I am aiming to write music for TV and have been trying to practise on TV programmes recorded onto video. I own an Atari computer, a Steinberg Midex+ SMPTE generator and a JVC HRDD865 VHS recorder. Having read several articles on TV composers in SOS who were using Ataris with separate VCRs (Nigel Beaham‑Powell and Bella Russell in April '96, Klive Humberstone in May '94 and Danny Chang in September '94), I assumed my setup would be up to the task, whether I used a pre‑timecoded VHS tape or striped timecode onto the tape myself.
The video has an audio dub facility, a frame shuttle/jog wheel, separate audio left and right outputs for stereo output, and stereo audio input. However, when I stripe video tape with SMPTE from the Midex+, I always lose the existing dialogue on the tape (although when I connect the VCR's audio outs to the Atari, it does chase to the video). If I turn up the volume on my TV, all I hear is SMPTE noise. Is this inevitable or have I missed something simple? I am unable to adjust the SMPTE input level as mentioned in the SOS Danny Chang interview.
The finished soundtrack will be transferred to DAT rather than video, so the music I put to picture is just for reference, but the loss of sound makes it impossible to construct the music around the actor's lines. I have been told by a couple of people that I can't keep the dialogue and sync the Atari to video, and would need to buy a Soundscape SSHDR1+ system with Fast Master AV card allowing me to transfer the whole video and dialogue. If this is true, I am baffled as to how the aforementioned TV composers were able to use separate VHS machines with their Ataris. Can you help?
Frank Martin Spears
Dave Shapton replies: Before answering your question, it's important to know something about how a VHS video tape is constructed, and how audio is recorded to it. On modern tape, audio is recorded in two places. The first is a low‑quality linear stereo track along one edge of the tape, sometimes referred to as the 'conventional' audio track. The other way audio is recorded is as a so‑called 'hi‑fi' non‑linear stereo signal, embedded with the video signal on the main part of the tape.
It's hard to know from what you say here how much access your VCR's audio dub facility offers you to its various audio tracks. Many domestic VHS recorders only allow you to overdub the 'conventional' linear stereo track, but some offer access to the 'hi‑fi' track as a whole or even the individual left and right 'hi‑fi' channels in isolation. Whatever your VCR's capabilities, it's most likely that its outputs contain a mix of the 'conventional' and 'hi‑fi' audio tracks. I suspect that when you use the audio dubbing facility to record the SMPTE code, it is only going onto the 'conventional' track, leaving the dialogue on the 'hi‑fi' track untouched. Nevertheless, because the output of the 'hi‑fi' and 'conventional' tracks has to emerge via the same stereo outputs, you will always hear timecode superimposed on your dialogue. Furthermore, you say that you have no control over the level of timecode that you input to the recorder, which suggests that it is going in pretty hot: so much so that it is completely swamping any dialogue tracks on playback (although at least it's loud enough for your Atari to recognise and chase to it).
What you need, of course, is a separate output for your timecode. Some VHS VCRs offer this facility, but like timecode DATs, their cost runs to thousands of pounds. Of course, you may be lucky, and own one of the domestic VCRs that lets you record to the individual left and right channels of the stereo 'hi‑fi' track. If this is the case, you could try the following 'quick and dirty' workaround.
Get a second video recorder so that you can dub videos to your existing machine. Copy the video directly using a composite (phono) lead, but route the right and left stereo audio tracks through a mixer where you can sum them to mono, and then route them to the right‑hand 'hi‑fi' channel of the recording video. As you record, route the timecode output from the Midex+ into the left‑hand 'hi‑fi' channel of the recording video. When you play back, you will have a mono guide dialogue track on the right‑hand 'hi‑fi' output and timecode on the left. Though separation between the left and right channels of a VHS 'hi‑fi' track on a VHS machine is pretty good, you'll inevitably get some timecode showing up on your dialogue track, but since this is only for reference you can probably live with it. If the level of the timecode signal is still a problem you could record that through the mixer as well.
Of course, it is perfectly possible to run such systems and make them work, as the existence of all the Atari‑based music‑for‑picture composers interviewed in SOS proves. However, there are a few things you should be aware of when you are working like this. You should get pretty good sync while the video is playing, but if you want to synchronise a musical event with a particular frame in the video then you'll have a problem. The SMPTE you have laid down to the tape is an example of Longitudinal Time Code (LTC); it is just an audio signal recorded alongside but separately from the video, read from the tape and interpreted by a SMPTE reader, much like a MIDI FSK Sync code. If you run the tape a little fast or slow, it shouldn't matter, as most timecode readers can work at speeds well above and below normal play speed. However, if you pause the tape, as you would do if you were trying to identify a specific frame to sync a musical event to, output of the timecode will stop, and immediately you'll lose sync with your sequencer. Furthermore, the sequencer will have no way of knowing where it should be until the tape starts again.
Professional setups get round this by using another variant of timecode called VITC (Vertical Interval Time Code, pronounced Vit‑Cee in its abbreviated form), which is actually an integral part of the video signal, normally recorded on two lines near the top of the picture, and usually out of view. Even in freeze‑frame, VITC is readable, because it is repeated every time the frame is shown. Hoiwever, VITC‑enabled equipment is, inevitably, expensive.
Another thing you should consider is what happens to your audio when you want to put it back to tape. For it to be perfectly timed to the video, you would need to put it back on to a timecoded medium. If you were going back to the original tape there would be no problem; but I don't know of any domestic video machine that will let you play back an audio track (for the timecode) while recording. Even if you striped your video with 25fps timecode, there is no fixed relation between that timecode and the actual frame rate of the video. Think of it like this: if, for whatever reason, your timecode is one percent fast, then your music will be slow when you record it slaved to a device that has timecode derived from the original video. Such variations in timecode can and will arise when you are using equipment with unmatched timebases. This is why all professional environments have a golden rule that every piece of equipment should be slaved, or referenced to, a master clock — preferably a master audio clock that can be divided down to give a video frame rate that is fixed to the audio sample rate.
Now, there is a way around these problems. In an ideal world, your video playback should be slaved to your audio playback. There is a simple way of achieving this, but (once again...) it does involve some expense. These days it is pretty easy to capture video to a PC or Mac and have a MIDI + Audio sequencer control its playback (although it's not so easy on an Atari, sadly). Most sequencers and audio packages now let you load a video file that will play in a window, following the cursor on the audio timeline perfectly. You will be able to scrub to an exact frame, and you will have no problems putting your audio onto a timecode‑enabled tape device because the sequencer will ensure not only that there is a proper relationship between the timecode you have been using and the video, but that the audio sample rate has a direct relationship to the timecode as well.
Video capture cards are getting better and cheaper; something like a Fast Master AV or a Pinnacle DC30+ would be ideal, and you should be able to pick one up pretty cheaply. If you want something more sophisticated, the Pinnacle DC50 works with most video formats and can do transitions in real‑time, while the Matrox RT2000 can even do real‑time 3D effects — so not only will your music be in sync, but you can make wedding videos as well...