Page 2: Recording Christmas Music

Scoring Santa

Once all the basic sounds were accommodated, we began by recording the title music. At this stage no timecoded video tape was available to us, so we worked to the storyboard and the supplied frame counts. It was accepted that there would probably be a lot of trimming and tweaking required when we were finally working to picture, but this was as good a starting point as any — the more material we had in place when the video cassette arrived, the better.

The title scene was to begin with a single snowflake making its way towards the ground, opening on the cartoon's title and then spiralling down towards Santa's house. The point at which the title appeared seemed to warrant a significant musical punctuation mark, and Richard had prepared a piece which took this into account. The section of music opened with a single string note, gradually building into the main theme and then coming to a climax before having a 'musical box' theme crossfading as the viewpoint floated through Santa's window. The significant timing points, such as the appearance of the titles, were marked in Cubase by cowbells, and time‑locked to enable us to change tempos without moving these reference markers.

Richard had recorded rough versions of the main themes into the sequencer of his Korg O1W/FD. We transferred these a track at a time (at half speed, to ensure timing accuracy) into Cubase, with Cubase taking its timing reference from the O1W's internal clock. On playback from the PC it was clear that something was awry. Strangely, Cubase was found to be quantising the notes as they came in, despite the fact that Auto‑quantise was switched off! The answer was to switch the quantise value to 'off', at which point everything began to transfer correctly. I had never encountered this problem with my Atari, so was quite unprepared for it.

For simplicity's sake, we recorded all the music in 1/4 time, leaving us to concentrate on tempos alone. Richard, being something of a 'traditional' musician, and an accomplished piano player to boot, was quite prepared to re‑play any parts we considered suspect. I was happy to go along with this, since it was often much quicker than using Cubase's editors and kept the creative juices flowing. When the time came for some serious editing and quantising sessions, Richard became quite fond of the new options afforded him by the technology! Some of the faster harp arpeggios benefited greatly from subtle quantisation and general tidying. The harp sound itself was found to be lacking in definition once it was blended into the mix, so the Yamaha TX802 was called upon to add an extra transient 'twang' to the sampled harp and this worked very well indeed. Richard had to do some fairly substantial trimming to fit into the time window before the appearance of the cartoon titles. This is part of the cruelty of working to video — as a stand‑alone piece of music, the original, unedited version was superior to that which will ultimately accompany the visuals!

As work progressed, I kept a cycle of backups between the PC and a couple of Jaz cartridges, so that in the event of a disk crash we would never lose everything. This proved to be very important as, on occasions, Richard would require me to return to previous versions to see how things had changed, and maybe restore elements that had been rejected in one piece for use in another.

Partly due to the nature of the project, but also due to the way in which both of us wanted to work, I made it a point to leave all the mixer channels set to a nominal 0dB level and set volumes from Cubase for all the instruments that could make use of MIDI volume commands. This meant that we could switch between pieces of music at a moment's notice without having to completely reset the mixer each time. The only problem with this approach occurs when MIDI‑controlled fades are brought into play, since any preset volume commands for the instrument are overridden as soon as the first volume command from the fader information is received. The K2000 is the exception to this rule, since it allows MIDI volume commands to define the maximum volume of the part, and then uses MIDI expression commands to fade between the maximum and fully off — brilliant! Some day all instruments will behave this way. For the S1100 I created loud and soft versions of each program, to allow some flexibility in setting basic levels.

Inevitably, there were times when 'loudest' just wasn't loud enough. On these occasions, I resorted to lifting the odd desk channel before recording the sound to hard disk in Richard's absence and setting a higher level from there. Any instruments that were beginning to show the strains of the demands on their polyphony were also recorded to hard disk during these breaks, freeing them up for further parts on the next session. This usually involved the K2000 string and S1100 stereo piano parts. I recorded these as cleanly as possible, from the digital outputs and including effects, in the hope that these recordings would make it to the final mix. For the MIDI‑driven sounds I deliberately avoided adding effects, other than those on the instruments themselves, which could be recalled by patch change messages. External effects would be added at mixdown.

The system generally worked well. There were one or two moments when program change messages were lost or ignored, but these were few and far between. Such problems were usually down to 'operator error', but a few just seemed to go away after a re‑boot and a coffee! One very annoying 'feature' of Cubase 3.05 manifested itself each time I deleted a track that had a patch change assigned to it, because the non‑existent track would continue to send its patch change regardless! Fortunately, nipping out and back into Cubase always cured this fault.

