Making A Video Demo
...On A Shoestring Budget
Having a problem getting your music listened to by film production companies? Paul D. Lehrman tells the tale of how he put together a showreel of his music without it costing him the earth.
"Yes, we got your tape. It went on the pile with the rest of them. We'll listen to it next time we need original music. No, I don't know when that will be. What was your name again?"
That was how the conversation often went when I'd call up a film production company to see if they had received the demo audio cassette I had sent them the week before. Now that everyone with a computer and a Roland MT32 wants to break into film scoring, getting someone to pay attention to your demo tape is harder than ever.
So I decided to follow the advice I was getting increasingly frequently from producers, which was to put together a video demo 'reel'. My strengths, I figured, weren't limited to just writing brilliant music — I was also a whiz at matching the music to picture, and that's what I needed to get across to these people.
I formulated my goals for the demo. First, I wanted to obtain the best possible audio quality, but hopefully not at the expense of video quality. Second, I should have a diverse selection of material on the tape. Third, it should be under 12 minutes long, so as not to strain the attention span of people who spend their lives looking at television. Fourth, and unfortunately not least, I wanted to spend as little money as possible.
My initial idea of how to do the job was to get hold of the masters, or at least good 3/4" dubs, of some of the best work I've done in the last couple of years, take them into a video studio, assemble together some of the more brilliant segments, run off a few dozen VHS copies and send them out.
But this was supposed to be a music demo, and the music mixes on the finished videos didn't exactly show off my scoring skills at their finest. I didn't have a hand in the final mixes for any of them (few composers do), and at best, the music tracks had little top or bottom, were heavily compressed, and were in glorious mono. At worst, they were totally inaudible.
What I needed to do, I decided, was to re-lay all the music tracks after I finished editing the video. All of the original music had been sequenced, and I still had all the data files, so I could reconstruct them fairly easily. Plus, I figured, since I've learned more about scoring and my equipment has improved over the years, I could probably even improve on the music in the process.
Had I been working with a professional budget, I would have gone into a video studio with a Macintosh-controlled MIDI setup and done the whole thing there. It would have meant re-programming all the sequences for the studio's particular arsenal of equipment, which, depending on how similar it was to my setup, could have been either relatively easy or a right royal pain. But even if it had just taken a couple of hours, that would have cost more money than I cared to spend: at least £500 just for the music preparation, to say nothing of the video editing and final layback.
No, I thought, I'll do as much of the job as I can in my home studio, where the clock counts minutes, not money. I've got SMPTE sync capability and plenty of experience of recording tracks on audio tape locked to timecode. Which led to another choice: should I produce an audio tape with timecode and take it to a video house for layback, or should I try to do the layback at home as well?
I realised that if I rented a 3/4" video recorder that could handle timecode, I could lay the audio directly from my mixing console on to audio tracks of the video master. This would save one generation of audio and one generation of video, and would probably be cheaper as well. If I created a VHS workprint from the edited video, and used it for all my music editing and mix preparation (I already have VHS equipment), I would only need the 3/4" recorder for the one day I was actually doing the audio layback.
A local rental company recommended a Sony BVU-150, a 3/4" machine that can handle SP tape for the highest possible quality, has a separate address track for timecode, stereo audio tracks with Dolby C noise reduction (important, considering the audio tracks are recorded linearly at a very slow speed), and audio dub capability. They agreed to let me have it over a weekend for the one-day rate of £150.
I called my friend Susu, who works for a local video production house that specialises in educational films, and she offered me the use of her modest but perfectly adequate 3/4" editing room, as well as her expert services, for a very attractive rate. I called the various producers of the material I wanted to include, and convinced them all to send me 3/4" dubs of the projects we had done together.
After a few days of going over the material and deciding which segments of which projects to include in my demo, Susu and I go into her studio with a stack of 3/4" cassettes. We assemble the six parts, and insert a title and credit before each segment, using a character generator. We're leaving the audio in place, because I'll need to refer to it later, and we're also laying SMPTE timecode on the address track.
