On Symphonies and Computers
by Mark Isaacs
This article is also published in Resonate magazine
In the early-to-mid-1990s I became aware that quite a number of my composer colleagues were doing their scores on computers, using Finale. By that time I’d been writing my scores by hand for around 25 years (and there were reams upon reams across numerous concert works and orchestral film scores). I dug my heels in: it wasn’t for me. I liked my tried and true methods.
The days of the hand copyist of earlier decades were gone by the turn of the millennium. For the orchestral commission I received in 2001 (a cor anglais concerto for Peter Duggan and the ASO) the score was entered into Sibelius from my handwritten copy by the team of two full-time music preparation staff (led by the redoubtable Julie Simonds, my high school sweetheart as it happens) that (then) Symphony Australia employed. My 2003 concerto for James Morrison and the MSO was entered by Laura Bishop into Finale. It was a similar situation for chamber works. The Australian Music Centre had a fund that generously payed for the “Copying of Parts” (by this time meaning “computer entry”) for Australia Council funded works, which most commissions were.
I lasted longer than most with my own “hair shirt” austerity in regards to notation software but ultimately recanted.
Given a copy of Sibelius 1.4 by my daughter’s high school music teacher in 2001, I began to use it just to produce simple charts as needed for my jazz groups. I still couldn’t conceive of doing a classical score that way. But in late 2003 I relented once again and prepared my Ave Maria for cello and piano in Sibelius, which I wrote for the fabulous cellist Trish O’Brien and myself to play (we were a touring classical duo at the time).
It was pretty obvious what the advantages were. Younger composers probably can’t conceive of a world without word processor-like score editing facilities. I’m sure sometimes I had avoided making revisions to hard copies as it was just so physically hard to pull off.
I don’t think I used the score playback facilities in Sibelius 1.4. If they were even present then, they would have produced what used to be called “MIDI files”, played back by the generic purely-synthesised imitation instruments that came built into the internal soundcards of computers.
Things started to get more interesting in 2005 with Sibelius 4, which came with Native Instrument’s Kontakt Player Gold featuring much better simulations of real instruments. In 2008 with Sibelius 5 came Sibelius’ own sampled sounds, which were proprietary adaptations of the Garritan Personal Orchestra and other libraries’ sounds. I also bought Garritan Personal Orchestra itself as well as the excellent Garritan Authorised Steinway piano samples. I became a regular on the Sibelius technical forum, initially getting much-needed help, and it’s a place at which I hover to this day, more often now helping out less-experienced users.
During this period I did find some limited uses for score playback. Firstly, it was great for aural proofreading. It is so easy, especially in busy chordal piano writing, to either leave out an accidental or overlook that one carries through the bar that needs to be natural-ised. So I was happy to catch any of these through playback and save on rehearsal time distraction over note corrections.
From 2004 I had become interested in outputting audio demos of my pieces for a small inner group to hear in the interval between a score’s completion and a performance/recording.
Even highly-experienced conductors take a great deal of time to nut out a new piece from score alone. In the old days I’d have to bash it out on the piano, trying to play configurations that don’t readily adapt to a pianist’s hands and singing in a nasal composer’s voice what couldn’t be played, yelling out “Flute!” or “Violins!” along the way. This had to be a better way. My wife and daughter certainly appreciated it a lot more! I loved to share with them what I had written at the end of the day, and this was so much more intelligible, especially for large ensemble or orchestral pieces. I’d email mp3s to my very musical parents and a couple of my close colleagues. I also began to offer these “demos” to the musicians who would be performing the piece, just so they could get the lay of the land before starting rehearsals. Some were put off in the early days by the then poor simulations of their own instruments, but nowadays many, if not most, musicians, and indeed conductors, request audio simulations.
The one thing I never did (until this year) was conceive of preparing any sort of “finished product” this way. Composers for TV (and even sometimes film) had started producing their final chamber or orchestral scores with samples. Not liking most television drama much, I would prefer to pick out what wasn’t a real orchestra (most wasn’t) along with snorting at the interminable and unimaginative take-offs of the marimba-based American Beauty (2000) music that dominated television scores for a decade.
That’s all changed: I’ve taken the plunge!
On May 10, I released the Digital World Premiere by the NotePerformer Virtual Orchestra of my Symphony No. 2 on the YouTube platform.
The reason to do this was a combination of duty and pragmatism.
My Symphony No. 1, commissioned by Kim Williams AM for the QSO, had a very successful first performance in 2013 by that orchestra conducted by Benjamin Northey. It was given very high praise by many luminaries from Vladimir Ashkenazy down and I co-produced, with video production company CBD, a TV program of the premiere which we licensed for several screenings each on Foxtel Arts TV in Australia and Sky Arts TV in New Zealand. On the strength of that success, plus letters of interest in presenting a new symphony by me from several Australian conductors as well as Richard Wenn, then Director of Artistic Planning at QSO, the Australia Council funded me directly to compose my Symphony No. 2 in early 2016.
