That Syncing Feeling (Part 2)
Chris Smith continues his series on synchronisation
Chris Smith locks us on to the second part of his series
Last issue's article introduced the basics of synchronisation, time code and MIDI, and really is essential reading if you are new to the subject. So if you missed out I'm sure a quick call to the MM office will assure you a back copy. As for those that did catch it who have gone out and bought a tape synchroniser, I hope you are now reaping the benefits as promised.
This second part looks more closely at recording and recovering codes, and some of the more advanced applications. You may not find it all so relevant just yet, but it is interesting to see how the basic principles apply in the most 'state of the art' audio/video recording studios.
What is worse than running out of booze before the end of the party? I'll tell you - losing your time code just before the final mix down. Both seem to happen all too often, and both can easily be avoided (usually by not drinking so much). Here are a few sobering tips.
1) Test out some tape first and find the best recording levels to suit your system. This should be around -7VU for SMPTE code, but higher is safer - as long as you don't cross talk to adjacent tracks.
2) Go direct if you can, straight into your synchroniser from your tape deck at line level (usually the phono socket input/output). Level, or impedance mismatch, is the most common 'no read' problem. Line level is the standard, about 1 volt P-P, which is that between the domestic cassette deck and hi-fi amplifier. Use one to check out the input/output characteristics of your tape synchroniser. You should get a VU reading a little below 0VU with the cassette VU control midway, when recording from the synchroniser or the tape deck. When levels from either cause a problem, you can boost/attenuate the time code signal through your desk, but you must avoid distortion and cross talk, or mixing music channels with the time code output.
3) The second most common problem is recording over the code. Protect that track. Use red tape over the record button, or chewing gum, or an upright razor blade (only joking).
4) Cut the noise reduction if you can, Dolby A,B,C,X,Y,Z etc, on the SMPTE track. Most readers can cope with NR, but cope better without it.
5) Top of the tips, record the SMPTE simultaneously onto two tracks to start with. That way if you lose one its no big deal, and you can record over the unwanted one at the end (check that they both read right through the song first though).
6) Which track to use? Well inside tracks can crosstalk onto two adjacent tracks, but outside tracks are more vulnerable to transit damage. Unless crosstalk is particularly audible on your deck I'd stick on the inside - but don't put anything too loud or too quiet on the tracks either side.
7) If you have varispeed, centralise it when recording the code, less you forget where it was, and make a note of the SMPTE stripe rate (usually 25 fps) and stick to it. Oh yes, don't physically cut and paste coded tape, because SMPTE doesn't heal.
If after all that marvellous advice you still crash and burn, here is a load more ...
1) Clean them tape heads. Give them a good scrub with cotton buds, a clean hankie, wire wool (no no, not wire wool!). You might discover tracks you forgot you ever recorded as well as hear the code clearly.
2) Experiment with the code off tape. Increase the level, decrease the level, EQ it, compress it, anything is worth a try.
3) The expensive solution - go and buy (or borrow) a synchroniser that regenerates code, and fills in for the bit you have trouble reading. You have to stripe a complete new track using a synchroniser that is synced to the existing stripe, and hope that it jumps cleanly over the drop out.
4) The inexpensive solution - same as above but with your own synchroniser, synced internally. All things being equal the restripe should not drift more than you can compensate for with subtle tempo changes, and you don't need to worry about start times if your synchroniser has a programmable start cue and delay in frames and bits (or msecs), but this can be very hit and miss.
5) The 'Chris Smith Special Method'. Say your tape drop out or misread occurs about two thirds through the song, go back to the nearest place where your sequencer isn't playing much (eg. end of a verse) and tape record the sequenced parts up to that point. Now go past the drop out point and use the autolocate feature to tape record sequences for the rest of the song. Now you can fill in the missing bit (punching-in on as yet unused tape tracks) with the sequencer synced internally, set to the more adventurous, you can restripe the tape on a separate track over the drop out point and record the sequenced 'fill in' synced to tape. This way you can use the synchroniser's cue, delay and tempo controls to line up the missing sequences just right - clever stuff eh?
So far we have only talked about syncing a MIDI sequencer to a multitrack tape recorder. Syncing two tape recorders together is rather more difficult and requires more complex and expensive equipment. Such synchronisers are usually made by the tape machine manufacturers, such as Fostex, Tascam, Otari etc. and can lock two 24 track machines together, for example, to give you up to 46 audio tracks.