One segment, from a recruiting film for a hospital for children, gives us trouble: we're actually splicing together three sections from different parts of the film, and if the music is going to go continuously through the splices, the cuts have to happen on the beat. Unfortunately, some of the scenes we want to use don't lend themselves to this, and they stop before the beat. We decide to grab a few frames from elsewhere in the film to fill in the gaps. Fortunately, we can edit the video without interrupting the audio, so we still hear the rhythm of the music to tell us how long to make each cut.
We finish in a few hours, and make a workprint of the master on a VHS Hi-Fi recorder I have brought in (despite its low cost and amazing audio superiority, very few video studios seem to own VHS Hi-Fi equipment), putting the audio — ie. dialogue, effects, and existing music — on one stereo channel, and the timecode on the other. This means I won't have to use the 3/4" tape while I'm preparing the music, which could get damaged from all the shuttling back and forth. It also means I can do a 'window burn', in which the timecode number for each frame appears on the screen, which is of tremendous help when editing.
I pick up the Sony 3/4" recorder from the rental company and discuss with them exactly what I need it for: I want to play back the video, sending the timecode to my sequencer, and as the sequence plays, the mix gets laid back to the two audio tracks on the same 3/4" tape. Everyone assures me this will work just fine. I even look in the manual, and it seems to agree.
The first thing that happens when I get home is that I discover my sequencer won't sync to the VHS workprint we've made. For some reason, every 15 seconds the 'lock' light on my SMPTE-to-MTC (MIDI Time Code) convertor goes out. I play with levels, try different convertors, and even try a timecode re-shaper, but nothing works. The code just isn't happening.
Fortunately, I have a beta-test version of Mark Of The Unicorn's new Video Time Piece, which is not only a very versatile SMPTE reader/writer/convertor, but it can also do window burns. So I cue up the 3/4" tape, run it through the Video Time Piece, and make a new window burn dub on my VHS Hi-Fi deck. It goes smoothly, and when I play back the new workprint, the sequencer locks perfectly. So much for 'professional' SMPTE gear.
Re-creating the musical score for these segments is not going to be hard, but there are also sound effects, voice-overs, and onscreen dialogue which I cannot recreate, because I don't have the original source tapes for any of them. What I am forced to do is strip that non-musical audio from the 3/4" dubs I do have, and store the individual cues in a format from which they can be played back in sync with the sequenced music.
However, the audio I'm stripping also has my music tracks underneath it, and there's no way to filter them out, so this means that when I mix them together, the sync between the sequenced music and the stripped audio must be perfect. If it's out-of-sync by more than a millisecond or so, the low-level music will clash with the sequenced music, and create phasing effects.
Therefore, I have to sync the sound cues to the music, and the music to the picture. When people are talking on screen, I also have to sync the sound cues to the picture. I begin to worry whether this really is possible, but I figure I won't know until I try it.
But how to do it? Multitrack tape is not an option, because I don't have a multitrack recorder that will sync to timecode. I consider Digidesign's Deck and Q-Sheet A/V programs, believing there must be a way to take advantage of their ability to record and play hard disk audio and MIDI Files simultaneously.
Deck, however, requires the digital audio to be stored as one continuous file, and there isn't enough room on my Macintosh's hard disk for all the cues plus all the silences between them: it would occupy about 60 Megabytes. More seriously, Deck does not offer continuous timecode sync — instead, it triggers playback when it reads a specific SMPTE (actually MTC) number, and then 'freewheels' from its own internal clock. Over the course of 12 minutes, I'm afraid, it may drift away from the sequencer (Passport Designs' Pro 4), which does continuously follow the timecode, not to mention the picture.
Q-Sheet A/V allows you to call up any number of discrete digital audio events instead of one continuous one, and it does theoretically sync continuously, but adjusting start times for MIDI playback — essential for the process I was facing — is a very clumsy and slow operation. In addition, the program does not support multi-cable MIDI Files, and most of my music sequences used more than 16 MIDI channels.
