First off, thanks for keeping up with my blog and all the comments and emails I’ve received.
I was kind of putting this together as a bit of a personal journal of my scoring efforts on my latest project. It’s definitely turned into a bit more than that. Kind of an educational tool, and from some of the comments I’ve received, a kind of “Composers Anonymous” group as well with others commiserating.
At The End of the Tunnel…
Also, for those of you checking in regularly – I know it’s been a bit of a stretch since my last post.
The good news is that I actually have a brief slice of time to write this post as I’m proctoring the final exam of my theory class at Ai.
The bad news is that I have quite a load of work to do when I get home tonight. The deadline is tomorrow afternoon – that light at the end of the tunnel is getting closer…
If all goes as planned, it won’t be a train heading the opposite way.
I’m pretty darn certain that what you’re about to read is in no way an “industry standard”, but it’s what seems to work for me.
This all comes after I’ve imported the picture into Logic, synchronized the SMPTE, chatted with the director about where and when music happens, and have my “orchestra” all set up ready and raring to go!
- Marking Up – I’ll insert markers at the start and end points of the cue and then watch the video clip a few times. I’ll then place subsequent markers at points where I think the music needs to “hit” – either have a beat land or get some sort of emphasis. Once that’s in place, I make sure that the markers and the picture audio are locked to SMPTE. If they aren’t then as I change the tempo, the markers and audio will remain fixed to musical time rather than “real” time.
- Timing is Everything – With markers in place, I’ll then go through a pass of the clip trying to determine what the base tempo is. Sometimes this is one tempo and others, it might be multiple changes. Regardless, I seem to find that there’s an implied tempo in the picture – it could be sourced from the editing (cuts and transitions) pace or action in the picture (movement and speech pace).
I find that either a rhythm or melody starts forming in my head – usually some sort of rhythm that seems to “fit” to help flesh out a tempo. I’ll then enable the click on playback and go into the tempo map to adjust the tempo line to that bpm (beats per minute) tempo is.
- Shifting Sands – A general tempo is just that – it’s a general guideline. I’ll shift the beginning of the cue to start on a downbeat and then look for important markers that I think need to be lined up specifically with a musical beat. Usually, I’ll do incremental changes – a few bpm increase or decrease that’s not noticeable.
There’s no hard and fast rule, but I try and change the tempo starting at the previous main marker that needed to be “hit”.
When changing tempos, the further you are upstream from a point that needs to be “hit”, the less drastic the change needs to be.
Sometimes, the cue needs a drastic change in tempo – that’s fine and dandy. It can be abrupt or on a slope – it’s really to taste.
- Player Piano – In my post about the “orchestra” I’m using in this project, I mentioned that I have three tracks that are piano samples in the EXS|24 sampler that I use for sketching things out.
Once I’m satisfied with the tempo map, I’ll start putting ideas down on the piano tracks. I usually try and sing or beat out something while I’m watching the film and then try and see what I can work out on the piano.
I like to keep three tracks open – I’ve found that sometimes I’ll need all three to flesh out melodic, harmonic and rhythmic elements I’m thinking. Keeping things on three different tracks allows me to easily zero in on specific elements to change things without messing up the other ideas.
Again, there’s no hard-and-fast rule that I use. I’ll admit, sometimes I completely struggle with things – it seems as though I’m fighting with my ideas to fit. Usually those get scrapped after a while and I start again. Other times, things just seem to fall into place. I could play through things a bar at a time and use the piano roll to move notes around extensively, but there are other times that I find that I “nail it” in one take.
- I’m with the Band – Once the piano sketch is working for me, I’ll collapse and combine the tracks into one and then bring up a notation view of the track so that I can use that to “perform” the parts.
I can’t say that it happens every time, but by the time I get to performing the parts, I usually have a good idea of how I want it to sound – which instruments are going to play what… or at least, which section of the “orchestra” I’m going to be pulling up first.
“Why perform it again, when it’s already MIDI information?“
It may be true, that I have MIDI notes that I’m duplicating in a lot of the parts, but there are a number of other things that makes simple copy ‘n paste not work very well:
– velocity information may be wrong for the intended instrument,
– convincing performances on other instruments need additional controller information such as expression, modulation, pitch and other continuous controller “performances”,
– chords played on a piano don’t translate well into convincing solo/monophonic lines,
– there may be additional nuances and liberties that need to be inserted live,
– it takes a LOT longer to draw all that stuff in rather than perform it live.
- Chicken, Egg, Something Else? – I can’t really say that I start or end with a certain instrument or part. If it’s a mainly orchestral score like this one, I find myself gravitating to the string section as a place to start about half the time. I did find that I got some good inspiration out of the synth elements that I had loaded into the template – either the pads or the rhythmic/gated/sequenced synth parts.
I relied a lot on double-reed instruments (Oboe, Bassoon and English Horn) to carry melodic lines, so if the cue had stronger thematic elements, then I’d start with those parts.
There were a couple of “action” cues that made use of heavy rhythms on drums – roto toms, timpani, bass drum. In those cues, I really found that getting a solid percussion sound first helped drive the intensity of the rest of the cue.
- Spit ‘n Polish – I mentioned above that I play all the parts to get a “performance” feel. That’s not to say that I don’t go in and edit the MIDI information after I’ve played them.
Everything’s on the table – expression, volume, velocity, MIDI on/off and quantization all get touched… sometimes heavily.
The string section is a great example – jumping between articulations can have some drastic “whoa!” moments that need adjustments to volume and velocities. Another big area that gets touched is volume – swells and other volume changes sometimes need to be adjusted and synchronized. I’ll freely admit it – there’s nothing wrong with a straight line volume curve from point “A”to point “B” if you’re looking to have everything blend nicely.
- Boing! – Finally, it’s bounce-out time. I’ll mute the dialogue track and select the start and end points of the cue. I’ll return the playhead to the start point and note the SMPTE position.
When I’m ready, then I’ll bounce out the cue to MP3 for the director to review. I use a consistent file-naming convention that includes:
– the cue “number” (usually 1M##)
– the SMPTE start (eight numbers in the format: HH:MM:SS:FF)
– version number
The file name might look like: 1M3-01_03_56_22-V1.MP3 (you can’t use colons in file names – I replace those with underscore characters)
All that stuff gives everyone all the information they need about the cue in one tidy package.
- Final Countdown – If I’m “on fire”, then I won’t have a lot of edit requests coming back. Edits can basically loop me back to start updating the cue at any one of the above points – it all depends on how trivial the change is.
Once I’ve got a green light on a cue, I’ll bounce it out as a WAV file – in Broadcast Wave Format. BWF is essentially a WAV file, but with extended header information that includes SMPTE start time. This allows post mixers to insert the file in the appropriate track and then tell the file to go to it’s SMPTE start time. Of course, if this screws up, the file has all the appropriate information in it too.
On this project, I got in touch with the mixer to ask him what he needed. His answer was pretty standard – all he wanted was 24-bit, 48KHz WAV files split into stems. As for number and content of stems, it was really up to the director and I. He also wanted the stems to be in split stereo (two separate mono tracks) rather than interlaced stereo.
Based on the feedback I got from the director I determined which sections needed to be grouped / separated into stems – there were some cues that actually only needed one stem and there were others that needed up to eight.
I’m zipping all the related cue stems into separate files and uploading them to my FTP server…
Tomorrow is the deadline … so far, I can’t hear any train whistles…