Everything But The Kitchen... (Part 4)
We conclude our series on synchronisation by looking at Time Code and its applications in the video and music industries. Film editor Richard Atherton is your guide.
The concluding episode of our series on hardware syncing looks at SMPTE time code and its many audio-visual applications.
Of all the essential qualities normally demanded of a film editor, the need to be able to refer to and identify any one frame from rolls of film is an extremely important element. It's a process eased by the film manufacturer printing numbers on the edge of the film: these are printed through automatically on any subsequent copies or prints so that any one frame is always accessible by reference to this 'edge number'.
Video is different. You can't hold video-tape up to the light and see a picture through it, neither can you print numbers on any part of it. Pictures recorded on video-tape can only be viewed by playing the tape on a suitable player and watching the output on a television monitor. Picture information is still made up as frames, however, and, in Europe, there are 25 frames per second (the Americans use 30). So with the advent of video-tape recording, some system of labelling each frame had to be developed. And although several systems came to light the Society of Motion Picture and Television Engineers (SMPTE) of America and the European Broadcasting Union (EBU) standardised on a system known as Time Code.
The code is a sequential electronic 80-bit digital signal that can be recorded either on a dedicated time code track or on a spare audio track. Although it was originally developed for identifying video-tape frames, it also has many applications in the sound studio. As its name suggests, the code is recorded in the form of a 'time' consisting of Hours, Minutes, Seconds and Frames. The starting point can be arbitrary or, to follow common studio practice, set to the actual time of day (24-Hour clock, of course). As an example, the last frame before midnight would have the address 23 Hours, 59 Minutes, 59 Seconds and 24 Frames (23:59: 59:29 in the USA). At midnight the address would change to 00:00:00:00. This Time Code Address, together with some additional information, makes up the Time Code Word. This word is divided into 80 equal parts called Bits (in case you didn't know, it's derived from 'binary digits'), each of which can have a value of either zero or one.
Bits are created electronically by fluctuations or shifts in the voltage of the time code signal, using a technique known as Manchester Bi-Phase Modulation. A new bit, equal to zero, is created whenever the signal shifts from one state to the other. Figure 1 shows a string of zeros: to create a 'one', there's a second voltage shift half-way through a bit period, and Figure 2 shows a pattern of zeros and ones.
In order to represent the time code address, each decimal digit is encoded into four bits using the well-known Binary Coded Decimal (BCD) technique. Figure 3 shows the decimal digits of 0 to 9 as four-bit binary codes - note that within each set of four bits, bit zero is recorded first.
The actual time code address uses only 26 of the 80 bits of a time code word. Of the remaining 54, 32 can be used to enter a further eight decimal digits which stay constant throughout the recording of the code: these are called User Bits. A fairly common practice is to enter the date of the recording (eg. 00:12:11:84), but more often than not, these bits are not used at all.
The remaining 22 bits are used for technical reference information. For example, the last 16 bits are used as a fixed synchronising word: among other things, this indicates which direction the tape is travelling in, thereby ensuring that the time code is always read correctly. As for the remaining six bits, they're used in television applications to indicate correct colour framing and to distinguish between 25 or 30 frames per second modes.
Obviously, as time code was developed for television, this has been the medium that's seen its use more than any other. Until recently, at least.
A time code generator lays down time code on a dedicated channel whenever shooting is done on video. (A spare audio channel can be used if a dedicated time code track is not available). When the tapes (called 'rushes', a hangover from film days) are played in the editing suite, a time code reader allows instant access to any part or frame of the tape. Most edit suite controllers these days are computerised, and if the required time is entered into the controller via a QWERTY keyboard, the tape player will search for that time code value automatically. For example, if a piece of action starts at 01:20:32:15 this address is entered into the controller and the video-tape player will then search for that frame. Using this technique, a programme can be quickly and efficiently compiled with the minimum of fuss.
One trend in video-tape editing these days lies in the field of 'off-line editing'. Never heard of it? I'll explain. A copy of all the rushes is made on a cheap video format (VHS, say), with the time code shown as part of the picture: this is created by an output from the time code reader which also generates characters. The VHS copies can then be edited 'off-line' away from the (expensive) 'on-line' edit suites. When the off-line is completed a list of all the edit points (ie. where one shot ends and another starts) is made. This edit decision list is then taken to the main on-line suite where, after the list of time code numbers has been entered into the controller, the programme can be compiled automatically. If the data is stored on a floppy disk from the off-line suite, all the editor has to do is transfer it to the on-line computer. This leaves you with nothing to do except change the tapes when instructed to do so (Luxury - Ed). In practice though, you won't find any editor giving that much power to a machine. They do make mistakes, you know.
And just in case you're wondering what relevance this has to music, remember that it's an awful lot cheaper to do your creative editing on a format like VHS if you've got a SMPTE time code that makes it possible. Also of importance is the Sync Word (the last 16 of the 80 bits), which can be thought of as an electronic sprocket hole. The tape machine motors are controlled via a servo, which is itself controlled by the reading of the time code sync word. Hence the actual tape speed is precisely controlled, any small fluctuations being quickly counter-acted: it's this ability to regulate tape speeds that makes time code so useful in audio recording studios.