(Faithful Sound On Sound readers will ask, "How about Studio Vision ?" Opcode's sequencing/hard disk recording program [reviewed February 1991 issue] would actually have been a perfect solution, but at the time I undertook this project it had yet to be released. Had it been available, Studio Vision would have made things go somewhat faster, but the same basic issues would have remained.)
The solution, I decide, is to load all of the cues into a sampler, and trigger it from the MIDI sequencer as the sequencer plays the music. The sampler I have available, a Roland S770, has just enough on-board memory to hold all of the voice cues, as long as I sample them at the lowest sampling rate (22.05kHz) and trim every cue carefully. I spend the rest of the day recording the voices and effects into the sampler from the 3/4" video tapes, and even at the reduced sampling rate they sound fine.
I edit the sequences, and assemble the synthesizer patches that I used on the various projects into a single set of banks, so that I can play the entire project as one continuous sequence. I then spend some time sprucing up the sound of the older tracks. I can't make any changes in rhythm or tempo, because the music has to match the music underneath the voice samples, but I can re-orchestrate, add instruments, and improve the stereo imaging. So I cut, copy, and paste the various segments of music together into one sequence, leaving a few bars of silence between each section for the title.
In the Roland sampler, I set all the envelopes on the voice samples, some of which are 20 or 30 seconds long, to infinite sustain. This way I don't have to use 30-second long notes to play them, but instead can trigger them with notes of any duration. On the sequence track that is triggering the sampler, I set up fade-ins and fade-outs for each audio cue by drawing in MIDI Controller 7 (MIDI Volume) data. Remember: all of this is being done 'off-line', without the video tape running. The final off-line task is to adjust the timings of the samples against the music, by moving the Note-On times of the samples ever so slightly, to try to eliminate any echoes or phasing effects caused by the music underneath the voice samples.
At this point I discover that when the audio mix on one of the films was originally done, the tape machine playing the music was obviously not locked to timecode, and in fact was running off speed. The result is that the music underneath the voice samples in that segment is too fast and slightly sharp (in pitch), when compared with the music being played by the MIDI sequencer. Ordinarily I could send a bit of pitch bend to the sample, but the voice is coming from a woman who appears on screen, and that would ruin her lip-sync. I toy with the idea of loading the sample into Alchemy, a program which lets you alter the pitch of a sample while maintaining its duration (Digidesign's Sound Designer II, Version 2.0, offers the same function, but again it wasn't available at the time), but it looks like it might be a long process of trial and error, so I figure it's not worth it.
Instead, I fool around with the sequencer tempo until the rhythm is just right. The pitch discrepancy sounds like a chorusing effect, but since the main instrument underneath the voice is a 12-string guitar, it doesn't sound bad at all. I could try to compensate for the pitch discrepancy by putting pitch bend on all of the sequencer tracks, but then I'd have to worry about how the tuning of that segment would clash with the segments immediately before and after it. I decide to leave well enough alone.
Next, I turn on the video and put the sequencer into MTC Sync mode. Using the window burn, I find my starting frame number and programme that into the sequencer. Then I find the starting points of each segment, and adjust the silences between the segments in the sequencer, using the program's 'Fit Time' function to get them to match the frame numbers. I discover that some of the video edits that Susu and I tried to make happen exactly on musical beats are actually a hair's breadth off, but I can make small tempo adjustments in some of the segments (ones with no voice-overs) to make these fit.
Today's the day for layback. I cue up my 3/4" master, re-route my audio lines, and discover to my horror that none of this is going to work!
Although the Sony BVU-150 does record in stereo, and although it does have an 'audio dub' function, you can only lay back audio to an already-recorded video tape on channel 1, but not on channel 2. Why? Only God and Sony know. The rental company sure didn't, and this all-important information was buried deep in the manual, so I didn't notice it during my cursory reading.