The time code used is still SMPTE, which is striped on one audio track of both tapes. The synchroniser has to remotely control the capstan speed and tape transport (fast forward, rewind, play) of one of the tape recorders (the slave) to keep it locked to the other (the master). When the tapes are both running the synchroniser reads and compares the two SMPTE codes and if it sees that the master is running slightly ahead or behind the slave it increases/decreases its output voltage to the slave capstan motor to speed it up/slow it down. This feed back loop has to be very accurate (to about ±50 micro secs) or you would hear pitch or phase variations between music recorded on the two tapes.
The synchroniser also has to auto-loc the slave tape recorder, with pre-roll and cue points set for the master. The time code has a fundamental frequency around 2-4 KHz, which any audio circuitry can cope with, but when the tape is fast forwarded at 25 times play speed this frequency can increase to 100 KHz. To chase and autoloc the synchroniser must be able to read code at this speed (forwards and backwards) and the tape recorder preamp must have enough bandwidth to reply it. Most synchronisers get round these problems by counting pulses from the timer rollers on the tape machines during fast winding, and then resort to the SMPTE code when the tape slows down.
So a lot of control is needed — reading of two time codes, feedback capstan control, remote transport control, and tape counter reading. The complexity and expense of such equipment is the main reason why so many studios have chosen MIDI sequencers with SMPTE/MIDI converters to increase their track recording capacity.
Connectors and interface protocols vary between different tape machines and as yet only one synchroniser (the Adam Smith Zeta-Three at around £2200) provides a universal interface - but competition is on the way, watch this space.
SMPTE time code was actually developed by the American TV industry as a means of cueing video material for accurate and synchronised editing. Finished productions can consist of hundreds of sections of film taken from different places and times, joined together with animation, video effects and, computer graphics. Before looking at how the music is added to the finished film, it is interesting to see how and why time code plays such a central role in the production of the video program itself.
Longitudinal Time Code (LTC) is the type used for audio and video work most of the time. It consists of a series of pulses or 'bits' running sequentially along the tape in the direction of tape travel. Each group of 80 bits (pulses or non-pulses) make up a frame with a unique binary pattern identifying the frame number, second, minute and hour. LTC is recorded by the audio head of a video cassette recorder (VCR) along the outside edge of the video tape next to picture data, which is recorded in a zig-zag pattern by the video head across the rest of the tape.
One frame of video is like a single TV screen full of picture information and is recorded as a single diagonal line across the video tape. For the LTC to be useful for video editing it must line up precisely with the video signal, frame for frame, so the video and LTC have to be synced together and recorded simultaneously.
Most of a typical show or current affairs program is filmed in the studio using three or four cameras with a central video recorder (see diagram C). The video signal generated by each camera is synced by a single time code unit, which also stripes the video tape with SMPTE. The video input to the VCR is simply switched between cameras so the director can do most of the pre-production editing 'on air' or straight to the master recording. The result is something like 'Top of the Pops' (ie. pretty rough) compared to a highly edited and polished production such as 'The Chart Show' on Channel 4.
Many productions use original films recorded out of the studio, which have to be copied onto a master with SMPTE added, synced frame for frame with the video picture. This requires a synchroniser that can generate SMPTE time code synced to the frame sync video signal from the original recording (see diagram D). At the same time a working cassette copy is recorded with 'burned in SMPTE'. This is SMPTE code modulated onto the video signal which actually replaces some of the video picture (usually the bottom right hand corner of the TV screen) with a video image of frames, secs, mins, hrs.
The SMPTE burnt copy (you may have seen such preproduction videos on TV) is then used by the director and technicians to make a list of 'cue points' - frame numbers throughout the tape recording where the tape is to be edited out, joined together, and where other films and visual effects are to be inserted.
You can't edit video tape physically by cutting and pasting - it is all done automatically by machines. The list of cue points is fed into a specialised computer system which sorts and sends the appropriate data to an edit controller (see diagram E). The edit controller now reads the SMPTE code from each of the master VCR's (there may be two or three of these) which will play the various film parts that make up the final production.
Under computer control, the edit controller will fast forward, rewind, play, stop, pre-roll and synchronise the VTR's according to the cue list. Video signals are switched between the players and visual effects, fade ins/outs, animation and computer graphics can all be mixed in under computer control. This way the director can see a dry run of the complete production before recording anything, and the technician can try subtle changes to edit points just by amending his computer program.
