Scores and Sequencers – Separate but Equal
I thought I’d take a moment to talk about the files used to make notated scores versus the files used to make audio mockups. My personal opinion is that these two processes are quite different and are best kept separate.
Of course, most people would love to combine them. The most common shortcut is to create a notated score…and then import that data into a sequencer as a MIDI file. This saves a lot of time but reduces the quality of the final audio product.
First, the underlying MIDI data necessary for the ideal mockup and ideal score are typically quite different (even for the same piece of music).
In a sequence, the exact rhythmic performance of a part typically differs (both in placement and duration of notes) from an exact interpretation of the notation. In your sequenced performance, notes are likely to overlap slightly. You may slightly ritard or accelerando here and there. A quarter note may not be a full beat…but something like 90% of a beat.
In addition, you may have several tracks in your sequence for a single instrument, where you would only have a single staff in a score. If you are using different programs, the samples and effects available from your sequencer are typically different than those available from your notation program.
For all of these reasons, you will always need two separate files if your goal is to make the absolute best audio mockup and the absolute best notated score. This is true even if you are using the same software program for each purpose.
It’s worth noting that most software programs are designed to do one or the other really well. The sequencing programs are designed to create the best audio recording possible. If you are creating a recording, then you should be using your sequencer of choice. The notation programs are designed to create the best notated score possible. If you are creating a notated score, then you should be using your notation program of choice.
Notably, the two functions are rarely combined in professional settings. Most professional film composers write in their sequencer. When satisfied, they pass this on to their orchestrator, who creates a notated score for the recording session. In this case, the sequence and the score are created in two different programs by two different people.
In short, creating the best possible score and the best possible audio recording are two different processes. While it would be nice to do them simultaneously, doing so inevitably compromises one or the other…and you end up with the perfect recording and a less-than-perfect notated score, or the perfect notated score and a less-than-perfect recording.
Too true – as proven in the last few weeks. However, as a composer, I think going through the score process first really helps me properly organize the music and “compose” vs. simply “arrange” good samples. Then, that drives a more cerebral mockup of “why” I can make 2 bars last 2 hrs, not just running macros on the lines. Ironically, when a talented orchestrator “hears” the mockup he/she can notate the proper score as it sounds, not necessarily as it is found as midi and/or metadata. That’s an amazing skill.
These are very good points to remember and young composers will do well to follow this practice.
Personally, I’ve found that notation software vs. sequencers really depend on the kind of performance you really want. I once made a mockup of a composition for piano and strings for my performers. The problem I ran into was that some rhythms and meters were too intricate for me to sequence accurately on my own. So 2/3 of my time was spent actually composing (notating) and 1/3 was constantly editing performance data. The result would have been indistinguishable from a real performance had it not been for: a) I didn’t have any good tremolo samples for single string instruments, b) I didn’t have any string sound fx (like sul ponticello) samples, c) I didn’t have any trill samples.
I also wrote background incidental music for a play. Given the dark mood of the drama, I was once again composing music that I could never perform myself. I was using sampled instruments and synthesizers in this case. Again, I started out in music notation. My main motivation here was keeping strict timing consistent with the notation software. I even managed to program much of the performance information into the notation file itself and had very little left to do once I moved to the sequencer.
I think the main thing to remember is MOST music isn’t composed with robots and computer-precision in mind. The projects I’ve worked on since then would not have worked as well if I hadn’t played the instrument parts into the sequencer in quasi-realtime (meaning I slowed the tempo WAY down). I find I rely on quantization less and less these days. I haven’t taken the step of composing in the sequencer and transcribing sequences with notation software yet, though. I imagine that will happen in the next few years.
Hey Ben –
This is something I have struggled with all along, because I tend to think in terms of notes on the staff and using the piano roll in a sequencer is a weird new world for me (despite taking two classes from you!)
Now, I compromise and will write a thematic element with notation software and then export the MIDI snippet and bring it into the sequencer where I can copy and paste, transpose, break apart, apply articulations, etc. and orchestrate as needed.
I’ve started using a music typesetting system called Lilypond for doing my scores. It’s not a graphical environment like Sibelius but done entirely with text markup (like HTML). It produces really gorgeous scores. And it appeals to the software engineer in me because I can write little programs to generate scores from templates very quickly!
– Brett
PS. Just enrolled in Berklee’s Master Certificate Program for Orchestration in Film & TV. 2010 is gonna be a good year!
This is so true! I’ve been sequencing for a long time now, and I’m more recently considering writing for orchestra, and you definitely have to notate very differently. I think it basically comes down to leaving room for a musician to express their own personality through the music. If you’re sequencing, the engineer has to program the computer to play the music how he wants it. If you sequence the same way you would score, it will sound mechanical.