Because of the frequency of the time code signal, it's usually laid down on the outer edge track of a multitrack tape. The code is a well-known source of crosstalk and must be kept away from tracks which have bass lines and low frequencies on them.
As with video-tape recorders, the multitrack's motor servos are controlled by the reading of the sync word, yielding exact record and replay speeds. And one of the companies in the forefront of time code use and development in studios is Audio-Kinetics; their system, Q-locks, has become synonymous with all time code control systems.
Quite apart from controlling tape transport speed, the nature of time code lends itself to several other important applications in the sound studio. As with video recorders, it allows rapid access to any segment of the tape, and the Q-lock system has five memories for storing significant locations such as start of track, start of vocals, and so on. It also has the ability to play a tape continuously between any two points - obviously useful if you have an instrument solo to 'drop in' between verses, say, and you think you're going to need several attempts before you get it absolutely right.
Another popular feature of Q-lock is its ability to control up to three recorders simultaneously. If you have three 16-track recorders, these can be linked to produce a 45-track recording system. (Three tracks are used to lay the time code on.) Better still, if you're fortunate enough to have three 24-track machines you can create a 69-track studio, and even Trevor Horn would probably have trouble filling that many. The recorders don't even have to have the same time code laid down on them, because any offsets can be calculated automatically by the controller.
Mixing desks can also be controlled by time code points, so that if you want a fade at any particular point, it can be triggered by the controller. And not surprisingly, the controller can be interfaced with a computer so that all such points can be stored as data and mix-downs performed from the comfort of your own VDU terminal. As well as making the mixing process a lot less physically demanding, it also means that several identical master tapes can be produced simply by the touch of a button or two.
Because video-tape recorders have only two audio channels, as often as not the audio tracks are produced in the sound studio on a multitrack recorder: this is then laid back onto the master video tape.
A 16-track tape, for example, is recorded with the time code on channel 16 matching that of the time code on the video tape. Two other tracks are used to record what the videotape editor has laid down on the video-tape, and further effects can be added to the multitrack once it has locked to a copy of the video-tape so that the two tapes are always in sync with each other. Let's suppose a gunshot is to be added to the effects track. At the point where a small puff of smoke is seen coming from the gun on the video-tape, the time code is read and entered into one of the Q-lock controller's memories. This is then set to trigger an instant-start cartridge player. The multitrack winds back several seconds before going into normal play and at some point just before the gunshot, it drops into record mode, triggering the cartridge player to start at the exact time code point. The result is a gunshot that occurs simultaneously in pictures and in sound. As many as three cartridge machines can be controlled in this way.
Similarly, a voice-over can be added (the loop facility is particularly useful at this stage) as the artist watches a small monitor in a sound booth. When that's been done, the effects and voice-over can be mixed down onto two spare multitrack channels before being laid back onto the master video-tape.
This technique is also used in the film world, where editors work with a standard speed of 24 frames per second. The sound of Indiana Jones reaching inside his gun holster for a gun that isn't there (see Indiana Jones and the Temple of Doom) is actually a piece of card being ruffled inside an old motor-cycle boot, though you'd never guess it, because like all the other sound effects in the film, it was added after shooting (no pun intended) using a time-coded multitrack system.
The burgeoning field of pop video promos also takes advantage of the accuracy that a time code-locked system can give. During the shooting, the artists mime to a time code-controlled playback to make sure everything's going to be at the correct speed. When the video editor has finished putting the pictures together, his or her time code edit decision list is handed over to the audio engineer, who compiles an audio track that conforms exactly to the timing of the video. Again, this ensures that sound and pictures are always in exact sync.
Finally, musicians are now finding a use for time code to control the playing of sequencers. Syco recently gave me a demonstration of a unit made by Friend Chip by the name of the SMPTE Reading Clock (SRC for short). This allows several sequencers to be started simultaneously, and also controls the overall tempo of the music. Staggered starts are not, as yet, possible but I am informed that this is being developed. As it happens manufacturers are only just starting to realise the potential of SMPTE for use in dedicated music-related products, but one recent example of what can be done is the Roland Sync Box reviewed elsewhere this issue, which allows drum machines and sequencers to be locked to a SMPTE time code on tape. And the code itself can either be pre-recorded or generated by the Sync Box itself.
With the cost of SMPTE generating and reading systems falling dramatically, largely due to the design of special dedicated ICs, it's likely that there'll be plenty of new synchronising systems taking advantage of the sophistication of SMPTE format in the near future. Thus the average musician will have access to a time code system that outperforms conventional sync pulse standards but should remain relatively affordable.
Within the professional fields of television, film and audio studios, it's obvious that time code is rapidly becoming an essential fixture. Part of the reason for that is the system's inherent flexibility, as more uses are being developed for it all the time. Even the BBC Micro will soon find use as a time code reader and control unit, if current developments are anything to go by.
The world of video syncing is that rare thing, a hi-tech field where a product has been standardised and hence universally accepted. All we need to do now is persuade the Americans that they only need 25 frames per second, and life will become easier still.
This is the only part of this series active so far.
Feature by Richard Atherton