Up until recently the audio content of productions has been very much secondary to the video pictures and effects. Now large companies are beginning to realise the value of high quality audio and music, especially in commercials. Advertising agencies spend a lot of money on specially commissioned music, or buying up copy rights, to find that illusive 'cult' or seductive sound track.
The audio content can consist of up to 5 different elements in a quality production: location sound, lip sync, voice over, music and audio effects. Location sound is usually laid down on one of the VCR audio tracks while filming the original video, the rest have to be added while synced to the video picture and SMPTE code.
Location sound (street noise etc.) usually 'fits' the picture and if it is sufficiently clear it may be carried along with the video pictures right through the editing process with its level adjusted for the final production. In many cases poor quality sound is replaced by library effects (eg. rain, car engines, doors slamming) from CD or samplers. Most of the sound track in 'Miami vice' for example, is as fake as the acting.
Studio sound (usually voices and audience response) is recorded live through a mixer, either onto the VCR soundtracks, or a two track master synced to the video. Location speech is often buried by location noise so the actors have to dub over their spoken parts while watching the picture in the studio - lip sync. SMPTE cues help the artist's alignment.
Traditionally film music is recorded with the orchestra watching the conductor, and the conductor watching the picture on a screen behind the orchestra. He controls their tempo and accent while they play from a score specially written for the film. Listen hard next time you see a 'Tom and Jerry' cartoon. All the music and sound is recorded in this way and the skills of the conductor and musicians greatly enhance the production.
These days the film musicians and score writers have a wealth of equipment - samplers, sequencers, synths, effects units, all of which have to be synced to the video via SMPTE. The musician is usually presented with a video cassette of the complete production, striped with SMPTE on a cue track and with corresponding SMPTE burned into the picture. This comes with a set of notes from the director/producer, listing the cue points where they want and don't want different types of music. Sequencers are very useful here, synced directly to the video time code using a SMPTE/MIDI synchroniser. MIDI events can be programmed to occur at specific frames read off the video (such as samples of doors slamming, machine gun fire etc.) and delayed/advanced to fit the picture action exactly.
Inevitably some music and effects come from audio tape, recorded at some earlier stage in production. In some cases the complete audio track is mixed down to one multitrack tape before being recorded onto the master video tape. For this a tape to tape jam sync synchroniser (like that described earlier) is needed between the VCR and the multitrack tape recorder. The video is usually chosen as master because the synchroniser is more likely to have remote transport and capstan controls for the audio machine. In theory it would be best to slave the video as variations in speed of vision are less noticeable than musical pitch.
The final audio track recording system may look something like diagram F. Everything is synced from the VCR and the engineer will monitor the sound and vision throughout the complete production before recording onto the VCR audio tracks, adjusting offsets, recording levels and mixer levels.
Although very few production studios use it yet, automated mix down equipment is ideally suited to this environment, and I have just enough space left to say a few words about that.
Nearly all mixer manufacturers offer some kind of automation system, from fully digital total recall desks of SSL (and the Yamaha DMP7) to simple track muting. Existing desks can be retrofitted with computer controlled fader override or insert point attenuation systems. All of these systems depend on time code, usually SMPTE, to coordinate the track muting and fader movements in time with the music.
Most systems look something like diagram G. The recording is monitored by the engineer who moves the mixer faders to fade in and fade out the various tracks and groups from the multitrack machine (or live from keyboards and samplers in a sequencer based studio). The fader movements and mutes are digitally recorded by the computer as a sequence of events synced against the SMPTE time code played from the tape machine. Atari based systems usually require a MIDI sync input, derived from a SMPTE/MIDI synchroniser as described last issue.
On play back the automation system will duplicate the fader movements and record adjustments and additions made on the mixer. The engineer can run the production from any point in time repeatedly, making fine adjustments until he is happy with the final mix down onto the master two track or CD.
The introduction of satellite and cable TV is leading to a high demand for low budget productions. With increasing air time the trend is towards quantity, rather than quality. Corporate promotional and training videos are also a new growth area, as well as locally sponsored and personal video productions.
All of these mediums will require decent sound tracks if they are to compete with existing broadcasts, but few have the facilities of the radiophonic workshop at their disposal. There is a whole spectrum of production facilities and opportunities opening up as film companies farm out their audio work to specialists. Next time I will be looking at the actual equipment used by these people, and how small studios, or even home studios, can be used for A/V work.
On the theory side, we will be looking at VITC (a SMPTE format specially for video work), music videos, recording of the video picture, and the new MIDI Time Code standard.
Feature by Chris Smith
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!