Our first major production challenge came when we were required to simulate the sound of singing reindeer.

In The Picture

After a few days we were sent a 'line test' video tape. This is a rough copy of the cartoon done entirely from basic line drawings, and is created to allow the director to confirm that the overall pace and timings work properly. This tape did not have any timecode added, but did include the dialogue, so we only had to hit the play button along with the relevant scenes to get a feel for whether the music was in the right ballpark. It was important that the music did not cover any of the dialogue, so although the music could be faded down at the dubbing stage, we tried to avoid using anything excessive over any of the important lines. Apart from the title scene, which required a fair amount of re‑working, we only had to make a few tempo tweaks and everything seemed to run pretty much as expected.

I considered it important to establish that my system would be able to cope with the timecoded tape we would eventually receive, so I asked the studio to send me a test tape containing the same format of timecode. If there was to be any problem I wanted to know about it earlier rather than later! A tape of one of the studio's previous cartoons was duly sent, complete with dialogue on one of the stereo tracks and timecode on the other. In the event, there proved to be no problem at all. My XRI XR300 read the timecode and duly poured MIDI Time Code into Cubase without any complaints. My confidence was considerably boosted.

Getting It Taped

Our work pattern was now well established, with Richard putting pen to paper during the day, while I was taking care of the day job, and both of us working to get the parts recorded into the computer at night. This regime worked very well, pausing only to accommodate my holiday halfway through!

On my return from holiday, the timecoded video tape was available and we could now start running our music in sync with the action on screen. To our dismay, the opening title sequence had once again been trimmed, forcing us to make yet more painful cuts to the musical accompaniment. Nevertheless, our work prior to receiving the timecoded tape really paid dividends, since in most cases very little needed to be done other than setting a start point.

We carefully edited starting points and tempos to hit crucial pieces of action, and I added a reference 'clapperboard' click at the start of each section of music, noting the timecode position of this click on the DAT label, so that the music could be accurately synchronised by the engineer at the dubbing stage. Some of the reference clicks unfortunately overlapped into the music preceding the cue, but the dubbing engineer assured us that he could use a second pair of tracks to alleviate the problem.

Work progressed well, and with a week and a half to go we had just over half of the soundtrack scored and recorded into Cubase. The circus scenes in particular were great fun to work on, with manic organs and crazy calliope sounds really defining the mood. When it came to specific musical effects, such as the singing reindeer (see 'Rudolph the Singing Reindeer' box), Santa's trumpet and the twinkly magic spells, I decided to record the musical backing and the effect separately, to give the dubbing engineer a chance to balance the relative levels. After all, it was possible that the producer would want the music considerably quieter than we might anticipate, while the effect was arguably more important to the storyline.

I can honestly say that this project has been amongst the most satisfying I've undertaken.

Merry Mixing

When the time came to mix the music, our job was made considerably easier by the fact that we had used MIDI volume and expression commands to do much of the work as we went along. Very little EQ‑ing was necessary, since Richard's skills as an orchestral arranger had ensured that few parts were fighting for space. The main job was to boost those parts that had run out of steam despite being set to a maximum MIDI volume of 127. Effects were generally used sparingly, to keep a strong sense of 'realism' to the sounds. I decided not to compress the overall mix, as this could be done at the dubbing stage, if necessary, although I did patch a Drawmer noise gate across the masters (and individually across the noisiest channels) to keep everything as clean as possible.

I can honestly say that this project has been amongst the most satisfying I've undertaken, and there was a real sense of achievement when it all came together as a coherent piece of work at the end. It's something of a cliché to say that the best soundtracks are those which are not consciously heard during a film, but complement the action in such a way as to blend into the background. I agree with this wholeheartedly, but I still harbour a secret wish that someone, somewhere notices the sounds accompanying the film and believes it's a real orchestra!

So how would I do things differently next time? I think I would firstly record the whole piece as a single Cubase song. I had believed that splitting the soundtrack into sections would simplify the process, but it just led to extra confusion, with some MIDI channels being assigned different sounds in different sections, and various anomalous 'hang‑overs' from one section to the next. I would also like to invest in a video recorder with more advanced jog and shuttle features, to make searching for hit points less of a drag. An automated mixer, such as Yamaha's O2R, would have made it easier to mix as we went along, saving both time and trouble. An investment of this amount and nature would have to be justified by a steady flow of similar work. But judging by the last few calls we received from the director of Father Christmas and the Missing Reindeer, and the dubbing engineer, you never know!