With Richard Wenn no longer at QSO, I experienced a quite unexpected level of general inertia regarding programming my new symphony from the many Australian conductors and orchestral artistic administrators who I began to let know about it from the moment it was funded. Those that returned my emails (and several unaccountably didn’t even get that far!) wrote courteously in acknowledgment without any enthusiasm or real interest it seemed, since they did not come back to say they’d investigated the score and basic audio demo which I circulated as soon as the work was completed in early 2017. Though there is one Australian conductor who still maintains a palpable warm interest in the work, nonetheless by early this year it was still not programmed for premiere.
My sense of “duty” was simply this: the Australian public payed for this work (symphonies don’t come cheaply) and after four long years they surely deserved to hear something? My sense of “pragmatism” was along the lines that the public release of a very high level “virtual performance” may itself help finally kindle a real performance.
I knew it had to be as good as it could be if I were to go down this path. I couldn’t afford the top level libraries that sampled the Vienna Philharmonic or BBC Symphony Orchestra, and in any case I’d found out from haunting web forums how difficult these libraries were to integrate seamlessly with notation software. Fortunately, there was NotePerformer, an excellent orchestral sample library from Sweden’s Arne Wallander, which was relatively inexpensive and integrated easily with notation software. I set about creating a “first mix” of my symphony with these sounds, aided by Arne’s same-day answers to email technical questions.
Though NotePerformer is designed to work “inside the box” with Sibelius, there were limitations to what could be achieved there, due to the way Sibelius operates. Basic reverb and EQ settings can be applied on a staff-by-staff basis only. If several instruments share a staff (e.g. percussion staffs or woodwind doublings) you cannot control parameters for each individually. Fine control over balance and the audio soundstage are very difficult to achieve.
I knew that those who were serious about final product (like the TV composers) mixed their output separately in a multi-track DAW (Digital Audio Workstation) like ProTools, Logic, Cubase or Sequoia. This was both beyond my pay grade and a level to which I in any case did not wish to promote myself (mainly because if I were to do it, I’d have to learn to do it properly, it would take over my life and I would end up being a sound engineer too, a bridge too far with the already onerously-variegated skill set I am trying to maintain in music).
Fortunately, last year I had the great pleasure of working with musician and audio producer Lachlan Bramble. A terrific violinist (he is Associate Principal Second Violin in ASO as well as leading the distinguished Benaud piano trio) and an in-demand chamber music mentor, he is also an audio producer. Lachlan produced two solo piano recordings I made in Adelaide last year (one of classical repertoire, the other original music) and so at the beginning of this year I asked him if he would mix the symphony, which he was very keen to do.
Sibelius can export each staff of a score as a separate audio file, and so I exported a set of what is called in the audio trade “stems”, one stereo “stem” for each staff, and Google Drive-d them over to Lachlan. His starting point was thus my onboard first mix, which we called the “Reference Mix”, though I gave him the audio “dry” without the in-house reverb from NotePerformer I had been using so that he was free to stretch his wings in any direction.
From there we collaborated back and forth by stages, with the final product being ‘Mix Version 5’. As well as all the niceties of balance and EQ, Lachlan’s great achievement was in the soundstage, both left to right, and front to back. An orchestra is a large 3-dimensional object, and being such an experienced orchestral player himself, Lachlan knew how to work toward a sense of realism. For example, the stereo renditions of each string section already had spatial width, but Lachlan widened them so that the strings covered the entire sound stage from the centre to both the left and right flanks as they do in real life.
Once I had the finished audio, I made a basic video with title cards to go with it, and at the end of the symphony included a short excerpt from QSO/Northey performing the previous symphony along with some testimonials about it (rather like the back cover of a book, I thought).
From there it was simply an upload to YouTube for its premiere, which was, as it happened, on Mother’s Day. I felt the use of the term “premiere” was justified. TV shows premiere at a particular date/time, then hang around to be streamed on iview and suchlike. Similarly, in the COVID-19 crisis there have been a number of YouTube or Facebook premieres that also go out at a specific day/time and then stay online for streaming. Without intending it, the timing of this online premiere seemed to fit the lockdown zeitgeist whereby these web platforms are the only public place for new music.
Another decision was to clearly signal that this was not a real orchestra by crediting it as being played by the “NotePerformer Virtual Orchestra”. Often it can sound deceptively real, at other moments not so much. Without a frank disclosure to the listener at the outset, I felt that there could potentially be a kind of “Is it? Isn’t it?” distraction from the music, rather like my aforementioned musings over TV drama background music. I hope that the listener will accept the premise and then put it to one side.
Though this remains a “duty/pragmatism” solution to the programming travails of a symphony that was indubitably written for a real orchestra – and does not replace it – I now feel that at some point I might go a step further. That step would be to write a future symphony in my ongoing cycle specifically for a virtual orchestra. The reason would be to write things that would be unplayable by a real life orchestra (particularly rhythmically, an area where computers are wholly undaunted), or if not absolutely unplayable, would fall over to a greater or lesser extent given the limited rehearsal time allotted to new works. And further to the “sound stage” matter, think of the 19th/20th century precedents for placing sub-ensembles around the auditorium space, such as Mahler’s offstage and gallery brass, or indeed its apotheosis in Stockhausen’s Gruppen. A virtual soundstage would allow for the wildest things: a woodwind section flying around the auditorium while playing, anyone?
© Mark Isaacs, May 2020