface with all the attendant problems."
However Lloyd Billing of The Tape Gallery was positively visionary: "forget dubbing for a moment, what you really have to look at digit-wise are the opportunities of post-syncing. Not only can you create multi-layered effects (for instance, a car engine 'ringing' like a telephone) but within a very short time you'll be able to superimpose the speech-patterns of one actor's voice on another - which means goodbye to ADR with difficult actors! The key issue then will be who owns the phonetics library."
But it takes an experienced technician to remind everyone of the basic facts. De Lane Lea's Norman Brown: "people make an awful lot of fuss about the technology but the end product's the same however you do it. It's got to be. Sound is there to enhance the pictures and that's all there is to it.