I'm later told that Sony never considered that anyone might want to do this, because these portable video machines are generally used for field recording. The only reason anyone would ever want to do an audio dub on such a machine would be to replace a voice-over, and that, of course, only needs one channel. What would it have cost Sony to put in stereo audio dubbing? I reckon about 25p for the second switch!
A few phone calls later, I find that a friend in another studio has a 3/4" machine that really will do stereo overdubs, but he can't really let it out of his studio because it weighs 90 pounds and is worth a lot of money, and perhaps there's some way I could bring all of my equipment — synths, samplers, reverbs, compressors, mixer, synchroniser, and computer — over there?
After the blood returns to my brain I realise there actually is a solution, and it's right at hand: master the thing on VHS. As it happens, I have two VHS video recorders, one of which offers the super high quality S-VHS format. I can dub the video signal from the 3/4" to it, feeding the timecode to my sequencer, which will produce the music, which I can record simultaneously on to the stereo Hi-Fi tracks of my S-VHS machine. It means an extra generation of video, and unfortunately I can't take full advantage of S-VHS's high quality, because I'm not feeding it a component video signal. The audio quality, on the other hand, will be a lot better than it would be on 3/4", because I will be recording in Hi-Fi. In addition, because my other VHS recorder is also Hi-Fi, when I'm all done I can make my own dubs for potential clients, thus removing the need to give the 3/4" master to a duplication house. This will save me money, and will also give me control over the audio quality of the dubs, something duplication houses are not known for paying a lot of attention to.
I make a few final adjustments to the mix, then do a trial pass. The synchronisation works perfectly. Another pass, and I'm done. I like the results, and the first few people I show it to are impressed. They're even more impressed when I tell them how much I spent on the project: less than £500 in total.
After I had sent my video demo reel to a few people, I began to receive some feedback. Some of the parts were received well, and some were thought to be too long or repetitious. I was also told by several people that the music was too prominent in places. I replied to these people, "It's supposed to be. It's a music demo." They would then patiently explain: "Producers are not only interested in how your music works with images, they're also listening for how it fits underneath the dialogue and sound effects."
I realised they were right, and decided a remix wouldn't hurt. And as long as I was doing that, I thought, I could spend another hour or so in Susu's studio, eliminating some of the more boring segments and inserting a new commercial I had just finished. So we made the changes in the music: I re-edited the music, which took all of about half an hour; I rented the same Sony 3/4" video recorder; I contracted the services of another friend who mixes audio for one of the local TV stations to lend his ears; and prepared to do battle.
Trouble was, I couldn't get the video image from the 3/4" machine to stand still. It was jumping all over the place, and the timecode it was producing would last for only a second, then disappear for two seconds, then come back. In a panic, I took the tape back to Susu's studio to see what was wrong.
Somehow, after we had finished all the editing, the master tape had become chewed up, so that the control track on the bottom edge was no longer readable, and therefore the whole thing was now unplayable. The logical culprit was the rental machine, but the rental company swore they could find nothing wrong with it.
The solution? There isn't any. At least I didn't pay for the rental, but now I've got to start over again. I will have to retrieve the 3/4" dubs again from the producers, since in the interim I had been a nice guy and returned most of them, and we'll have to do the video editing from scratch. Of course, now that we know what we're doing, things should go much faster, but it's still wasted time and money. Fortunately I have my S-VHS master intact, so at least I can continue to make copies of my original demo.
The lesson to be learned here folks is that you can indeed produce a video demo in your home studio, provided you have the patience and you know your gear well. Just be prepared for Murphy's Law to manifest itself in new and interesting ways amazingly often. Make backups of everything. Above all, don't trust anything a video dealer, service person, or instruction manual tells you about how their equipment handles audio — try what you need to do yourself. If it works, great. If it doesn't, try something else. You'll get the job done, somehow.
Paul D. Lehrman is an author and composer, based in Boston, USA, who specialises in manuals for high-end music software and hardware, and music for industrial and educational films.
Feature by Paul D. Lehrman
Previous article in this issue:
Next article in this issue:
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!