I think it’s a bit closer now than it once was due to Rewire support in notation programs and support for high-end sound sets.
For example, Sibelius sound sets allow you to employ libraries with program switches and keyswitches, so you can end up using VSL, LASS, EWQL, Symphobia, et. al. for notation playback, and have things like staccato symbols in the score automatically keyswitch or program switch to the correct articulation, so your staves really do correspond one-to-one with your channels.
You can also use Rewire now (at least with Sibelius, unsure about FInale), so you can feed both notation and DAW midi channel data into the same virtual instruments at the same time, using one to sketch and another to tweak without breaking into into multiple cumbersome import/export steps. The DAW and notation program don’t need to be on the same machine, either (you could use a tablet PC for manually writing into a notation program, for example, while running Reason, Logic, Pro Tools, etc. on another machine at the same time).
Unfortunately there’s still a lot of controller data missing if you don’t spend some time in a DAW while notating, so the playback even with the best libraries doesn’t sound good enough for a final delivery without these other approaches, just as you say — but I’ve found that it really is better than it used to be even a year or two ago, which turns out to be a meaningful improvement in the case in which you are both composer and orchestrator/copyist and can’t afford the time of a separate midi-to-copyist stage (assuming you are going to have at least one live part performed and that notation is therefore needed at all).
Merely my experiences, ymmv.
Peace,
Sean
Ben:
Thanks for the insight. I have discovered that the hard way–by wondering why my Sibelius scores converted into midi and imported into my sequencer don;t sound as I would like. I much prefer writing in notation and don’t find it as efficient to write in a sequencer, either using the piano roll or entering notes by midi controller, particularly with a complicated orchestral score where I would have to enter a lot of lines, many of which are beyond my keyboard skills. I also find it very difficult to see what I am doing in the piano roll, particularly in seeing how multiple lines relate. I was wondering if you have any advice on workarounds for that situation. What about exporting a midi file from Sibelius into a sequencer, clearing all the non-Note midi data, setting all the velocities to a fixed value, and then going through and adjusting velocity, timing, note length and midi volume levels on a note by note basis? I realize that is a lot of work, but at least keeps the conceptual work in notation and allows one to take advantage of some of the work you have already done in entering the score. Any further advice or ideas would be appreciated.
Best,
John
This thing looks simple but it’s a very important concept for modern composers. I often write music in score roughly and then do the arranging by sequencers. Most of the tracks won’t looks like the parts of the score. Some parts may even need 2 or more tracks to performance all articulations. Sometimes if needed I make a full version of score. So I have to export the sequencer project to a MIDI file then import it in Sibelius. But it is also a hard work to port the sequence file to a good-looking score.
So for me, I have 3 separated work. Do I get the right track to do the work?
Thanks for the clarification, from my experience this is very true, just never gave it too much thaught. Problems arise when there is a budget / time limit, then unfortunately the two processes get unified, and as stated one or the other gets compromised.
These are very good points to remember and young composers will do well to follow this practice. Personally, I’ve found that notation software vs. sequencers really depend on the kind of performance you really want. I once made a mockup of a composition for piano and strings for my performers. The problem I ran into was that some rhythms and meters were too intricate for me to sequence accurately on my own. So 2/3 of my time was spent actually composing (notating) and 1/3 was constantly editing performance data. The result would have been indistinguishable from a real performance had it not been for: a) I didn’t have any good tremolo samples for single string instruments, b) I didn’t have any string sound fx (like sul ponticello) samples, c) I didn’t have any trill samples. I also wrote background incidental music for a play. Given the dark mood of the drama, I was once again composing music that I could never perform myself. I was using sampled instruments and synthesizers in this case. Again, I started out in music notation. My main motivation here was keeping strict timing consistent with the notation software. I even managed to program much of the performance information into the notation file itself and had very little left to do once I moved to the sequencer. I think the main thing to remember is MOST music isn’t composed with robots and computer-precision in mind. The projects I’ve worked on since then would not have worked as well if I hadn’t played the instrument parts into the sequencer in quasi-realtime (meaning I slowed the tempo WAY down). I find I rely on quantization less and less these days. I haven’t taken the step of composing in the sequencer and transcribing sequences with notation software yet, though. I imagine that will happen in the next few years.
Thanks for the insight. It is so much clearer now about the processes, and so true that each process has the best tool for the job.
Thanks!
KC
just wrestling with this stuff right now… I’m from a sequencing background but i’m about to buy sibelius. I have a feeling that for orchestral writing one can flow faster with notation, and then go into production afterwards . Sibelius 7 seems like it’s become quite advanced soundwise, and I see one can even do video and hit points.
Rewire is also a thought, as Sean says..
Only problem would be, as Songming says, the if the score evolves afterwards one has to go back and renotate.
Oh well, I will keep experimenting!!
Leave a Comment