The Technical Aspect and Work Flow of Orchestration in the 21st Century Pt.1

The job of being an orchestrator for media music has changed fundamentally over the last decades. 

While the job until the early 2000s was mainly blowing up handwritten sketches to full score or in some cases do some additional arrangement work, the overwhelming part of the job today is to transfer more or less complete DAW sessions to notation.

In a sense the job has reverted from blowing up sketches to reducing down multi layered and produced sessions.

Orchestration nowadays is also more tech heavy than ever before. It is expected  for an orchestrator to own and be able to handle any major DAW platform alongside the major notation softwares.

Personally, I have been working as a professional orchestrator since 2007, so I basically got into the game after the transition to DAW based composition happened already.

Yet, it took me several years to develop a streamlined work flow. Particularly in my first few years softwares were more limited in their capabilities than they are now which caused considerably more work when doing orchestration. For instance, notation programs didn't have any sort of "magnetic layout" which caused literally every symbol to visually collide with something else needing to manually adjust each and every dynamic and hairpin etc.

Yet, even in 2020 the general mode of operation when transfering from DAW to Notation still is via MIDI files. You would export a MIDI file from DAW and import it into Notation software and bring it to a state in there where it is readable for live players.

Softwares like Dorico have started to push the boundaries between DAW and Notation but still there is no seamless cross-integration so the step of a MIDI file is still necessary.

In general, messy MIDI files are a lot harder to clean up in Notation software than in DAW so there usually is a necessity to load the file up for clean-up in a DAW first.

Depending on the composer you work with (or your own work flow if you orchestrate your own work), they will either send you the Session File they produced or a straight forward export of a MIDI from it or maybe even a cleaned up MIDI. Usually, you might also get a Demo and/or Stems of the sections.

In any case, it is recommendable to do a clean up in a DAW before moving to notation.

Here are the steps I usually do (differing slightly from DAW to DAW):

Quantize: 

With the need to trigger certain transitions (like legato) early to have them sound in time, actual starting points of notes might not lay exactly on grid. If you import those straight away into notation you will end up with a lot of super small note values that are tied over bar lines or beats and are super annoying to clean up in notation.

Quantize Note Ends:

With tracks that are recorded by playing them into the DAW, you might get inconsistencies in note length. For instance if the composer has played in a row of staccato eighth notes, they might have played them in a way where they play them very short resulting in really short note events. If you import something like this into notation, you will often not get eighth notes but rather something like 32note followed by dotted 16th rest, 16th note, 16th rest etc. Again, this is much quicker fixed in a DAW than it is in any notation. In Dorico, there is a function where you can extend the length of the note until the next note effectively getting rid of all rests in between, in Sibelius there is no such function. Yet, it is still easier in DAW to just quickly fix that.

Particularly annoying are also imports of notes that overlap into each other so this is where this step will save a lot of work as well.

Merge Midi Events: 

There are two sides to this. In Logic for instance, several midi events on the same instrument track will each export as a distinct Midi tracks in the Midi file. In very busy cues with lots of midi events you might end up with hundreds of tracks in the Midi file and respective an incredible amount of staves in the notation software. So it is very advisable to merge all midi events to just end up with one long midi event per track. 

The other part of this step is slightly depending on the material and the workflow of the composer.  With composers who prefer to work on a "one track per articulation" basis, you might for instance have several tracks of "Violin 1", one with all legato notes, one with staccato, one with pizz etc. In some cases it might be advisable to keep it that way before exporting to Midi and sort that in the Notation software. Particularly when there are many switches between articulations and only few articulations used. This will save you from carefully listening through the stems or cross referencing the session later on to catch all the switches in articulations.

In other cases (for instance where the different tracks are used for layering) it might be useful to merge these all to one track. Careful though that you don't lose any notes that way that might cancel each other out when merging two tracks that layer each other.

It should be noted that in one of the latest updates, Sibelius introduced a more advanced Midi Import function that let's you adress such issues dring the import.

Key Switches: 

In this case I separate between the notation software I'm working with. If I'm going to Sibelius, I would delete all key switches in the Midi File as it is a nightmare to manually delete them in Sibelius. You can theoretically in more recent versions of the program filter notes by pitch (range) and delete those but this still will leave you often with long notes that are split into tiny fractions and tied over rather than one long properly notated note.

If I go to Dorico instead, I would in many cases keep the keyswitches as they will give an easy to spot way to see articulation changes and adjust the notated articulations accordingly. In Dorico you can filter these notes again later on with the big advantage that after deletion, Dorico will automatically rewrite the notation of any long notes reverting back to readable rhythms.

In general, it is of course easier to delete the keyswitches in DAW as this will save you from being overwhelmed with endless amounts of ledger lines after import but if those don't scare you, in some cases keeping Keyswitches to import into notation might speed up things.

Bar Numbers/Track Lengths:

In some scoring situations it might be useful to write the cues in one big DAW session. This might result in some cues starting at something like bar 1748. If you export this and import it to notation you will end up exactly with these 1748 empty bars notated before seeing the first note. So of course it makes sense to move these cues to bar 1 before export. Careful though as the composer might want to keep the super high starting bar number for reference, however I usually try to convince them to not do this as it is super awkward in a scoring session to say something like "Cello, bar 1748, can this be a bit louder?"

Additionally, some DAWs when accidentally set to weird track boundaries might export thousands of empty bars after the actual track into the midi file which might even crash some notation programs on import. So generally, try to set the track range to reasonable values before exporting.

CC Data:

Usually, there is no use for keeping any of those in Midi files with the exception of maybe CC64 (Piano Pedal) which will import in Dorico for instance as pedal lines. Other than that, it really makes no sense to keep them as they will clutter the page in Sibelius and mess with any playback in Notation software. Modwheel and/or CC11 are worth having a look at in the DAW while transfering to notation as they usually are used for dynamics and you can see where there are crescendi etc.

When you import into notation, make sure to carefully set any import options. In spite of all the preparation from above, some things still import weirdly if not taken care of. Usually you can select the smallest note value that should be notated and how tuplets are interpreted. If you set these values properly, it will save you already a lot of work.

I generally never work on the actual import of the midi file but use this as a intermediate to copypaste material into my prepared notation template. Usually you will spend way too much time fixing notation options, instrument names etc. in a Midi Import so it is just much easier to copy paste into a properly prepared template.

There are some really important functions that save a lot of time in either Sibelius and Dorico (don't know about Finale). In Sibelius it is called "Renotate Performance" which will really help to clean up things quickly. In Dorico it is caled "Requantize" which will do the same. Both functions work better when used in smaller passages to address import problems rather than applying to the whole piece.

Also, don't understimate the power of filters in notation software. All of them come with a vast variety to filter objects by certain properties which can speed up things a lot in notation.

When working your way through the piece, it of course helps to have the session open in DAW on another screen to cross check as well as listen to the demo/stems.

There are a few more common issues that touch more on notation than the techhnical side. For instance, inexperienced orchestrators tend to try to transfer every CC curve into notation ending up with a lot of dynamic markings and hairpins that are really not necessary and are implied by the line or instrument anyway. I will cover these common issues in another part.